Mine restoration effect intelligent evaluation system and method based on multi-source data

Abstract

The invention discloses an intelligent evaluation system and method for a mine restoration effect based on multi-source data, and relates to the technical field of data analysis. Environmental data, engineering data and biological data in original mine restoration are collected; performing space-time alignment on all the data cubes; performing weighted summation on the precision, the coverage rate and the freshness of each type of repair data to obtain a quality score, and constructing a quality scoring mechanism; calculating the dynamic weight of each type of repair data, constructing a score fusion model by using the knowledge graph correction coefficient and the dynamic weight of each type of repair data, collecting a historical mine repair score and a professional lowest requirement score, and calculating to obtain a quality score threshold system; judging the fusion score by using a quality score threshold system, and outputting a score report; according to a fusion score calculation process, setting an original data node, a data processing node and a conclusion node to construct a traceability graph; and outputting the traceability map while outputting the scoring report.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a system and method for intelligently evaluating mine restoration effects based on multi-source data. Background Art

[0002] Mining activities can severely disturb the geological environment, triggering geological hazards such as rock shifting and surface subsidence. They can also damage groundwater, soil, and vegetation, leading to ecosystem degradation in mining areas. Even after mining activities cease, remaining goafs and other areas can remain "reactivated" by various factors, triggering secondary hazards and restricting the safe development and utilization of land and closed mine resources. Traditional assessments of mine restoration effectiveness rely primarily on manual observation, empirical judgment, on-site surveys, sampling and analysis, and long-term observations of vegetation recovery. These methods are not only time-consuming and labor-intensive, but also susceptible to subjective factors and lack scientific and systematic validity. Furthermore, traditional remote sensing monitoring technology also presents challenges in evaluating the effectiveness of mine ecological restoration. For example, traditional vegetation indices perform poorly in certain situations, and existing remote sensing data processing methods lack information extraction and comprehensive analysis, making it difficult to comprehensively and accurately assess mine restoration results. The rapid development of the Internet of Things, cloud computing, big data, artificial intelligence, and remote sensing technologies has provided new technical approaches and approaches for evaluating mine restoration effectiveness.

[0003] In today's mine restoration, environmental, engineering, and biological data are mostly collected by different teams, with incompatible formats, making them difficult to integrate and analyze. In addition, after the completion of traditional projects, there is a lack of data archiving and review mechanisms, making it difficult to summarize experience. Summary of the Invention

[0004] The purpose of the present invention is to provide a system and method for intelligent evaluation of mine restoration effects based on multi-source data to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions: A method for intelligently evaluating mine restoration effects based on multi-source data, comprising the following steps: S100, collecting environmental data, engineering data, and biological data from the original mine restoration; normalizing the collected restoration data, and encapsulating the normalized restoration data to obtain a data cube; Furthermore, the specific steps of encapsulating the normalized repair data to obtain a data cube are as follows: S101. Collect all restoration data from the original mine and divide the restoration data into three types: environmental data, engineering data, and biological data. Calculate the mean and standard deviation of each type of restoration data. Use the mean and standard deviation of each type of restoration data to normalize the original restoration data. The formula is: ; In the formula, x norm represents the normalized value of each repair data, x represents each original repair data, xp represents the mean value of each repair data, and xz represents the standard deviation of each repair data; S102, extract the normalized timestamp, spatial longitude and latitude coordinates and data type of each repair data; construct a data cube structure as {timestamp: time, spatial longitude and latitude coordinates: (Lat, Lon), type: (environment / engineering / biology), data: x norm}; A data cube is constructed for each normalized repair data.

[0006] Through normalization processing (such as unified units and value range scaling), data format differences are eliminated to ensure that data from different sources (such as sensors and manual surveys) can be directly compared and analyzed, thereby improving data compatibility.

[0007] The data cube encapsulates multi-dimensional data into a unified structure, facilitating subsequent spatiotemporal analysis and model calls, and allowing for rapid retrieval of restoration progress in specific areas from a three-dimensional perspective of “time-space-indicators”.

[0008] S200, dividing all data cubes into time and space, analyzing all repair data according to the divided time and space benchmarks, and performing time and space alignment on all data cubes; Furthermore, the specific steps for spatiotemporal alignment of all data cubes are as follows: S201. Using one day as the time base and a 10m×10m grid as the spatial base, the mine restoration area is divided into grid areas using the spatial base. In each grid, mine restoration data is collected according to the time base. Using the time base as the judgment standard, the collection period of each type of restoration data is extracted. The restoration data with a collection period less than the time base is regarded as high-frequency data, and the restoration data with a collection period greater than the time base is regarded as low-frequency data. Data aggregation is performed on the high-frequency data, specifically by calculating the average value of all the restoration data collected within the time base as the standard data. Data interpolation is performed on the low-frequency data, specifically by extracting the data collected twice adjacent to the low-frequency data, and using the data collected twice adjacent to fill the data not collected on any day in between. The formula is: ; In the formula, V interp Indicates the interpolation of the middle date t days, t represents the date between two consecutive data collections, t2 and t1 represent the dates of two consecutive data collections, V t2 and V t1Represents data collected twice adjacently; by performing spatiotemporal alignment on high-frequency data and low-frequency data respectively, the daily data value of each data type is obtained in each spatial reference grid.

[0009] Spatiotemporal alignment eliminates coordinate bias (such as differences in coordinate systems used by different survey teams), ensuring accurate overlay of data across periods and regions, and avoiding assessment errors caused by spatial misalignment (such as mistakenly attributing data from Area A to projects in Area B). Structured spatiotemporal data supports dynamic visualization, helping managers intuitively understand progress delays or regional differences in restoration results.

[0010] S300: For each collected data cube, calculate the accuracy, coverage, and freshness of the repaired data; perform a weighted sum of the accuracy, coverage, and freshness of each repaired data to obtain a quality score, and establish a quality scoring mechanism; Furthermore, the specific steps for building a quality scoring mechanism are as follows: S301: When collecting each type of repair data based on the time reference and the space reference, extract the collection error of each sensor, and use the sensor error to calculate the accuracy of collecting each type of repair data. The formula is: ; In the formula, Acc k Indicates the accuracy of collecting the kth type of repair data, er k represents the sensor error during the k-th data collection; When extracting repair data from each spatial reference grid, if the corresponding repair data can be extracted, it is marked as a valid grid. The coverage of each repair data is calculated using the formula: ; In the formula, Cov k Indicates the coverage of each repair data, P a Indicates the number of valid squares for each repair data, P z represents the total number of squares in the mine; Extract the delay time when collecting each type of repair data, use the time base as the unit delay time, and calculate the freshness of each type of repair data. The formula is: , in the formula, Fresh k Indicates the freshness of the k-th repair data, Ty k Indicates the delay time of the k-th repair data; S302: A quality score is obtained by weighted summing the accuracy, coverage, and freshness of each repair data. The quality scoring mechanism is constructed as follows: ; In the formula, Q krepresents the quality score of the kth repair data, α1 represents the weight of the accuracy of each repair data, α2 represents the weight of the coverage of each repair data, and α3 represents the weight of the freshness of each repair data.

[0011] Accuracy assessment can identify data errors and avoid decisions based on erroneous data; coverage ensures that no key indicators are missed; and freshness reflects whether the data is updated in a timely manner, ensuring the timeliness of the assessment.

[0012] Weighted scoring converts abstract quality into specific numerical values, facilitating horizontal comparison of the reliability of different data types and giving priority to the use of high-quality data-driven evaluation models.

[0013] S400: pre-define a knowledge graph rule base based on expert knowledge and experience, analyze and set a knowledge graph correction coefficient for each type of repair data based on the expert knowledge and knowledge graph rule base, calculate the dynamic weight of each type of repair data, construct a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each type of repair data, and output a fusion score; Furthermore, the specific steps for constructing a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data are as follows: S401. Predefine a knowledge graph rule base based on expert knowledge and experience, where the knowledge graph rule base contains risk thresholds and corresponding judgment rules for each type of repair data; analyze and set a knowledge graph correction coefficient for each type of repair data based on expert knowledge and the knowledge graph rule base; S402: Calculate the dynamic weight of each type of repair data using the following formula: ; In the formula, W k represents the dynamic weight of the kth type of restoration data, and n represents the total number of restoration data collected in the mine; S403. Build a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data. The formula is: ; In the formula, E final represents the fusion score of all repair data, KG (Q k ) represents the knowledge graph correction coefficient for each type of restoration data quality score; when collecting mine restoration data in real time, the scoring fusion model is used to calculate and output the final fusion score.

[0014] The knowledge graph rule base converts expert experience into computable logical rules (such as IF-THEN conditional statements), avoiding professional logic that may be ignored by relying solely on data-driven models (such as key parameters for slope stability in engineering mechanics), and improving the scientific nature of the assessment.

[0015] Correction coefficients and dynamic weights can flexibly adjust the importance of different indicators (for example, in water source protection areas, the weight of water quality indicators is higher than that of project progress), adapting to the environmental sensitivity characteristics and remediation objectives of different mining areas.

[0016] Dynamic weights can be automatically adjusted based on data quality. For example, when a sensor failure in a certain area causes a decrease in data freshness, the model automatically reduces the contribution of that data and instead relies on historical data or data from neighboring areas, enhancing the model's ability to resist interference.

[0017] The fusion score integrates data quality and domain knowledge to avoid misjudgment of a single indicator (such as only looking at vegetation coverage and ignoring soil pollution residues) and provide a more comprehensive assessment of restoration effects.

[0018] S500: Collect historical mine restoration scores and professional minimum requirement scores, calculate a quality score threshold system; use the quality score threshold system to judge the fusion score and output a score report; Furthermore, the specific steps for outputting the scoring report are: S501. Collect historical mine restoration scores and professional minimum requirement scores, and calculate the quality score threshold system. The formula is: ; In the formula, B base represents the quality scoring benchmark, b1 represents the historical mine restoration score coefficient, b2 represents the professional minimum requirement score coefficient, H history represents the historical mine restoration score, S standard Indicates the minimum score required for the major; divide the quality score benchmark into three levels, with 0.6 times the quality score benchmark as the qualified quality score threshold, 0.8 times the quality score benchmark as the good quality score threshold, and 1 times the quality score benchmark as the excellent quality score threshold; construct a quality score threshold system using three levels of thresholds; S502, using the quality score threshold system to judge the fusion score, when E final If the value is less than 0.6 times of the quality score benchmark, the repair is deemed unqualified. final When the quality score is between 0.6 and 0.8 times the benchmark, the repair is considered qualified. final If the repair is between 0.8 and 1 times of the quality score benchmark, it is considered to be good. final If the value is greater than 1 times the quality score benchmark, the repair is considered excellent and a score report is output.

[0019] The threshold system combines historical experience (such as the average score of similar mine restoration projects) with industry standards to provide a clear benchmark for scoring, eliminating subjective judgment. For example, if a mine's integration score falls below the minimum professional requirement, an immediate warning can be triggered, suspending subsequent work or adjusting the restoration plan.

[0020] The scoring report presents key conclusions (such as the restoration level of each area and major shortcomings indicators) in a structured form (such as tables and charts), making it easier for decision makers to quickly obtain information.

[0021] S600. According to the fusion scoring calculation process, the original data nodes, data processing nodes and conclusion nodes are respectively set to construct a traceability map; while outputting the scoring report, the traceability map is output.

[0022] Furthermore, the specific steps for outputting the traceability graph are as follows: S601. According to the fusion score calculation process, the collected original repair data is used to set the original data node, and the shape is set to be circular; the processing of the original repair data, including normalization, spatiotemporal alignment, quality scoring, and fusion scoring, is set as a processing node, and the shape is set to be square; the judgment result in the output scoring report is set as a conclusion node, and the shape is diamond; the flow direction during data processing is used as an edge to construct a traceability map; and the traceability map is output at the same time as the scoring report; S602. When the repair quality is judged to be unqualified in the scoring report, reverse tracing is performed in the traceability map to highlight the processing path of the unqualified repair data type.

[0023] The traceability graph uses a graph structure to record data sources, processing logic, and model parameters (such as knowledge graph rule versions), forming a complete "data-to-conclusion" chain. When evaluation results are questionable (such as sudden score fluctuations), the problem can be quickly traced back to the specific node (such as raw data anomalies or processing algorithm errors), improving fault location efficiency.

[0024] The traceability graph preserves historical calculation logic, providing a reference for model iteration. For example, if the correction coefficient for a certain type of data is found to be improperly set, the traceability graph can be used to query historical parameters and compare the results, optimizing the knowledge graph rule base and forming a positive cycle of "evaluation-feedback-improvement."

[0025] An intelligent evaluation system for mine restoration effects based on multi-source data, comprising a data acquisition module, a data preprocessing module, a quality scoring module, a scoring fusion module, a report generation module, and a traceability module; The data acquisition module is used to collect environmental data, engineering data and biological data in the original mine restoration; The data preprocessing module is used to normalize and align the collected mine restoration data in time and space; The quality scoring module is used to calculate the accuracy, coverage and freshness of the repair data for each collected data cube; the weighted sum of the accuracy, coverage and freshness of each repair data is used to obtain a quality score, thereby building a quality scoring mechanism; The scoring fusion module is used to calculate the dynamic weight of each repair data, build a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data, and output a fusion score; The report generation module is used to collect historical mine restoration scores and professional minimum requirement scores, calculate a quality score threshold system; use the quality score threshold system to judge the fusion score and output a score report; The traceability module is used to set the original data nodes, data processing nodes and scoring nodes respectively according to the fusion scoring calculation process to construct a traceability map; and output the traceability map at the same time as outputting the scoring report.

[0026] The data preprocessing module includes a normalization unit and a spatiotemporal alignment unit; The normalization unit is used to normalize the collected repair data and encapsulate the normalized repair data to obtain a data cube; The time-space alignment unit is used to divide all data cubes into time and space, analyze all repair data according to the divided time-space reference, and perform time-space alignment on all data cubes.

[0027] The scoring fusion module includes a dynamic weight calculation unit and a scoring fusion model unit; The dynamic weight calculation unit is used to calculate the dynamic weight using the quality score of each repair data; The scoring fusion model unit is used to construct a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data.

[0028] Compared with the prior art, the present invention has the following beneficial effects: 1. The quality scoring mechanism in this invention uses three-dimensional quantitative evaluation of accuracy, coverage, and freshness, and forms a quantitative score by weighted summation, replacing traditional manual experience judgment and improving the objectivity of data quality assessment.

[0029] 2. The scoring fusion model in this invention combines expert experience and dynamic data weights to achieve adaptive adjustment of restoration effect evaluation. The knowledge graph correction coefficient introduces industry standards to avoid the limitations of a single data dimension. For example, it integrates multiple factors such as geological structure and vegetation type to correct the score and improve the scientific nature of the evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a module distribution diagram of an intelligent evaluation system for mine restoration effects based on multi-source data according to the present invention; Figure 2 This is a schematic diagram of the steps of an intelligent evaluation method for mine restoration effects based on multi-source data of the present invention. DETAILED DESCRIPTION

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0032] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution. A method for intelligently evaluating mine restoration effects based on multi-source data, comprising the following steps: S100, collecting environmental data, engineering data, and biological data from the original mine restoration; normalizing the collected restoration data, and encapsulating the normalized restoration data to obtain a data cube; The specific steps for encapsulating the normalized repair data to obtain a data cube are: S101. Collect all restoration data from the original mine and divide the restoration data into three types: environmental data, engineering data, and biological data. Calculate the mean and standard deviation of each type of restoration data. Use the mean and standard deviation of each type of restoration data to normalize the original restoration data. The formula is: ; In the formula, x norm represents the normalized value of each repair data, x represents each original repair data, xp represents the mean value of each repair data, and xz represents the standard deviation of each repair data; S102, extract the normalized timestamp, spatial longitude and latitude coordinates and data type of each repair data; construct a data cube structure as {timestamp: time, spatial longitude and latitude coordinates: (Lat, Lon), type: (environment / engineering / biology), data: x norm}; A data cube is constructed for each normalized repair data.

[0033] Through normalization processing (such as unified units and value range scaling), data format differences are eliminated to ensure that data from different sources (such as sensors and manual surveys) can be directly compared and analyzed, thereby improving data compatibility.

[0034] The data cube encapsulates multi-dimensional data into a unified structure, facilitating subsequent spatiotemporal analysis and model calls, and allowing for rapid retrieval of restoration progress in specific areas from a three-dimensional perspective of “time-space-indicators”.

[0035] S200, dividing all data cubes into time and space, analyzing all repair data according to the divided time and space benchmarks, and performing time and space alignment on all data cubes; The specific steps for spatiotemporal alignment of all data cubes are: S201. Using one day as the time base and a 10m×10m grid as the spatial base, the mine restoration area is divided into grid areas using the spatial base. In each grid, mine restoration data is collected according to the time base. Using the time base as the judgment standard, the collection period of each type of restoration data is extracted. The restoration data with a collection period less than the time base is regarded as high-frequency data, and the restoration data with a collection period greater than the time base is regarded as low-frequency data. Data aggregation is performed on the high-frequency data, specifically by calculating the average value of all the restoration data collected within the time base as the standard data. Data interpolation is performed on the low-frequency data, specifically by extracting the data collected twice adjacent to the low-frequency data, and using the data collected twice adjacent to fill the data not collected on any day in between. The formula is: ; In the formula, V interp Indicates the interpolation of the middle date t days, t represents the date between two consecutive data collections, t2 and t1 represent the dates of two consecutive data collections, V t2 and V t1 Represents data collected twice adjacently; by performing spatiotemporal alignment on high-frequency data and low-frequency data respectively, the daily data value of each data type is obtained in each spatial reference grid.

[0036] Spatiotemporal alignment eliminates coordinate bias (such as differences in coordinate systems used by different survey teams), ensuring accurate overlay of data across periods and regions, and avoiding assessment errors caused by spatial misalignment (such as mistakenly attributing data from Area A to projects in Area B). Structured spatiotemporal data supports dynamic visualization, helping managers intuitively understand progress delays or regional differences in restoration results.

[0037] S300: For each collected data cube, calculate the accuracy, coverage, and freshness of the repaired data; perform a weighted sum of the accuracy, coverage, and freshness of each repaired data to obtain a quality score, and establish a quality scoring mechanism; The specific steps to build a quality scoring mechanism are: S301: When collecting each type of repair data based on the time reference and the space reference, extract the collection error of each sensor, and use the sensor error to calculate the accuracy of collecting each type of repair data. The formula is: ; In the formula, Acc k Indicates the accuracy of collecting the kth type of repair data, er k represents the sensor error during the k-th data collection; When extracting repair data from each spatial reference grid, if the corresponding repair data can be extracted, it is marked as a valid grid. The coverage of each repair data is calculated using the formula: ; In the formula, Cov k Indicates the coverage of each repair data, P a Indicates the number of valid squares for each repair data, P z represents the total number of squares in the mine; Extract the delay time when collecting each type of repair data, use the time base as the unit delay time, and calculate the freshness of each type of repair data. The formula is: , in the formula, Fresh k Indicates the freshness of the k-th repair data, Ty k Indicates the delay time of the k-th repair data; S302: A quality score is obtained by weighted summing the accuracy, coverage, and freshness of each repair data. The quality scoring mechanism is constructed as follows: ; In the formula, Q k represents the quality score of the kth repair data, α1 represents the weight of the accuracy of each repair data, α2 represents the weight of the coverage of each repair data, and α3 represents the weight of the freshness of each repair data.

[0038] Accuracy assessment can identify data errors and avoid decisions based on erroneous data; coverage ensures that no key indicators are missed; and freshness reflects whether the data is updated in a timely manner, ensuring the timeliness of the assessment.

[0039] Weighted scoring converts abstract quality into specific numerical values, facilitating horizontal comparison of the reliability of different data types and giving priority to the use of high-quality data-driven evaluation models.

[0040] S400: pre-define a knowledge graph rule base based on expert knowledge and experience, analyze and set a knowledge graph correction coefficient for each type of repair data based on the expert knowledge and knowledge graph rule base, calculate the dynamic weight of each type of repair data, construct a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each type of repair data, and output a fusion score; The specific steps for constructing a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data are as follows: S401. Predefine a knowledge graph rule base based on expert knowledge and experience, where the knowledge graph rule base contains risk thresholds and corresponding judgment rules for each type of repair data; analyze and set a knowledge graph correction coefficient for each type of repair data based on expert knowledge and the knowledge graph rule base; S402: Calculate the dynamic weight of each type of repair data using the following formula: ; In the formula, W k represents the dynamic weight of the kth type of restoration data, and n represents the total number of restoration data collected in the mine; S403. Build a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data. The formula is: ; In the formula, E final represents the fusion score of all repair data, KG (Q k ) represents the knowledge graph correction coefficient for each type of restoration data quality score; when collecting mine restoration data in real time, the scoring fusion model is used to calculate and output the final fusion score.

[0041] The knowledge graph rule base converts expert experience into computable logical rules (such as IF-THEN conditional statements), avoiding professional logic that may be ignored by relying solely on data-driven models (such as key parameters for slope stability in engineering mechanics), and improving the scientific nature of the assessment.

[0042] Correction coefficients and dynamic weights can flexibly adjust the importance of different indicators (for example, in water source protection areas, the weight of water quality indicators is higher than that of project progress), adapting to the environmental sensitivity characteristics and remediation objectives of different mining areas.

[0043] Dynamic weights can be automatically adjusted based on data quality. For example, when a sensor failure in a certain area causes a decrease in data freshness, the model automatically reduces the contribution of that data and instead relies on historical data or data from neighboring areas, enhancing the model's ability to resist interference.

[0044] The fusion score integrates data quality and domain knowledge to avoid misjudgment of a single indicator (such as only looking at vegetation coverage and ignoring soil pollution residues) and provide a more comprehensive assessment of restoration effects.

[0045] S500: Collect historical mine restoration scores and professional minimum requirement scores, calculate a quality score threshold system; use the quality score threshold system to judge the fusion score and output a score report; The specific steps to output the scoring report are: S501. Collect historical mine restoration scores and professional minimum requirement scores, and calculate the quality score threshold system. The formula is: ; In the formula, B base represents the quality scoring benchmark, b1 represents the historical mine restoration score coefficient, b2 represents the professional minimum requirement score coefficient, H history represents the historical mine restoration score, Sstandard Indicates the minimum score required for the major; divide the quality score benchmark into three levels, with 0.6 times the quality score benchmark as the qualified quality score threshold, 0.8 times the quality score benchmark as the good quality score threshold, and 1 times the quality score benchmark as the excellent quality score threshold; construct a quality score threshold system using three levels of thresholds; S502, using the quality score threshold system to judge the fusion score, when E final If the value is less than 0.6 times of the quality score benchmark, the repair is deemed unqualified. final When the quality score is between 0.6 and 0.8 times the benchmark, the repair is considered qualified. final If the repair is between 0.8 and 1 times of the quality score benchmark, it is considered to be good. final If the value is greater than 1 times the quality score benchmark, the repair is considered excellent and a score report is output.

[0046] The threshold system combines historical experience (such as the average score of similar mine restoration projects) with industry standards to provide a clear benchmark for scoring, eliminating subjective judgment. For example, if a mine's integration score falls below the minimum professional requirement, an immediate warning can be triggered, suspending subsequent work or adjusting the restoration plan.

[0047] The scoring report presents key conclusions (such as the restoration level of each area and major shortcomings indicators) in a structured form (such as tables and charts), making it easier for decision makers to quickly obtain information.

[0048] S600. According to the fusion scoring calculation process, the original data nodes, data processing nodes and conclusion nodes are respectively set to construct a traceability map; while outputting the scoring report, the traceability map is output.

[0049] The specific steps to output the traceability map are: S601. According to the fusion score calculation process, the collected original repair data is used to set the original data node, and the shape is set to be circular; the processing of the original repair data, including normalization, spatiotemporal alignment, quality scoring, and fusion scoring, is set as a processing node, and the shape is set to be square; the judgment result in the output scoring report is set as a conclusion node, and the shape is diamond; the flow direction during data processing is used as an edge to construct a traceability map; and the traceability map is output at the same time as the scoring report; S602. When the repair quality is judged to be unqualified in the scoring report, reverse tracing is performed in the traceability map to highlight the processing path of the unqualified repair data type.

[0050] The traceability graph uses a graph structure to record data sources, processing logic, and model parameters (such as knowledge graph rule versions), forming a complete "data-to-conclusion" chain. When evaluation results are questionable (such as sudden score fluctuations), the problem can be quickly traced back to the specific node (such as raw data anomalies or processing algorithm errors), improving fault location efficiency.

[0051] The traceability graph preserves historical calculation logic, providing a reference for model iteration. For example, if the correction coefficient for a certain type of data is found to be improperly set, the traceability graph can be used to query historical parameters and compare the results, optimizing the knowledge graph rule base and forming a positive cycle of "evaluation-feedback-improvement."

[0052] An intelligent evaluation system for mine restoration effects based on multi-source data, comprising a data acquisition module, a data preprocessing module, a quality scoring module, a scoring fusion module, a report generation module, and a traceability module; The data acquisition module is used to collect environmental data, engineering data and biological data in the original mine restoration; The data preprocessing module is used to normalize and align the collected mine restoration data in time and space; The quality scoring module is used to calculate the accuracy, coverage and freshness of the repair data for each collected data cube; the weighted sum of the accuracy, coverage and freshness of each repair data is used to obtain a quality score, thereby building a quality scoring mechanism; The scoring fusion module is used to calculate the dynamic weight of each repair data, build a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data, and output a fusion score; The report generation module is used to collect historical mine restoration scores and professional minimum requirement scores, calculate a quality score threshold system; use the quality score threshold system to judge the fusion score and output a score report; The traceability module is used to set the original data nodes, data processing nodes and scoring nodes respectively according to the fusion scoring calculation process to construct a traceability map; and output the traceability map at the same time as outputting the scoring report.

[0053] The data preprocessing module includes a normalization unit and a spatiotemporal alignment unit; The normalization unit is used to normalize the collected repair data and encapsulate the normalized repair data to obtain a data cube; The time-space alignment unit is used to divide all data cubes into time and space, analyze all repair data according to the divided time-space reference, and perform time-space alignment on all data cubes.

[0054] The scoring fusion module includes a dynamic weight calculation unit and a scoring fusion model unit; The dynamic weight calculation unit is used to calculate the dynamic weight using the quality score of each repair data; The scoring fusion model unit is used to construct a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data.

[0055] Example: Examples of building a knowledge graph rule base based on domain expertise include: # Geological Rules if G_slope>25 degrees: Soil and water loss risk += 20% # Ecosystem Rules if B_shannon<2.5 and V_NDVI<0.6: Ecological stability = "poor" Among them, G_slope represents the terrain slope, B_shannon represents the Shannon-Wiener diversity index; If the rule base shows that the regional slope is >25 degrees, KG=0.8 to reduce the score; Assuming that the weight of terrain slope data in the collected mine restoration data is 0.6, the quality score is 70, and the knowledge graph correction coefficient is 0.8, the calculated score of terrain slope in the scoring fusion model is 33.6; It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for intelligent evaluation of mine restoration effects based on multi-source data, characterized by: The method comprises the following steps: S100, collecting environmental data, engineering data, and biological data from the original mine restoration; normalizing the collected restoration data, and encapsulating the normalized restoration data to obtain a data cube; S200, dividing all data cubes into time and space, analyzing all repair data according to the divided time and space benchmarks, and performing time and space alignment on all data cubes; S300: For each collected data cube, calculate the accuracy, coverage, and freshness of the repaired data; perform a weighted sum of the accuracy, coverage, and freshness of each repaired data to obtain a quality score, and establish a quality scoring mechanism; S400: pre-define a knowledge graph rule base based on expert knowledge and experience, analyze and set a knowledge graph correction coefficient for each type of repair data based on the expert knowledge and knowledge graph rule base, calculate the dynamic weight of each type of repair data, construct a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each type of repair data, and output a fusion score; S500: Collect historical mine restoration scores and professional minimum requirement scores, calculate a quality score threshold system; use the quality score threshold system to judge the fusion score and output a score report; S600. According to the fusion scoring calculation process, the original data nodes, data processing nodes and conclusion nodes are respectively set to construct a traceability map; while outputting the scoring report, the traceability map is output.

2. The method for intelligently evaluating mine restoration effects based on multi-source data according to claim 1, characterized in that: The specific steps of encapsulating the normalized repair data to obtain a data cube in S100 are: S101. Collect all restoration data from the original mine and divide the restoration data into three types: environmental data, engineering data, and biological data. Calculate the mean and standard deviation of each type of restoration data. Use the mean and standard deviation of each type of restoration data to normalize the original restoration data. The formula is: ； In the formula, x norm represents the normalized value of each repair data, x represents each original repair data, xp represents the mean value of each repair data, and xz represents the standard deviation of each repair data; S102, extract the normalized timestamp, spatial longitude and latitude coordinates and data type of each repair data; construct a data cube structure as {timestamp: time, spatial longitude and latitude coordinates: (Lat, Lon), type: (environment / engineering / biology), data: x norm }; A data cube is constructed for each normalized repair data.

3. The method for intelligently evaluating mine restoration effects based on multi-source data according to claim 1, characterized in that: The specific steps of performing spatiotemporal alignment on all data cubes in S200 are as follows: S201. Using one day as the time base and a 10m×10m grid as the spatial base, the mine restoration area is divided into grid areas using the spatial base. In each grid, mine restoration data is collected according to the time base. Using the time base as the judgment standard, the collection period of each type of restoration data is extracted. The restoration data with a collection period less than the time base is regarded as high-frequency data, and the restoration data with a collection period greater than the time base is regarded as low-frequency data. Data aggregation is performed on the high-frequency data, specifically by calculating the average value of all the restoration data collected within the time base as the standard data. Data interpolation is performed on the low-frequency data, specifically by extracting the data collected twice adjacent to the low-frequency data, and using the data collected twice adjacent to fill the data not collected on any day in between. The formula is: ； In the formula, V interp Indicates the interpolation of the middle date t days, t represents the date between two consecutive data collections, t2 and t1 represent the dates of two consecutive data collections, V t2 and V t1 Represents data collected twice adjacently; by performing spatiotemporal alignment on high-frequency data and low-frequency data respectively, the daily data value of each data type is obtained in each spatial reference grid.

4. The method for intelligently evaluating mine restoration effects based on multi-source data according to claim 3 is characterized in that: The specific steps of constructing the quality scoring mechanism in S300 are: S301: When collecting each type of repair data based on the time reference and the space reference, extract the collection error of each sensor, and use the sensor error to calculate the accuracy of collecting each type of repair data. The formula is: ； In the formula, Acc k Indicates the accuracy of collecting the kth type of repair data, er k represents the sensor error during the k-th data collection; When extracting repair data from each spatial reference grid, if the corresponding repair data can be extracted, it is marked as a valid grid. The coverage of each repair data is calculated using the formula: ； In the formula, Cov k Indicates the coverage of each repair data, P a Indicates the number of valid squares for each repair data, P z represents the total number of squares in the mine; Extract the delay time when collecting each type of repair data, use the time base as the unit delay time, and calculate the freshness of each type of repair data. The formula is: , in the formula, Fresh k Indicates the freshness of the k-th repair data, Ty k Indicates the delay time of the k-th repair data; S302: A quality score is obtained by weighted summing the accuracy, coverage, and freshness of each repair data. The quality scoring mechanism is constructed as follows: ； In the formula, Q k represents the quality score of the kth repair data, α1 represents the weight of the accuracy of each repair data, α2 represents the weight of the coverage of each repair data, and α3 represents the weight of the freshness of each repair data.

5. The method for intelligently evaluating mine restoration effects based on multi-source data according to claim 4 is characterized in that: The specific steps of constructing a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data in S400 are as follows: S401. Predefine a knowledge graph rule base based on expert knowledge and experience, where the knowledge graph rule base contains risk thresholds and corresponding judgment rules for each type of repair data; analyze and set a knowledge graph correction coefficient for each type of repair data based on expert knowledge and the knowledge graph rule base; S402: Calculate the dynamic weight of each type of repair data using the following formula: ； In the formula, W k represents the dynamic weight of the kth type of restoration data, and n represents the total number of restoration data collected in the mine; S403. Build a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data. The formula is: ； In the formula, E final represents the fusion score of all repair data, KG (Q k ) represents the knowledge graph correction coefficient for each repair data quality score; When collecting mine restoration data in real time, the scoring fusion model is used to calculate and output the final fusion score.

6. The method for intelligently evaluating mine restoration effects based on multi-source data according to claim 5, characterized in that: The specific steps of outputting the scoring report in S500 are: S501. Collect historical mine restoration scores and professional minimum requirement scores, and calculate the quality score threshold system. The formula is: ； In the formula, B base represents the quality scoring benchmark, b1 represents the historical mine restoration score coefficient, b2 represents the professional minimum requirement score coefficient, H history represents the historical mine restoration score, S standard Indicates the minimum score required for the major; divide the quality score benchmark into three levels, with 0.6 times the quality score benchmark as the qualified quality score threshold, 0.8 times the quality score benchmark as the good quality score threshold, and 1 times the quality score benchmark as the excellent quality score threshold; construct a quality score threshold system using three levels of thresholds; S502, using the quality score threshold system to judge the fusion score, when E final If the value is less than 0.6 times of the quality score benchmark, the repair is deemed unqualified. final When the quality score is between 0.6 and 0.8 times the benchmark, the repair is considered qualified. final If the repair is between 0.8 and 1 times of the quality score benchmark, it is considered to be good. final If the value is greater than 1 times the quality score benchmark, the repair is considered excellent and a score report is output.

7. The method for intelligently evaluating mine restoration effects based on multi-source data according to claim 6, characterized in that: The specific steps of outputting the traceability map in S600 are: S601. According to the fusion score calculation process, the collected original repair data is used to set the original data node, and the shape is set to be circular; the processing of the original repair data, including normalization, spatiotemporal alignment, quality scoring, and fusion scoring, is set as a processing node, and the shape is set to be square; the judgment result in the output scoring report is set as a conclusion node, and the shape is diamond; the flow direction during data processing is used as an edge to construct a traceability map; and the traceability map is output at the same time as the scoring report; S602. When the repair quality is judged to be unqualified in the scoring report, reverse tracing is performed in the traceability map to highlight the processing path of the unqualified repair data type.

8. An intelligent assessment system for mine restoration effects based on multi-source data, characterized by: The intelligent evaluation system for mine restoration effects includes data acquisition module, data preprocessing module, quality scoring module, scoring fusion module, report generation module and traceability module; The data acquisition module is used to collect environmental data, engineering data and biological data in the original mine restoration; The data preprocessing module is used to normalize and align the collected mine restoration data in time and space; The quality scoring module is used to calculate the accuracy, coverage and freshness of the repair data for each collected data cube; the weighted sum of the accuracy, coverage and freshness of each repair data is used to obtain a quality score, thereby building a quality scoring mechanism; The scoring fusion module is used to calculate the dynamic weight of each repair data, build a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data, and output a fusion score; The report generation module is used to collect historical mine restoration scores and professional minimum requirement scores, calculate a quality score threshold system; use the quality score threshold system to judge the fusion score and output a score report; The traceability module is used to set the original data nodes, data processing nodes and scoring nodes respectively according to the fusion scoring calculation process to construct a traceability map; and output the traceability map at the same time as outputting the scoring report.

9. The intelligent assessment system for mine restoration effects based on multi-source data according to claim 8, characterized in that: The data preprocessing module includes a normalization unit and a spatiotemporal alignment unit; The normalization unit is used to normalize the collected repair data and encapsulate the normalized repair data to obtain a data cube; The time-space alignment unit is used to divide all data cubes into time and space, analyze all repair data according to the divided time-space reference, and perform time-space alignment on all data cubes.

10. The intelligent assessment system for mine restoration effects based on multi-source data according to claim 8, characterized in that: The scoring fusion module includes a dynamic weight calculation unit and a scoring fusion model unit; The dynamic weight calculation unit is used to calculate the dynamic weight using the quality score of each repair data; The scoring fusion model unit is used to construct a scoring fusion model using the knowledge graph correction coefficient and dynamic weight of each repair data.

Images