An ecological project environmental impact assessment result evaluation method and device

Abstract

The present application relates to the technical field of ecological protection analysis of power plant construction, and particularly relates to an ecological project EIA result evaluation method and device. The method comprises: based on natural language processing technology and a pre-constructed ecological knowledge graph, performing structured analysis on an EIA report under semantic constraints to obtain a structured EIA parameter set; according to the structured EIA parameter set, the ecological knowledge graph and a pre-constructed historical benchmark library, generating evaluation basic data containing an EIA quality evaluation score and a verification index set; based on the evaluation basic data, actual monitoring values and historical benchmark mean values of similar projects obtained from the historical benchmark library, using a triangular mutual verification method and a comprehensive evaluation method to obtain an evaluation data set; and according to the cumulative evaluation data sets of multiple projects, updating the prediction correction information in the knowledge graph and the historical benchmark library. The present application can not only improve the accuracy of EIA prediction results, but also improve the EIA prediction accuracy of subsequent similar projects.

Description

Technical Field

[0001] This invention relates to the field of ecological protection analysis technology in power plant construction, and in particular to an evaluation method and apparatus for the results of environmental impact assessments of ecological projects. Background Technology

[0002] As an important component of clean energy, wind power generation has seen a continuous increase in the number of power plants put into operation and the scale of their operation. During the construction of wind power projects, activities such as wind turbine platform excavation, road construction, and power collection line erection inevitably impact the ecological environment of the construction site. Therefore, in order to protect and restore the ecology, an environmental impact assessment must be conducted before the project starts to predict the degree of ecological impact and propose mitigation measures. The accuracy of the environmental impact assessment prediction results directly affects the actual effectiveness of subsequent ecological protection.

[0003] In related technologies, wind power ecological projects generally adopt a post-construction environmental impact assessment (EIA) approach, focusing only on whether ecological measures such as vegetation restoration and soil erosion control meet standards. However, they do not systematically evaluate the accuracy of predictions in the EIA report. This approach masks numerous EIA prediction biases, and even when discrepancies are found between actual impacts and predictions, it's impossible to trace the source of the bias—whether it stems from the prediction method, parameter selection, or measure design. Furthermore, post-evaluation conclusions typically only apply to the rectification of a single project, concluding once the assessment is complete. This fails to translate into industry-wide knowledge accumulation, leading to recurring EIA prediction biases in similar projects.

[0004] Therefore, there is an urgent need to provide an evaluation method and apparatus for the environmental impact assessment results of ecological projects. Summary of the Invention

[0005] This invention provides a method and apparatus for evaluating the environmental impact assessment (EIA) results of ecological projects. It can solve the problem that existing EIA methods for wind power ecological projects only focus on the effectiveness of ecological measures, cannot verify the accuracy of the predicted conclusions of the EIA report, and the evaluation conclusions cannot provide a reference for the EIA prediction of subsequent similar projects.

[0006] In a first aspect, embodiments of the present invention provide a method for evaluating the environmental impact assessment results of ecological projects, comprising: Step S1: Obtain the environmental impact assessment report of the ecological project to be evaluated, and perform structured parsing of the environmental impact assessment report under semantic constraints based on natural language processing technology and a pre-constructed wind power ecological knowledge graph to obtain a structured environmental impact assessment parameter set; Step S2: Based on the structured environmental impact assessment parameter set, the wind power ecological knowledge graph, and the pre-built historical benchmark library, generate assessment base data containing environmental impact assessment quality evaluation scores and verification indicator sets; Step S3: Obtain the actual monitoring values ​​after the project is completed. Based on the assessment base data, the actual monitoring values, and the average historical benchmark values ​​of similar projects obtained from the historical benchmark database, the assessment dataset is obtained by using the triangulation method and the comprehensive evaluation method. Step S4: Based on the accumulated evaluation dataset of multiple projects, update the prediction correction information in the knowledge graph and the historical benchmark database, so that the evaluation of new projects is based on the updated knowledge graph and historical benchmark database, and perform the above steps S1 to S3.

[0007] Secondly, embodiments of this specification also provide an evaluation device for the environmental impact assessment results of ecological projects, comprising: The parsing unit is used to obtain the environmental impact assessment report of the ecological project to be evaluated. Based on natural language processing technology and a pre-built wind power ecological knowledge graph, the environmental impact assessment report is subjected to structured parsing under semantic constraints to obtain a structured environmental impact assessment parameter set. The generation unit is used to generate assessment base data containing environmental impact assessment quality assessment scores and verification indicator sets based on the structured environmental impact assessment parameter set, the wind power ecological knowledge graph, and the pre-built historical benchmark library; The evaluation unit is used to obtain the actual monitoring values ​​after the project is completed. Based on the evaluation base data, the actual monitoring values, and the average historical benchmark values ​​of similar projects obtained from the historical benchmark database, the evaluation dataset is obtained by using the triangulation method and the comprehensive evaluation method. The update unit is used to update the prediction correction information in the knowledge graph and the historical benchmark database based on the accumulated evaluation dataset of multiple projects, so that the evaluation of new projects is based on the updated knowledge graph and historical benchmark database and performs the steps of the parsing unit, generation unit and evaluation unit described above.

[0008] Thirdly, embodiments of this specification also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the method described in any embodiment of this specification.

[0009] Fourthly, embodiments of this specification also provide a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the methods described in any embodiment of this specification.

[0010] This invention provides a method and apparatus for evaluating the environmental impact assessment (EIA) results of ecological projects. First, a wind power ecological knowledge graph is constructed, and natural language processing technology is used to perform structured parsing of the EIA report under the semantic constraints of the knowledge graph, resulting in a structured EIA parameter set. Based on this, combined with the knowledge graph and a pre-built historical benchmark database, evaluation foundation data containing EIA quality assessment scores and a set of verification indicators is generated, and the indicators in the verification indicator set are assigned differentiated weights. Then, actual monitoring data after project completion is introduced, using historical benchmark values ​​of similar projects in the historical benchmark database as a third-party reference. This data, along with the EIA prediction values ​​in the verification indicator set, forms a triangular cross-validation, enabling attribution identification of deviations in each indicator. Finally, a comprehensive evaluation method is used to obtain the overall credibility level of the EIA. As the number of assessment projects accumulates, when the assessment data for similar projects reaches a certain scale, systematic prediction deviation patterns across projects are extracted. Based on this, prediction correction information is generated and stored in a knowledge graph. Upon completion of the assessment of each new project, the correction information in the knowledge graph and the statistical values ​​in the historical benchmark database are continuously updated based on all assessment data, including newly added data. This allows subsequent assessments to be performed based on the updated correction information and the historical benchmark database, forming a continuous iterative closed loop from assessment to feedback and from feedback to optimization. In this way, not only can the accuracy of environmental impact assessment (EIA) prediction results be improved, but the accuracy of EIA predictions for subsequent similar projects can also be improved. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating an evaluation method for environmental impact assessment results of an ecological project, provided in one embodiment of this specification. Figure 2 This is a hardware architecture diagram of an electronic device provided in one embodiment of this specification; Figure 3 This is a structural diagram of an assessment device for environmental impact assessment results of an ecological project, provided in one embodiment of this specification. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the embodiments in this specification clearer, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments in this specification without creative effort are within the scope of protection of this specification.

[0014] Please refer to Figure 1 This specification provides an embodiment of a method for evaluating the environmental impact assessment results of an ecological project, the method comprising: Step S1: Obtain the environmental impact assessment report of the ecological project to be evaluated, and perform structured parsing of the environmental impact assessment report under semantic constraints based on natural language processing technology and a pre-built ecological knowledge graph to obtain a structured environmental impact assessment parameter set; Step S2: Based on the structured environmental impact assessment parameter set, the electroecological knowledge graph, and the pre-built historical benchmark library, generate assessment base data containing environmental impact assessment quality evaluation scores and a set of verification indicators; wherein, the set of verification indicators includes the environmental impact assessment predicted values ​​corresponding to each verification indicator; Step S3: Obtain the actual monitoring values ​​after the project is completed. Based on the assessment base data, the actual monitoring values, and the historical benchmark average of similar projects obtained from the historical benchmark database, the assessment dataset is obtained by using the triangulation method and the comprehensive evaluation method. Step S4: Based on the accumulated evaluation datasets of multiple projects, update the prediction correction information in the knowledge graph and the historical benchmark database, so that the evaluation of new projects is based on the updated knowledge graph and historical benchmark database, and perform the above steps S1 to S3.

[0015] In this embodiment, a wind power ecological knowledge graph is first constructed, and natural language processing technology is used to perform structured parsing of the environmental impact assessment report under the semantic constraints of the knowledge graph, resulting in a structured environmental impact assessment parameter set. Based on this, the knowledge graph and a pre-built historical benchmark library are combined to generate assessment base data containing environmental impact assessment quality evaluation scores and a set of verification indicators, and the indicators in the verification indicator set are assigned differentiated weights. Then, actual monitoring data after the project is completed are introduced, and the historical benchmark values ​​of similar projects in the historical benchmark library are used as a third-party reference to form a triangular cross-validation with the environmental impact assessment prediction values ​​in the verification indicator set, thereby achieving attribution identification of deviations in each indicator. Finally, the comprehensive evaluation method is combined to obtain the comprehensive credibility level of the environmental impact assessment. As the number of assessment projects accumulates, when the assessment data for similar projects reaches a certain scale, systematic prediction deviation patterns across projects are extracted. Based on this, prediction correction information is generated and stored in a knowledge graph. Upon completion of the assessment of each new project, the correction information in the knowledge graph and the statistical values ​​in the historical benchmark database are continuously updated based on all assessment data, including newly added data. This allows subsequent assessments to be performed based on the updated correction information and the historical benchmark database, forming a continuous iterative closed loop from assessment to feedback and from feedback to optimization. In this way, not only can the accuracy of environmental impact assessment (EIA) prediction results be improved, but the accuracy of EIA predictions for subsequent similar projects can also be improved.

[0016] The following description Figure 1 The execution method of each step is shown.

[0017] Regarding step S1: In one embodiment of this specification, the ecological project is a wind power ecological project; the ecological knowledge graph is an ecological knowledge graph in the wind power field. The ecological knowledge graph is constructed based on environmental impact assessment industry standards and specifications, several environmental impact assessment approval reports, and several monitoring data of wind farm ecological projects.

[0018] In this embodiment, a wind power ecological knowledge graph is first constructed based on industry guidelines such as the "Technical Guidelines for Environmental Impact Assessment: Ecological Impact," approved wind power environmental impact assessment reports (e.g., no fewer than 200 reports), and monitoring data of ecological projects in operating wind farms (including annual monitoring reports and vegetation restoration data). The environmental impact assessment industry standards and specifications provide the basis for ontology definitions in the wind power field, including the classification of entity types, the standardization of relationship types, and the threshold ranges of key indicators. Approved wind power environmental impact assessment reports provide abundant example data and expert judgment experience, from which typical ecological indicator values, commonly used prediction methods, conventional ecological measures, and their applicable conditions for various wind farm projects can be extracted. Monitoring data of ecological projects in operating wind farms provides feedback information on actual operational effects, allowing the prior knowledge and mapping relationships in the graph to be continuously calibrated with the accumulation of practical data.

[0019] In one embodiment of this specification, the wind power ecological knowledge graph includes a schema layer and a data layer. The schema layer is used to define entity types and the relationships between entity types. Entity types include wind farm entities, ecological region entities, ecological indicator entities, ecological measure entities, bird species entities, fuzzy semantic terminology entities, and prediction method entities, as well as spatial location relationships, indicator association relationships, measure adaptation relationships, method correspondence relationships, and fuzzy semantic numerical mapping relationships between entities. The data layer stores specific domain instance knowledge generated based on the entities and relationships defined in the schema layer.

[0020] In this embodiment, the wind farm entity describes the basic attributes of the wind power project, such as wind farm type (mountain wind farm, plain wind farm, or offshore wind farm), installed capacity, number of wind turbines, and altitude range; the ecological region entity describes the ecological functional zoning (such as the North China wind energy region) and typical ecological characteristics (such as typical vegetation types) of the project area; the ecological indicator entity describes quantifiable ecological assessment indicators, such as vegetation coverage, bird collision rate, and soil erosion modulus; the ecological measures entity describes the parameters and applicable conditions of various ecological protection and restoration measures such as vegetation restoration, bird protection, and soil conservation; the bird species entity describes information on key bird species that may appear in the wind power project area, including protection level, migration season, and flight altitude characteristics; the fuzzy semantic term entity describes common qualitative expressions in environmental impact assessment reports, such as "minor impact" and "good effect," as well as the corresponding numerical mapping ranges of these expressions in different scenarios; and the prediction method entity describes the applicable conditions and common defects of various ecological impact prediction methods.

[0021] Spatial location relationship is used to describe the geographical location relationship between wind farms and ecological areas; indicator correlation relationship is used to describe the correspondence between project types and typical evaluation indicators; measure adaptation relationship is used to describe the types of ecological measures applicable under different terrain conditions and ecological zoning; method correspondence relationship is used to describe the matching relationship between each ecological indicator and the recommended prediction method; fuzzy semantic numerical mapping relationship is used to describe the quantitative correlation between fuzzy semantic terms and specific numerical ranges, thereby transforming the qualitative statements in the environmental impact assessment report into calculable quantitative indicators.

[0022] It should be noted that the specific construction methods of the above knowledge graphs are common knowledge in this field, and those skilled in the art can complete the basic construction of knowledge graphs based on general methods, so they will not be elaborated here.

[0023] In one embodiment of this specification, the historical benchmark database is a pre-built database independent of the wind power ecological knowledge graph, used to store actual monitoring data and statistical values ​​of completed wind power ecological projects. Its data sources include ecological monitoring reports during the operation period of completed wind power projects, environmental protection acceptance data upon completion, historical post-environmental impact assessment data, and scientific research monitoring data. The database is stored hierarchically according to four dimensions: wind farm type, ecological zoning, altitude zone, and installed capacity. Each original record includes at least the project number, wind farm type, ecological zoning, altitude range, installed capacity, indicator name, indicator category, assessment year, environmental impact assessment prediction value, measured value, data collection method, and data source. Based on the original data, historical benchmark mean, standard deviation, deviation rate quantiles, and sample size are pre-calculated and stored at each level.

[0024] In one embodiment of this specification, step S1 may specifically include: Based on the pattern layer constraints of the wind power ecological knowledge graph, natural language processing technology is used to perform named entity recognition and semantic relation extraction on the environmental impact assessment reports of the wind power ecological projects to be evaluated, generating an initial knowledge set consisting of multiple triples. Identify fuzzy semantic entities in the initial knowledge set, and convert qualitative fuzzy descriptions into quantitative numerical ranges with confidence levels according to the fuzzy semantic quantification mapping rules pre-stored in the knowledge graph, to obtain the quantified knowledge set. Based on the data layer of the wind power ecological knowledge graph, the quantified knowledge set is verified and integrated to obtain a structured environmental impact assessment parameter set. The structured environmental impact assessment parameter set includes basic project information, evaluation method information, data source information, measure plan information, risk assessment information, and a list of prediction indicators.

[0025] In this embodiment, before parsing the environmental impact assessment report of the wind power ecological project to be evaluated, the report needs to be preprocessed. This includes extracting the main text and tables, preserving the chapter structure, and filtering sentences related to ecological prediction, measure parameters, and risk assessment. Then, using natural language processing technology and based on the pattern layer constraints of a pre-constructed ecological knowledge graph, named entity recognition is performed on the preprocessed text. Entities are identified, and the time commitment relationship between indicators and values, the configuration relationship between measures and parameters, and the correspondence between prediction methods and indicators are extracted from the sentences containing the identified entities, forming multiple triples. For qualitatively ambiguous expressions in the environmental impact assessment report, dependency parsing is used to determine the modifiers of the ambiguous terms. Pre-stored fuzzy semantic numerical mapping relationships are queried in the knowledge graph to obtain the corresponding numerical intervals. The confidence level of the assigned numerical intervals is dynamically adjusted based on uncertain modifiers in the context, thereby converting the qualitative description into a quantitative numerical interval with confidence. Finally, based on the data layer of the wind power ecological knowledge graph, the quantified knowledge set is fused, aligned, and verified to output a structured environmental impact assessment parameter set. This parameter set includes basic project information, evaluation method information, data source information, measure plan information, risk assessment information, and a list of prediction indicators. The values ​​in the list of prediction indicators are either precise values ​​or fuzzy ranges with confidence levels.

[0026] Regarding step S2: In one embodiment of this specification, the environmental impact assessment quality evaluation score in step S2 is generated using the following method: Construct a template library for environmental impact assessment quality evaluation items; each evaluation item template includes the evaluation item name, the dimension it belongs to, the applicable conditions, the data source mapping path, and the scoring rules; the dimensions it belongs to include the reliability of basic data, the scientific nature of the evaluation method, the feasibility of ecological measures, and the comprehensiveness of risk assessment; Traverse the evaluation item template library. If the applicable conditions of the current template match the item type in the structured parameter set, extract the corresponding field values ​​from the structured parameter set according to the data source mapping path, calculate the score according to the scoring rules, and obtain the evaluation item instance. By summarizing the evaluation items belonging to the same dimension, calculating the score rate of each dimension, and obtaining the environmental impact assessment quality score.

[0027] In this embodiment of the invention, the environmental impact assessment (EIA) quality evaluation score comprehensively reviews the quality of the EIA report from four dimensions: completeness of data collection, rationality of prediction method selection, operability of measure design, and adequacy of risk identification. Each assessment item uses a three-level scoring system: full marks, partially satisfactory, and unsatisfactory. After all assessment items are scored, the score rate for each dimension is calculated.

[0028] In one embodiment of this specification, the evaluation items for the reliability dimension of basic data include: whether the bird survey covers the entire migration season (such as spring from March to May and autumn from September to November), with full marks when both spring and autumn are covered, partial marks when only one season is covered, and unqualified when neither is covered; whether the number of quadrats in the vegetation survey meets the statistical requirements, with full marks when the number of quadrats reaches the minimum requirement of the guidelines, otherwise unqualified.

[0029] The evaluation items for the scientific dimension of the evaluation method include: whether the bird collision rate prediction uses a standard collision model (full marks when a standard model is used, partial satisfaction when analogy analysis is used, and unqualified when relying solely on expert subjective judgment); and whether the vegetation restoration prediction cites successful cases from the same climate zone (full marks when the cited cases match the project's climate zone, partial satisfaction when the climate zone partially matches, and unqualified when no cases are cited or the climate zone does not match).

[0030] The assessment items for the feasibility dimension of ecological measures include: whether native species are used for slope vegetation restoration. A full score is given when the proportion of native species reaches the first threshold and no invasive species are introduced. A score within the middle range is considered partially satisfactory, while a score that is too low or introduces invasive species is considered unsatisfactory. Whether bird monitoring measures include a clear activation threshold and emergency plan is also considered. A full score is given when both are included, a score that includes only one is considered partially satisfactory, and a score that includes only a general description is considered unsatisfactory.

[0031] The assessment items for the comprehensiveness dimension of risk assessment include: whether the cumulative impact of raptors is identified (full marks when the report includes a section analyzing the cumulative impact of raptors, partial satisfaction when it is only briefly mentioned without in-depth analysis, and unsatisfactory when no relevant content is involved); whether the superposition effect of multiple wind farms is analyzed (full marks when the report includes an analysis of the superposition effect of surrounding existing and planned wind farms, partial satisfaction when only the impact of this wind farm is analyzed, and unsatisfactory when no superposition analysis is conducted).

[0032] Thus, through the aforementioned template-based automatic scoring mechanism, the traditional environmental impact assessment report quality review, which relies on subjective expert judgment, is transformed into a quantifiable and traceable automated assessment process, ensuring the consistency of assessment standards and the objectivity of assessment results.

[0033] In one embodiment of this specification, the verification index set in step S2 is generated using the following method: Extract the indicator name, predicted value, and committed time for each indicator from the list of forecast indicators in the structured parameter set; The indicators are classified according to their names to obtain a set of validation indicators; the set of validation indicators includes categories such as biological performance, habitat structure, environmental media, and management execution.

[0034] Specifically, when classifying indicators, those reflecting the response characteristics at the individual or population level, such as population density, collision rate, or diversity index, are categorized as biological performance; those reflecting the physical habitat area, cover, or spatial pattern characteristics, such as vegetation cover, suitable habitat area, or landscape fragmentation, are categorized as habitat structure; those reflecting the quality characteristics of abiotic environmental elements such as soil and water, such as soil erosion modulus, water quality compliance rate, or soil organic matter, are categorized as environmental media; and those reflecting the degree of implementation and operational status of protection and restoration measures, such as measure completion rate, facility operation rate, or replanting survival rate, are categorized as management implementation. In this way, extracting the predicted indicators scattered across various chapters of the environmental impact assessment report and classifying them according to ecological attributes not only facilitates the subsequent differentiated weighting of each verification indicator but also provides a verifiable list of indicators for the assessment in step S3.

[0035] In one embodiment of this specification, the verification index set further includes a weight value and a deviation threshold corresponding to each verification index; wherein the deviation threshold is determined based on the deviation data of similar projects in the historical benchmark library.

[0036] In one embodiment of this specification, the weight values ​​corresponding to each verification indicator are determined as follows: Based on the basic information of the project in the structured parameter set, the initial weights of various verification indicators in similar projects are queried from the knowledge graph; Obtain deviation data of similar projects from the historical benchmark database, and calculate the objective weights of various verification indicators based on the deviation data; The combination coefficients are determined based on the number of samples of similar projects. The initial weights and objective weights are then combined in a weighted manner to obtain the final weights. The proportion of the initial weights in the combination decreases as the number of samples in the historical benchmark database increases. In one embodiment of this specification, the objective weights of various verification indicators are calculated using the following method: Among them, W obj，k E represents the objective weight of the j-th verification indicator. j Let be the information entropy of the j-th verification indicator, with a value range of [0,1]. p is the sum of the information entropy redundancy of n verification metrics. ij Let be the proportion of the normalized value of the i-th item under the j-th indicator to the sum of the normalized values ​​of all items under that indicator, m be the number of similar items in the historical benchmark database (i.e., the sample size), n be the number of validation indicators, and r be the number of items under that indicator. ij Let be the normalized value of the j-th indicator of the i-th item, with a value range of [0,1].

[0037] Considering the varying relative importance of verification indicators across different project attributes, this embodiment first retrieves the initial weights of various verification indicators for similar projects from a knowledge graph based on the basic project information in the structured parameter set. These initial weights are pre-stored expert prior knowledge within the knowledge graph, allowing domain experts to empirically assess the relative importance of various verification indicators across different types of wind power projects. Next, deviation data from similar projects in the historical benchmark database is obtained—that is, the historical deviation rate between the environmental impact assessment predictions and actual values ​​of each indicator in similar projects. The objective weights for each verification indicator are then calculated using the aforementioned method. Finally, the initial weights and objective weights are weighted and combined to obtain the final weights, thereby achieving differentiated weighting based on subjective experience and objective data.

[0038] In the aforementioned differentiated weighting process, the proportion of the initial weight in the combination decreases as the sample size in the historical benchmark database increases. When the sample size of similar projects is small, the statistical regularity of the deviation data is not yet reliable, and the combination coefficient makes the initial weight proportion relatively high, with the final weight mainly relying on expert experience. As the number of evaluation projects accumulates and the sample size gradually increases, the statistical regularity of the deviation data tends to stabilize, and the combination coefficient automatically adjusts, gradually increasing the proportion of the objective weight. The final weight becomes closer and closer to the objective regularity reflected by the actual data. Therefore, by dynamically adjusting the combination ratio of subjective and objective weights, it is beneficial to achieve an adaptive optimization effect where the weight size adapts to data accumulation.

[0039] In one embodiment of this specification, the environmental impact assessment quality evaluation score and verification index set generated in step S2 can also be used to conduct phased evaluations of the environmental impact assessment results of wind power ecological projects: When a project is in the approval period, the quality of the environmental impact assessment report itself is reviewed based on the environmental impact assessment quality evaluation score, and the environmental impact assessment report with defects is given a graded warning based on the score rate of each dimension; Once the project enters the operational phase, actual monitoring data for each verification indicator is obtained based on the verification indicator set, and step S3 is executed to obtain the comprehensive credibility level of the environmental impact assessment.

[0040] In this embodiment of the invention, by conducting phased assessments of the approval period and the operation period, full-process control of the environmental impact assessment results can be achieved; during the approval period, preliminary review and graded early warning are carried out based on the environmental impact assessment quality assessment score, and reports with quality defects are intercepted before the environmental impact assessment is approved; during the operation period, post-verification is carried out based on the verification indicator set and actual monitoring data, and the comprehensive credibility level of the environmental impact assessment is output.

[0041] Regarding step S3: In one embodiment of this specification, the evaluation dataset includes the prediction bias rate of each validation indicator, the attribution type of the indicator bias, and the environmental impact assessment credibility level.

[0042] In one embodiment of this specification, step S3 includes: The environmental impact assessment predictions, actual monitoring values, and historical benchmark averages of similar projects were processed to obtain the validation data set. Based on the validation data set, calculate the prediction deviation rate and measured deviation for each indicator; The predicted deviation rate and the measured deviation are compared with the deviation threshold and the preset historical deviation threshold respectively, and the deviation attribution type of each indicator is determined according to the preset attribution logic rules. Based on the weight values ​​corresponding to each verification indicator, the deviations of each indicator are weighted and aggregated using a fuzzy comprehensive evaluation model to obtain the overall credibility level of the environmental impact assessment.

[0043] In this embodiment of the invention, the environmental impact assessment prediction values ​​from the verification indicator set, the actual monitoring values ​​obtained after the project is completed (such as those collected through remote sensing imagery, UAV aerial photography, or ground quadrat surveys), and the historical benchmark average values ​​of similar projects retrieved from the historical benchmark database are first preprocessed. Spatial interpolation and time series matching algorithms are then used to unify the three sets of data into the same time window and spatial grid, forming a comparable verification data set. Next, for the verification data set, the prediction deviation rate and measured deviation of each indicator are calculated. When the sample size of similar projects in the historical benchmark database is less than a preset minimum, the measured deviation of that indicator is not calculated temporarily, and the corresponding attribution type is marked as uncertain.

[0044] In one embodiment of this specification, the prediction deviation rate is the ratio of the difference between the actual monitored value and the environmental impact assessment prediction value to the environmental impact assessment prediction value; the measured deviation is the ratio of the difference between the actual monitored value and the historical benchmark average of similar projects to the historical benchmark value; the prediction deviation rate mainly reflects the degree to which the measured results deviate from the environmental impact assessment prediction, while the measured deviation mainly reflects the degree to which the measured results of this project deviate from the normal performance of similar projects.

[0045] In one embodiment of this specification, based on the predicted deviation rate and measured deviation of each indicator, combined with a preset deviation threshold and a historical deviation threshold, the deviation attribution type of each indicator is determined according to the following attribution logic rules: When the prediction deviation rate is greater than the deviation threshold and the measured deviation is less than or equal to the historical deviation threshold, the deviation attribution type of the verification indicator is determined to be environmental impact assessment prediction inaccuracy. When the prediction deviation rate is greater than the deviation threshold and the measured deviation is greater than the historical deviation threshold, the deviation attribution type of the verification indicator is determined to be abnormal implementation of measures or external interference. When the prediction deviation rate is less than or equal to the deviation threshold, and the measured deviation is less than or equal to the historical deviation threshold, the deviation attribution type of the verification indicator is determined to be that the environmental impact assessment prediction is basically accurate and the measured value is normal. When the prediction deviation rate is less than or equal to the deviation threshold, but the degree of deviation is greater than the historical deviation threshold, the deviation attribution type of the verification indicator is determined to be a joint deviation between the environmental impact assessment prediction and the actual measurement from the historical benchmark.

[0046] Considering that simply comparing the predicted value with the actual monitoring value cannot identify the source of the deviation, this embodiment of the invention introduces the historical benchmark average of similar projects in the historical benchmark library as a third-party reference system. Together with the environmental impact assessment predicted value and the actual monitoring value, it forms a triangular cross-validation. The attribution type is determined according to the above-mentioned attribution logic rules. This not only quantifies the accuracy of the environmental impact assessment prediction, but also accurately distinguishes whether the root cause of the deviation is the inaccuracy of the environmental impact assessment prediction, the abnormal implementation of measures, or the regional specificity. This provides a basis for the formulation of subsequent targeted rectification measures.

[0047] It should be noted that in this embodiment of the invention, both the deviation threshold and the historical deviation threshold are determined based on actual data from similar projects in the historical benchmark database. Similar projects refer to completed projects that are hierarchically matched with the current project to be evaluated across four dimensions: wind farm type, ecological zoning, altitude zone, and installed capacity. The deviation threshold is calculated by arranging all predicted deviation rates of similar projects on the same indicator from smallest to largest, and taking the upper quartile as the deviation threshold for that indicator. That is, historically, 75% of similar projects have not exceeded this value in terms of predicted deviation for that indicator; exceeding this threshold is considered a significant deviation. The historical deviation threshold is calculated by determining the degree of deviation of the measured value of each similar project relative to the historical average of similar projects, forming a set of measured-historical deviation data for all similar projects under that indicator, and similarly, taking the upper quartile as the historical deviation threshold.

[0048] In one embodiment of this specification, the overall credibility level of the environmental impact assessment is determined by the following method: Multiple confidence levels are preset, and a membership function with prediction bias rate as the independent variable is established for each level; Based on the prediction deviation rate and membership function of each verification indicator, the membership vector of each indicator to each level is obtained. By using the weight values ​​corresponding to each verification indicator, the membership vector is weighted and fuzzy synthesized to obtain the initial comprehensive membership degree for each confidence level; The credibility adjustment factor is determined based on the environmental impact assessment quality evaluation score, and the initial comprehensive membership degree is corrected using the adjustment factor to obtain the corrected comprehensive membership degree. Based on the revised comprehensive membership degree, the comprehensive credibility level of the environmental impact assessment is determined according to the preset grading rules.

[0049] In one embodiment of this specification, the initial comprehensive membership degree of each confidence level is obtained using the following method: In the formula, B(0) k Let W be the initial overall membership degree of the project to the k-th credibility level, j be the j-th verification indicator, n be the total number of verification indicators, k be the k-th credibility level, K be the number of credibility levels, and W be the number of credibility levels. j μ is the weight value of the j-th verification indicator. k (D j ) represents the membership degree of the j-th indicator to the k-th level.

[0050] In one embodiment of this specification, the modified comprehensive membership degree is as follows: In the formula, B k Let λ be the overall membership degree of the project after adjustment for the k-th credibility level, and λ be the credibility adjustment factor. The weighted average of the score rates of each dimension in the environmental impact assessment quality evaluation is taken as the overall quality score Q, with a value ranging from 0 to 1. The adjustment factor λ = Q is determined based on the overall quality score. α , where α is the adjustment intensity coefficient, and α>0.

[0051] In one embodiment of this specification, a comprehensive credibility index can be further calculated to visually reflect the overall credibility level of the environmental impact assessment results in numerical form; wherein, the comprehensive credibility index is calculated using the following method: In the formula, I is the comprehensive credibility index, and S k This is the quantitative score for the k-th level, such as highly reliable = 95, basically reliable = 85, critically reliable = 65, and unreliable = 40.

[0052] In the environmental impact assessment of wind power ecological projects, the differences between the various credibility levels are not significant. When the deviation rate of an indicator is in the critical area between two levels, it is difficult to absolutely classify it into a certain level. Therefore, in this embodiment of the invention, a membership function is used to describe the degree of conformity of the deviation rate of each indicator to different credibility levels. An initial comprehensive membership degree is obtained by integrating the results according to the weight of each indicator. The environmental impact assessment quality evaluation score generated in step S2 is introduced as a credibility adjustment factor for correction. Finally, the comprehensive credibility level of the environmental impact assessment is determined based on the corrected comprehensive membership degree. In this way, not only can the gradual transition relationship between levels be reasonably handled, but the quality of the environmental impact assessment report can also be stored in the quantitative evaluation, avoiding the problem that the quality defects of the report itself may be masked by the accidental accuracy of the predicted value.

[0053] Regarding step S4: In one embodiment of this specification, step S4 includes: Accumulate evaluation datasets from multiple projects. When the number of accumulated projects reaches a preset threshold, group the evaluation datasets according to indicator type and project attributes, and extract the statistical characteristics of prediction deviation rate in each group. When the statistical features meet the preset conditions, the prediction correction information for the corresponding group is generated and stored in the knowledge graph; For each new evaluation dataset of the same type of project, statistical features are re-extracted based on all evaluation datasets under that group, including the new data, and the stored prediction correction information is updated.

[0054] In this embodiment, after accumulating evaluation datasets for multiple projects (e.g., more than 20), the datasets are grouped according to indicator type and project attributes (e.g., wind farm type, ecological zoning, altitude zone, and installed capacity). For each group, the prediction deviation rate of all similar projects within that group is collected. Then, the mean, standard deviation, and confidence interval of the prediction deviation rate for that group are calculated. Based on these statistical characteristics, a judgment is made. For example, if the confidence interval does not include 1.0, the group is determined to have systematic prediction bias, and the mean is taken as the prediction correction coefficient for that group and stored in the knowledge graph. If the confidence interval includes 1.0, there is no systematic bias, and no correction coefficient is generated. For each new evaluation data point for a similar project, the above statistical characteristics are recalculated based on all data in that group, including the new data, and the correction coefficients and their statistical attributes stored in the knowledge graph are updated. Thus, as the sample size gradually increases, the standard error decreases, the confidence interval narrows, and the accuracy of the correction coefficients continuously improves.

[0055] In one embodiment of this specification, step S4 further includes: The actual monitoring values ​​of new projects verified in step S3 are stored in the historical benchmark database according to project attributes. For each stratification of newly stored data, the statistical values ​​of each indicator under that stratification are recalculated based on all data in that stratification; the statistical values ​​include the historical benchmark mean and the quantiles of the deviation data. When the amount of data in any layer exceeds the preset limit, the earliest record in that layer is eliminated in chronological order to update the historical benchmark database.

[0056] In this embodiment, the actual monitoring data of the new project verified in step S3 is stored in the corresponding layer of the historical benchmark database according to the project attributes. For each layer containing new data, the statistical values ​​of each verification indicator are recalculated based on all data, including the newly added data, within that layer. The statistical values ​​include at least the historical benchmark mean and the deviation data quantiles (i.e., the statistical measures on which the deviation threshold is set). When the data record volume of any layer exceeds the preset upper limit, the earliest record in that layer is eliminated in chronological order to keep the data volume of that layer within the upper limit. Through the above rolling updates, the statistical values ​​in the historical benchmark database can continuously reflect the latest industry technology level and ecological restoration effectiveness, avoiding benchmark distortion due to outdated data.

[0057] In one embodiment of this specification, step S4 further includes: automatically correcting the parameters of the comprehensive evaluation model in each preset period (e.g., every 30 projects); wherein the model parameters include at least one of the weight values ​​of each verification indicator, the deviation threshold, and the membership function parameters; during correction, a prior distribution (e.g., Gaussian distribution) is set for the model parameters to be corrected; based on the consistency data between the comprehensive credibility evaluation results output by the accumulated evaluation projects and the expert review reference ratings of the corresponding projects, the likelihood estimate and variance of the model parameters are calculated; according to the prior distribution, the likelihood estimate, and the variance, the posterior mean is calculated and used as the updated model parameter value.

[0058] In one embodiment of this specification, the posterior mean μ p The following method was used to calculate: Where μ0 is the mean of the prior distribution, Let Variance be the variance of the prior distribution. These are likelihood estimates obtained based on consistent data. denoted as the variance of the likelihood estimate.

[0059] In summary, a continuously iterative closed-loop mechanism is formed through three parallel processes: knowledge graph correction information updates, rolling updates of the historical benchmark database, and self-evolutionary updates of the evaluation model parameters. Specifically, the correction information updates feed back cross-project deviation patterns to the knowledge graph, providing corrective references when analyzing new project environmental impact assessments, thus improving the rationality of environmental impact assessment predictions from the outset. The rolling updates of the historical benchmark database replace outdated records with measured data from new projects, providing more timely reference benchmarks for subsequent assessments. The self-evolutionary updates of model parameters automatically adjust weights, thresholds, and membership function parameters with the goal of ensuring consistency between system ratings and expert judgments, enabling the evaluation model to continuously optimize with data accumulation. In this way, the system can continuously evolve as the number of evaluated projects increases, and the objectivity and accuracy of the evaluation results can be continuously improved.

[0060] like Figure 2 , Figure 3 As shown in the illustration, this specification provides an evaluation device for the environmental impact assessment results of ecological projects. The device can be implemented through software, hardware, or a combination of both. From a hardware perspective, as... Figure 2 The diagram shown is a hardware architecture diagram of an electronic device for evaluating the environmental impact assessment results of an ecological project, as provided in an embodiment of this specification. (Except for...) Figure 2 In addition to the processor, memory, network interface, and non-volatile memory shown, the electronic device in the embodiment may also include other hardware, such as a forwarding chip responsible for processing packets. Taking software implementation as an example, such as... Figure 3 As shown, a device in a logical sense is formed by the CPU of the electronic device in which it is located reading the corresponding computer program from the non-volatile memory into the memory for execution.

[0061] like Figure 3 As shown in the figure, this embodiment provides an evaluation device for the environmental impact assessment results of ecological projects, comprising: The parsing unit 300 is used to obtain the environmental impact assessment report of the ecological project to be evaluated, and to perform structured parsing of the environmental impact assessment report under semantic constraints based on natural language processing technology and a pre-built ecological knowledge graph to obtain a structured environmental impact assessment parameter set. The generation unit 302 is used to generate assessment base data containing environmental impact assessment quality evaluation scores and verification indicator sets based on the structured environmental impact assessment parameter set, the knowledge graph, and the pre-built historical benchmark library; wherein, the verification indicator set includes the environmental impact assessment predicted values ​​corresponding to each verification indicator; and the historical benchmark library includes the historical benchmark mean and deviation rate of each indicator in various ecological projects. The evaluation unit 304 is used to obtain the actual monitoring values ​​after the project is completed. Based on the evaluation basic data, the actual monitoring values ​​and the average historical benchmark values ​​of similar projects obtained from the historical benchmark database, the evaluation dataset is obtained by using the triangulation method and the comprehensive evaluation method. The update unit 306 is used to update the prediction correction information in the knowledge graph and the historical benchmark library based on the accumulated evaluation dataset of multiple projects, so that the evaluation of new projects is based on the updated knowledge graph and historical benchmark library and performs the steps of the parsing unit, generation unit and evaluation unit described above.

[0062] In the embodiments of this specification, the parsing unit 300 can be used to execute step S1 in the above method embodiments, the generation unit 302 can be used to execute step S2 in the above method embodiments, the evaluation unit 304 can be used to execute step S3 in the above method embodiments, and the updating unit 306 can be used to execute step S4 in the above method embodiments.

[0063] It is understood that the structures illustrated in the embodiments of this specification do not constitute a specific limitation on an assessment device for the environmental impact assessment results of an ecological project. In other embodiments of this specification, an assessment device for the environmental impact assessment results of an ecological project may include more or fewer components than illustrated, or combine some components, or split some components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0064] The information interaction and execution process between the modules in the above-mentioned device are based on the same concept as the method embodiments in this specification, and the specific details can be found in the descriptions in the method embodiments in this specification, so they will not be repeated here.

[0065] This specification also provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements an evaluation method for the environmental impact assessment results of an ecological project according to any embodiment of this specification.

[0066] This specification also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform an evaluation method for environmental impact assessment results of an ecological project according to any embodiment of this specification.

[0067] Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage medium.

[0068] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute a part of this specification.

[0069] Storage media embodiments for providing program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0070] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0071] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.

[0072] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0073] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as ROM, RAM, magnetic disk, or optical disk.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this specification, and are not intended to limit them. Although this specification has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this specification.

Claims

1. A method for evaluating the results of environmental impact assessments for ecological projects, characterized in that, include: Step S1: Obtain the environmental impact assessment report of the ecological project to be evaluated, and perform structured parsing of the environmental impact assessment report under semantic constraints based on natural language processing technology and a pre-built ecological knowledge graph to obtain a structured environmental impact assessment parameter set; Step S2: Based on the structured environmental impact assessment parameter set, the knowledge graph, and the pre-built historical benchmark library, generate assessment base data containing environmental impact assessment quality evaluation scores and verification indicator sets; wherein, the verification indicator set includes the environmental impact assessment predicted values ​​corresponding to each verification indicator; the historical benchmark library includes the historical benchmark mean and deviation rate of each indicator in various ecological projects; Step S3: Obtain the actual monitoring values ​​after the project is completed. Based on the assessment base data, the actual monitoring values, and the average historical benchmark values ​​of similar projects from the historical benchmark database, the assessment dataset is obtained using the triangulation method and the comprehensive evaluation method. Step S4: Based on the accumulated evaluation dataset of multiple projects, update the prediction correction information in the knowledge graph and the historical benchmark database, so that the evaluation of new projects is based on the updated knowledge graph and historical benchmark database, and perform the above steps S1 to S3.

2. The method according to claim 1, characterized in that, The ecological project is a wind power ecological project; the ecological knowledge graph is an ecological knowledge graph in the field of wind power. The ecological knowledge graph is constructed based on environmental impact assessment industry standards and specifications, several environmental impact assessment approval reports, and several wind farm ecological project monitoring data.

3. The method according to claim 1, characterized in that, The environmental impact assessment (EIA) quality evaluation score is generated using the following method: matching the structured EIA parameter set with a pre-set evaluation item template library to obtain scores for each evaluation dimension, and calculating the corresponding score rate based on the scores for each evaluation dimension; wherein, each evaluation item template in the evaluation item template library includes its respective dimension, applicable conditions, and scoring rules, and the evaluation dimensions include at least the reliability of basic data, the scientific nature of the evaluation method, the feasibility of ecological measures, and the comprehensiveness of risk assessment; and / or, The validation indicator set is generated using the following method: extracting the indicator name, predicted value, and commitment time of each indicator from the prediction indicator list in the structured parameter set, and classifying each indicator according to the indicator name to obtain the validation indicator set; wherein, the validation indicator set includes biological performance, habitat structure, environmental media, and management execution categories.

4. The method according to claim 1, characterized in that, The verification indicator set also includes the weight value and deviation threshold corresponding to each verification indicator; wherein, the deviation threshold is determined based on the deviation rate of similar projects in the historical benchmark library; The weight values ​​for each verification indicator are determined as follows: Based on the basic information of the projects in the structured parameter set, the initial weights of various verification indicators in similar projects are queried from the knowledge graph. Obtain deviation data of similar projects in the historical benchmark library, and calculate the objective weights of various verification indicators based on the deviation data; The combination coefficient is determined based on the number of samples of the same type of projects. The initial weight and the objective weight are then combined in a weighted manner to obtain the final weight. The proportion of the initial weight in the combination decreases as the number of samples in the historical benchmark database increases.

5. The method according to claim 1, characterized in that, The evaluation dataset includes the prediction deviation rate of each validation indicator, the attribution type of the indicator deviation, and the environmental impact assessment credibility level. The evaluation dataset is obtained by using the assessment baseline data, the historical benchmark average of similar projects in the historical benchmark database, and the actual monitoring values ​​obtained after the project was completed, employing the triangulation method and the comprehensive evaluation method. The dataset includes: The environmental impact assessment prediction values, actual monitoring values, and historical benchmark averages of similar projects are preprocessed to obtain a verification data set. Based on the verification data set, calculate the prediction deviation rate and measured deviation for each indicator; The predicted deviation rate and the measured deviation are compared with the deviation threshold and the preset historical deviation threshold respectively, and the deviation attribution type of each indicator is determined according to the preset attribution logic rules. Based on the weight values ​​corresponding to each verification indicator, the deviations of each indicator are weighted and aggregated through a comprehensive evaluation model to obtain the overall credibility level of the environmental impact assessment.

6. The method according to claim 5, characterized in that, The step involves weighting and aggregating the deviations of each verification indicator using a comprehensive evaluation model based on their respective weight values ​​to obtain the overall credibility level of the environmental impact assessment, including: Multiple confidence levels are preset, and a membership function with the prediction deviation rate as the independent variable is established for each level; Based on the prediction deviation rate of each verification indicator and the membership function, the initial membership vector of each indicator to each level is obtained. The credibility adjustment factor is determined based on the environmental impact assessment quality evaluation score, and the initial comprehensive membership degree is corrected using the adjustment factor to obtain the corrected comprehensive membership degree. Based on the revised comprehensive membership degree, the comprehensive credibility level of the environmental impact assessment is determined according to the preset grading rules.

7. The method according to claim 1, characterized in that, The step of updating the prediction correction information in the knowledge graph based on the accumulated evaluation dataset of multiple projects includes: The evaluation datasets of multiple projects are accumulated. When the number of accumulated projects reaches a preset threshold, the evaluation datasets are grouped according to the indicator type and project attributes, and the statistical characteristics of the prediction deviation rate in each group are extracted. When the statistical features meet the preset conditions, the prediction correction information for the corresponding group is generated and stored in the knowledge graph; For each new evaluation dataset of the same type of project, the statistical features are re-extracted based on all evaluation datasets under that group, including the new data, and the stored prediction correction information is updated; and / or, The step of updating the historical benchmark database based on the accumulated evaluation datasets from multiple projects includes: The actual monitoring values ​​of the new project after verification in step S3 are stored in the historical benchmark database according to the project attributes. For each layer of newly stored data, the statistical values ​​of each indicator under that layer are recalculated based on all the data in that layer; wherein, the statistical values ​​include the historical benchmark mean and the deviation data quantiles; When the amount of data in any layer exceeds the preset limit, the earliest record in that layer is eliminated in chronological order to update the historical benchmark database.

8. An evaluation device for the environmental impact assessment results of an ecological project, characterized in that, include: The parsing unit is used to obtain the environmental impact assessment report of the ecological project to be evaluated. Based on natural language processing technology and a pre-built wind power ecological knowledge graph, the environmental impact assessment report is subjected to structured parsing under semantic constraints to obtain a structured environmental impact assessment parameter set. The generation unit is used to generate assessment base data containing environmental impact assessment quality assessment scores and verification indicator sets based on the structured environmental impact assessment parameter set, the wind power ecological knowledge graph, and the pre-built historical benchmark library; The evaluation unit is used to obtain the actual monitoring values ​​after the project is completed. Based on the evaluation base data, the actual monitoring values, and the average historical benchmark values ​​of similar projects obtained from the historical benchmark database, the evaluation dataset is obtained by using the triangulation method and the comprehensive evaluation method. The update unit is used to update the prediction correction information in the knowledge graph and the historical benchmark library based on the accumulated evaluation dataset of multiple projects, so that the evaluation of new projects is based on the updated knowledge graph and historical benchmark library and performs the steps of the parsing unit, generation unit and evaluation unit described above.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor, when executing the computer program, implements the method as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the method of any one of claims 1-7.

Images