An open-pit coal mine disturbance year identification method fusing multiple ground object features

Abstract

This invention provides a method for identifying the year of open-pit coal mine disturbance by integrating multiple land cover features, belonging to the field of mine ecological monitoring technology. It involves acquiring long-term remote sensing images of open-pit coal mines, performing radiometric calibration and cloud removal preprocessing to establish a high-quality annual dataset; calculating the spectral indices of vegetation, bare soil, and bare coal; and constructing a novel open-pit coal mine disturbance index, MDI_NBC, by comprehensively considering the changes in the spectral indices of these three land cover features during open-pit coal mining; comparing the index changes in undisturbed areas to determine the MDI_NBC threshold for distinguishing mining disturbances; and applying a time-series segmentation algorithm with optimized parameters to conduct long-term monitoring of open-pit coal mine disturbances in different vegetation gradient zones, identifying the year of open-pit coal mine disturbances. This invention achieves universal monitoring of open-pit coal mine disturbances across vegetation gradient zones, effectively distinguishing vegetation disturbances caused by climate change, and improving the efficiency and accuracy of identifying the year of open-pit coal mine disturbances.

Description

Technical Field

[0001] This invention provides a method for identifying the year of disturbance in open-pit coal mines by integrating multiple features of land cover, which belongs to the field of mine ecological monitoring technology. Background Technology

[0002] Open-pit coal mining is often accompanied by large-scale topsoil stripping and vegetation destruction, causing irreversible disturbances to regional soil structure, hydrological processes and ecosystems, and seriously threatening regional ecological security.

[0003] Remote sensing technology, with its advantages of long-term time series, large-scale coverage, non-contact operation, and high precision, has overcome the limitations of traditional methods such as field surveys and visual interpretation, which are costly, inefficient, and have limited monitoring range. It has become a core technology for monitoring disturbances in open-pit coal mines and is widely used in various scenarios, including long-term disturbance and remediation monitoring in mining areas. However, due to limitations imposed by environmental background, remote sensing indices, and data sources, the existing technology system faces numerous insurmountable technical challenges in monitoring disturbances in open-pit coal mines in different vegetation gradient zones, especially in arid and semi-arid regions. These challenges have become key issues restricting mine ecological supervision and the construction of green mines.

[0004] Given that open-pit coal mining is accompanied by changes in the characteristics of various land features, and in response to the limitations of existing technologies, there is an urgent need to innovate a monitoring method for open-pit coal mining disturbance that integrates multi-source land feature characteristics and is applicable to cross vegetation gradient zones. This method would fill the gap in monitoring technology in arid and semi-arid regions and achieve full-range, long-term, and high-precision monitoring of open-pit coal mining disturbance in different vegetation gradient zones.

[0005] In existing technologies, monitoring temporal disturbances in open-pit coal mines based on vegetation indices such as NDVI and EVI, and applying time-series segmentation algorithms, is currently the mainstream method. The core logic of this technical solution is to use changes in vegetation indices as the primary criterion for determining mining disturbances. However, this technology has the following drawbacks:

[0006] (1) Limited application scope. The core monitoring object of vegetation indices such as NDVI is surface vegetation. However, in the arid and semi-arid open-pit coal mining areas in the west, the vegetation signal is weak, and mining activities such as topsoil stripping and coal mining in open-pit mines cannot be identified, which cannot meet the monitoring needs of arid and semi-arid areas;

[0007] (2) Poor resistance to interference. The vegetation index is also highly sensitive to climate change. Extreme precipitation, drought and other climate factors can cause natural abrupt changes in the vegetation index. These abrupt changes are difficult to distinguish from the changes in the vegetation index caused by mining disturbances, which can easily lead to misjudgment of monitoring results. In particular, in semi-arid areas with frequent climate fluctuations, the monitoring accuracy is greatly reduced.

[0008] In addition, existing technologies also include constructing remote sensing indices such as bare coal or coal gangue based on medium-to-high resolution remote sensing imagery, combining mining area feature characteristics with object-oriented segmentation, and identifying open-pit coal mine disturbance boundaries at specific time points. The core of this technical solution is to achieve fine segmentation of mining area features through high-resolution imagery. This technology has the following technical drawbacks:

[0009] (1) Short image retrospective period. The retrospective period of medium and high resolution remote sensing images is mostly after 2010, which is short and cannot reveal the long-term mining disturbance process since the beginning of mining in open coal mines, making it difficult to support long-term research on ecological changes in mining areas;

[0010] (2) Indices are limited. Due to factors such as the geographical background, mining methods, and ecological conditions of the mining area, single land feature indices lack universality and cannot be directly applied to other mining areas. Summary of the Invention

[0011] To address the aforementioned technical problems, this invention provides a method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics, aiming to achieve the following objectives:

[0012] (1) In view of the problem that traditional vegetation index-based methods are limited in application and have poor anti-interference ability in arid and semi-arid mining areas, this invention constructs a comprehensive remote sensing index that integrates vegetation, bare soil and bare coal, and combines it with time series segmentation algorithm to establish a method for monitoring open-pit coal mining disturbances in different vegetation gradient zones.

[0013] (2) In view of the problems of short back-up period and incomplete monitoring of medium and high resolution images, this invention is based on Landsat long time series images and optimizes the time series segmentation algorithm parameters to realize large-scale, long-term and high-precision monitoring of disturbance years in open coal mines.

[0014] The specific technical solution provided by this invention is as follows:

[0015] A method for identifying the year of open-pit coal mine disturbance by integrating multiple land cover features is proposed. This method acquires long-term remote sensing images of open-pit coal mines, performs radiometric calibration and cloud removal preprocessing to establish a high-quality annual dataset. Spectral indices of vegetation, bare soil, and bare coal are calculated. By integrating the changes in these indices during open-pit coal mining, a novel open-pit coal mine disturbance index, MDI_NBC, is constructed. The index changes in undisturbed areas are compared to determine the MDI_NBC threshold for distinguishing mining disturbances. Finally, a time-series segmentation algorithm with optimized parameters is applied to conduct long-term monitoring of open-pit coal mine disturbances in different vegetation gradient zones, identifying the year of open-pit coal mine disturbance.

[0016] Specifically, the following steps are included:

[0017] Step 1: Data collection and preprocessing;

[0018] Long-term Landsat surface reflectance images of open-pit coal mines were collected, and preprocessing such as radiometric calibration and atmospheric correction was carried out. At the same time, the CFMASK function was applied to mask invalid ground features such as clouds, shadows, and snow to ensure that the mining area is covered by high-quality images every year and to establish an annual high-quality image dataset.

[0019] Data on the distribution of impermeable surfaces and water bodies were extracted from the latest publicly available land use data to cover open-pit mines and eliminate interference from non-mining human activities on the monitoring of mine disturbances.

[0020] Vectorize the open-pit coal mine boundary from the open-pit coal mine image dataset to serve as the research scope for open-pit coal mine mining disturbance at the current time point;

[0021] Unify the coordinate system, resolution, and spatial range of all data to ensure comparability of data from different periods and of different types.

[0022] Step 2: Constructing the disturbance index;

[0023] Spectral indices for three land cover types—vegetation, bare soil, and bare coal—were established through band calculations. Vegetation was characterized by NDVI, bare soil by BI, and bare coal by OCFI.

[0024] A novel open-pit coal mining disturbance index, MDI_NBC, is constructed by integrating the index changes of three core land features during open-pit coal mining: vegetation degradation, bare soil exposure, and bare coal exposure. The specific calculation process is shown in Formula 1. ,

[0025] In the formula, MDI_NBC y Let Max be the disturbance index for year y. [y,最新一年] OCFI is the maximum value of OCFI from year y to the latest year, Max. [1990,最新一年] MBI represents the maximum value of MBI from 1990 to the latest year. [y,最新一年] NDVI is the minimum NDVI value from year y to the latest year, OCFI y For the frequency of bare coal in year y, Max [1990,y] MBI represents the maximum MBI from 1990 to year y, NDVI y The mean NDVI value over y years.

[0026] The NDVI calculation process is shown in Formulas 2 and 3: ;

[0027] The MBI calculation process is shown in Formulas 4 and 5: , ;

[0028] The OCFI calculation process is shown in formulas 6 and 7: , ,

[0029] In the formula, ρ R ρ represents the red band reflectivity. NIR ρ represents the reflectivity in the near-infrared band. SWIR1 ρ represents the reflectivity of shortwave infrared 1. SWIR2 This represents the reflectivity of shortwave infrared 2.

[0030] Step 3: Determine the perturbation threshold;

[0031] The disturbance impact range of open-pit coal mining is defined as 2km, and the boundary of the open-pit coal mine is buffered by 4km. The 2-4km buffer ring is regarded as the undisturbed area.

[0032] Randomly select points inside the open-pit mine and in undisturbed areas to track the temporal change trajectory of the MDI_NBC index at each sample point, statistically analyze the trend, magnitude and duration of index changes in each region, establish disturbance source labels, and compare the numerical frequency distribution of different disturbance sources.

[0033] 80% of each category was randomly selected as training samples and 20% as validation samples. The random forest method was applied to determine the threshold for distinguishing between coal mining disturbances and non-coal mining disturbances.

[0034] Step 4: Monitoring coal mining disturbance;

[0035] Based on random sample points within the mining area, a time series segmentation algorithm based on MDI_NBC is constructed. The root mean square error of the sample points is calculated, as shown in Formula 8. The minimum value is used as the reference standard, and the corresponding parameter combination is selected as the optimal parameter for the time series segmentation algorithm in the open-pit coal mine area. ;

[0036] In the formula, y i Let i be the true value of sample point i. Let be the predicted value of sample point i, and n be the total number of sample points.

[0037] The optimized time series segmentation algorithm was applied to the time series remote sensing data of the mining area to reconstruct the MDI_NBC time series change trajectory of the entire mining area pixel by pixel. Disturbances in which the change trend, amplitude and time of each pixel exceeded the set threshold were regarded as open-pit coal mining disturbances, and the year information of the disturbance was identified.

[0038] Based on the preliminary identification results of the perturbed years, spatial filtering is carried out to reduce noise in the year information.

[0039] Based on the disturbance area in each year, a random stratified sampling method with unequal proportions is adopted to generate sample points in each mining area in each year, ensuring that the spatial interval of the sample points is not less than 100 meters, thereby reducing the spatial autocorrelation of the sample points.

[0040] High-resolution imagery was acquired from online maps and combined with Landsat time-series image tiling tools to determine the true year of disturbance for each sample point.

[0041] The disturbance years identified for each sample point are aggregated with the results of three consecutive years (including adjacent years) into one category to construct a confusion matrix for the disturbance years of each open-pit mine. The following calculations are then performed:

[0042] User precision: ;

[0043] Producer precision: ;

[0044] Overall accuracy: ;

[0045] F1 score: ;

[0046] In the formula, P ii The number of correctly classified samples, p i· p represents the number of samples predicted to be class i. ·i This represents the number of real samples of class i, m is the number of classes, and N is the total number of samples.

[0047] Beneficial effects of the technical solution of this invention

[0048] (1) Universal monitoring across vegetation gradient zones has been achieved. The MDI_NBC index integrates the characteristics of three land features: vegetation, bare soil, and bare coal. It can capture mining disturbance signals in vegetated areas and accurately identify core mining activities such as topsoil stripping and bare coal exposure in non-vegetated areas. It effectively solves the problem of monitoring failure of traditional vegetation indices in arid and semi-arid regions, and realizes unified and accurate monitoring of mining areas with different ecological backgrounds, significantly improving its universality.

[0049] (2) Improved the accuracy of open-pit coal mining disturbance identification. Compared with the traditional NDVI index, the MDI_NBC index has significantly improved identification accuracy. At the same time, the fusion of bare soil and bare coal characteristics effectively reduces the interference of natural abrupt changes in vegetation index caused by extreme precipitation, drought and other climatic factors on the monitoring results, greatly reducing the misjudgment rate and significantly improving the monitoring accuracy and stability;

[0050] (3) Comprehensive monitoring of open-pit coal mines over a long period of time has been achieved. Based on the Landsat series of long-term images, long-term monitoring of mining disturbances in open-pit coal mines has been achieved, which makes up for the short backtracking period of medium and high resolution images, and integrates the features of ground features throughout the mining process, avoiding the one-sidedness of monitoring with a single index. Attached Figure Description

[0051] Figure 1 This is a flowchart of the process of the present invention;

[0052] Figure 2 This is a schematic diagram illustrating the changes in spectral indices of multi-source ground cover at vegetation-covered sampling points, as shown in the example.

[0053] Figure 3 This is a schematic diagram illustrating the changes in the spectral indices of multi-source ground cover at unvegetated sampling points, as shown in the example.

[0054] Figure 4 This is a schematic diagram illustrating the threshold distinction between mining disturbances and non-mining disturbances in an open-pit mine, as shown in the example.

[0055] Figure 5 This is an example of the initial results for open-pit coal mine disturbance years based on the MDI_NBC index;

[0056] Figure 6 This is a spatial filtering result of open-pit coal mine disturbance years based on the MDI_NBC index, as an example.

[0057] Figure 7 The image shows the spatial filtering results of open-pit coal mine disturbance years based on the NDVI index, as an example. Detailed Implementation

[0058] The technical solution of the present invention will be described in conjunction with the accompanying drawings.

[0059] This invention provides a method for identifying the year of open-pit coal mine disturbance by integrating multiple land cover features. It acquires long-term remote sensing images of open-pit coal mines, performs radiometric calibration and cloud removal preprocessing to establish a high-quality annual dataset; calculates the spectral indices of vegetation, bare soil, and bare coal; and constructs a novel open-pit coal mine disturbance index (MDI_NBC) by integrating the changes in the spectral indices of these three land cover features during open-pit coal mining; compares the index changes in undisturbed areas to determine the MDI_NBC threshold for distinguishing mining disturbances; and applies a time-series segmentation algorithm with optimized parameters to conduct long-term monitoring of open-pit coal mine disturbances in different vegetation gradient zones, identifying the year of open-pit coal mine disturbance information, such as... Figure 1 As shown, the specific steps are as follows:

[0060] Step 1: Data Collection and Preprocessing

[0061] We collected Landsat 5 / 7 / 8 / 9 surface reflection images of open-pit coal mines from May to September 1990 to 2024, carried out preprocessing such as radiometric calibration and atmospheric correction, and applied the CFMASK function to mask invalid ground features such as clouds, shadows, and snow to ensure that the mining area is covered by high-quality images every year and to establish an annual high-quality image dataset.

[0062] Data on impermeable surfaces and water bodies were extracted from the 2024 CLCD (China Land Cover Dataset) to cover the open-pit mine area and eliminate interference from non-mining human activities on the monitoring of mine disturbance.

[0063] Vectorize the open-pit coal mine boundary from the open-pit coal mine image dataset to serve as the research scope for open-pit coal mine mining disturbance at the current time point;

[0064] Unify the coordinate system, resolution, and spatial range of all data to ensure comparability of data from different periods and of different types.

[0065] Step 2: Construction of the Disturbance Index

[0066] Spectral indices for three land cover types—vegetation, bare soil, and bare coal—were established using band operations on Landsat 5 / 7 / 8 / 9 images. Vegetation was represented by NDVI, bare soil by BI, and bare coal by OCFI.

[0067] Open-pit coal mining damages surface vegetation, leading to a decrease in NDVI (Natural Density Index), while the corresponding bare soil index shows a temporary increase. As underground coal is exposed, the bare coal index also shows an increasing trend, while the bare soil index decreases, and the vegetation index remains at a low level. To address the limitations of existing single-index monitoring, this paper integrates the changes in three core land cover characteristics during open-pit coal mining: vegetation degradation, bare soil exposure, and bare coal exposure. By combining the characteristics of vegetation, bare soil, and bare coal, a new open-pit coal mining disturbance index, MDI_NBC, is constructed. This index can simultaneously capture mining disturbance signals in both vegetated and non-vegetated areas, effectively solving the problem of monitoring failure in arid and semi-arid regions. Figure 2 and Figure 3 The specific calculation process is shown in Formula 1: ;

[0068] In the formula, MDI_NBC y Let Max be the disturbance index for year y. [y,2024] OCFI represents the maximum OCFI value from year y to 2024. [1990,2024] MBI represents the maximum MBI from 1990 to 2024. [y,2024] NDVI is the minimum NDVI value from year y to 2024, OCFI y For the frequency of bare coal in year y, Max [1990,y] MBI represents the maximum MBI from 1990 to year y, NDVI y The mean NDVI value over y years.

[0069] The NDVI calculation process is shown in Formula 2-3: , ;

[0070] The MBI calculation process is shown in formula 4-5: , ;

[0071] The OCFI calculation process is shown in formula 6-7: , ;

[0072] In the formula, ρ R ρ represents the red band reflectivity. NIR ρ represents the reflectivity in the near-infrared band. SWIR1 ρ represents the reflectivity of shortwave infrared 1. SWIR2 This represents the reflectivity of shortwave infrared 2.

[0073] Step 3: Determine the perturbation threshold;

[0074] Multi-ring buffers are implemented at the boundary of open-pit coal mines, with a 2km range defined as the impact range of open-pit coal mine mining disturbances, and a 2-4km buffer ring defined as the undisturbed area.

[0075] Randomly select points inside the open-pit mine and in undisturbed areas to track the temporal change trajectory of the MDI_NBC index at each sample point, statistically analyze the trend, magnitude and duration of index changes in each region, establish disturbance source labels, and compare the numerical frequency distribution of different disturbance sources.

[0076] 80% of each category was randomly selected as training samples, and 20% as validation samples. The random forest method was applied to determine the distinction threshold between mining disturbances and non-mining disturbances. Figure 4 As shown.

[0077] Step 4: Monitoring coal mining disturbance;

[0078] Based on random sample points within the mining area, a time series segmentation algorithm based on MDI_NBC is constructed. The root mean square error of the sample points is calculated (Formula 8). The minimum value is used as the reference standard, and the corresponding parameter combination is selected as the optimal parameter for the time series segmentation algorithm in the open-pit coal mine area. ;

[0079] In the formula, y i Let i be the true value of sample point i. Let be the predicted value of sample point i, and n be the total number of sample points.

[0080] The optimized time-series segmentation algorithm was applied to the time-series remote sensing data of the mining area. The MDI_NBC time-series change trajectory of the entire mining area was reconstructed pixel by pixel. Disturbances where the change trend, amplitude, and time of each pixel exceeded a set threshold were considered open-pit coal mining disturbances, and the year information of the disturbances was identified. Figure 5 ;

[0081] Based on the preliminary identification results of the perturbed years, spatial filtering is performed to reduce noise in the year information, such as... Figure 6 ;

[0082] Based on the disturbance area in each year, a random stratified sampling method with unequal proportions is adopted to generate sample points in each mining area in each year, ensuring that the spatial interval of the sample points is not less than 100 meters, thereby reducing the spatial autocorrelation of the sample points.

[0083] High-resolution images were obtained from online maps such as Google Earth and Tianditu, and Landsat time-series image tiling tool was used to determine the true year of disturbance for each sample point.

[0084] The results of the perturbation years identified for each sample point and the adjacent years (a total of three years) are aggregated into one class to construct a confusion matrix for the perturbation years of each open-pit mine. User accuracy (Formula 9), producer accuracy (Formula 10), overall accuracy (Formula 11), and F1 score (Formula 12) are calculated, and the perturbation identification performance based on MDI_NBC and NDVI is compared. Figure 7 .

[0085] , , , ,

[0086] In the formula, P ii The number of correctly classified samples, p i· p represents the number of samples predicted to be class i. ·i This represents the number of real samples of class i, m is the number of classes, and N is the total number of samples.

Claims

1. A method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics, characterized in that, Long-term remote sensing images of open-pit coal mines were acquired, and after radiometric calibration and cloud removal preprocessing, a high-quality annual dataset was established. Spectral indices of vegetation, bare soil, and bare coal were calculated. By integrating the changes in the spectral indices of these three land features during open-pit coal mining, a novel open-pit coal mining disturbance index, MDI_NBC, was constructed. The index changes in undisturbed areas were compared to determine the MDI_NBC threshold for distinguishing mining disturbances. A time-series segmentation algorithm with optimized parameters was applied to conduct long-term monitoring of open-pit coal mine mining disturbances in different vegetation gradient zones, identifying the year information of open-pit coal mine mining disturbances.

2. The method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics according to claim 1, characterized in that, Specifically, the following steps are included: Step 1: Data collection and preprocessing; We collected long-term Landsat surface reflectance images of open-pit coal mines, carried out radiometric calibration and atmospheric correction preprocessing, and applied the CFMASK function to mask invalid ground features such as clouds, shadows, and snow to ensure that the mining area is covered by high-quality images every year and to establish an annual high-quality image dataset. Data on the distribution of impermeable surfaces and water bodies were extracted from the latest publicly available land use data to cover open-pit mines and eliminate interference from non-mining human activities on the monitoring of mine disturbances. Vectorize the open-pit coal mine boundary from the open-pit coal mine image dataset to serve as the research scope for open-pit coal mine mining disturbance at the current time point; Unify the coordinate system, resolution, and spatial range of all data to ensure that data from different periods and of different types are comparable; Step 2: Constructing the disturbance index; Spectral indices for three land cover types—vegetation, bare soil, and bare coal—were established through band calculations. Vegetation was characterized by NDVI, bare soil by BI, and bare coal by OCFI. By integrating the index changes of three core land features during open-pit coal mining—vegetation degradation, bare soil exposure, and bare coal exposure—a new open-pit coal mining disturbance index, MDI_NBC, is constructed. Step 3: Determine the perturbation threshold; The disturbance impact range of open-pit coal mining is defined as 2km, and the boundary of the open-pit coal mine is buffered by 4km. The 2-4km buffer ring is regarded as the undisturbed area. Randomly selected points were taken within the open-pit mine area and undisturbed areas to track the temporal change trajectory of the MDI_NBC index at each sample point. The trend, magnitude, and duration of the index changes in each region were statistically analyzed to determine the threshold for distinguishing between mining disturbances and non-mining disturbances. Step 4: Monitoring coal mining disturbance; Based on random sample points within the mining area, a time series segmentation algorithm based on MDI_NBC is constructed. The minimum root mean square error of the sample points is used as a reference standard, and the corresponding parameter combination is selected as the optimal parameter for the time series segmentation algorithm in open-pit coal mines. The optimized time series segmentation algorithm was applied to the time series remote sensing data of the mining area to reconstruct the MDI_NBC time series change trajectory of the entire mining area pixel by pixel. Disturbances in which the change trend, amplitude and time of each pixel exceeded the set threshold were regarded as open-pit coal mining disturbances, and the year information of the disturbance was identified. Based on the preliminary identification results of the perturbed years, spatial filtering is carried out to reduce noise in the year information. Based on the disturbance area in each year, a random stratified sampling method with unequal proportions is adopted to generate sample points in each mining area in each year, ensuring that the spatial interval of the sample points is not less than 100 meters, thereby reducing the spatial autocorrelation of the sample points. High-resolution imagery was acquired from online maps and combined with Landsat time-series image tiling tools to determine the true year of disturbance for each sample point. The results of the three years of disturbance years identified by each sample point are aggregated with the results of the adjacent years into one class. A confusion matrix of disturbance years for each open-pit mine is constructed, and user accuracy, producer accuracy, overall accuracy and F1 score are calculated to evaluate the accuracy of disturbance year identification.

3. The method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics according to claim 2, characterized in that, The construction of the novel open-pit coal mining disturbance index MDI_NBC in step 2 is specifically calculated as shown in Formula 1: ， In the formula, MDI_NBC y Let Max be the disturbance index for year y. [y,最新一年] OCFI is the maximum value of OCFI from year y to the latest year, Max. [1990,最新一年] MBI represents the maximum value of MBI from 1990 to the latest year. [y,最新一年] NDVI is the minimum NDVI value from year y to the latest year, OCFI y For the frequency of bare coal in year y, Max [1990,y] MBI represents the maximum MBI from 1990 to year y, NDVI y The mean NDVI over y years; The NDVI calculation process is shown in Formulas 2 and 3: ， ， The MBI calculation process is shown in Formulas 4 and 5: ， ， The OCFI calculation process is shown in formulas 6 and 7: ， ， In the formula, ρ R ρ represents the red band reflectivity. NIR ρ represents the reflectivity in the near-infrared band. SWIR1 ρ represents the reflectivity of shortwave infrared 1. SWIR2 This represents the reflectivity of shortwave infrared 2.

4. The method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics according to claim 2, characterized in that, In step 3, the threshold for the disturbance index MDI_NBC in new open-pit coal mines is determined: The study investigated the trend, magnitude, and duration of the maximum changes in the MDI_NBC index at various sampling points in open-pit coal mines and undisturbed areas during the statistical research period. Disturbance source labels were established, and the numerical frequency distributions of different disturbance sources were compared. 80% of each category was randomly selected as training samples and 20% as validation samples. The random forest method was applied to determine the threshold for distinguishing between coal mining disturbances and non-coal mining disturbances.

5. The method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics according to claim 2, characterized in that, Step 4: Parameter optimization of the time series algorithm: Based on multiple core parameters of time series algorithms, a cyclical application based on sample points is carried out, and the minimum root mean square error of sample points is used as the screening criterion to establish an optimization parameter set for time series segmentation algorithms. The formula for calculating the root mean square error is: ， In the formula, y i Let i be the true value of sample point i. Let be the predicted value of sample point i, and n be the total number of sample points.

6. The method for identifying the year of disturbance in open-pit coal mines by integrating multiple feature characteristics according to claim 2, characterized in that, The specific calculation formula for evaluating the year recognition accuracy in step 4 is as follows: User precision: ， Producer precision: ， Overall accuracy: ， F1 score: ， In the formula, P ii The number of correctly classified samples, p i· p represents the number of samples predicted to be class i. ·i This represents the number of real samples of class i, m is the number of classes, and N is the total number of samples.

Images