Mine ecological restoration monitoring system based on unmanned aerial vehicle inspection

Abstract

The invention provides a mine ecological restoration monitoring system based on unmanned aerial vehicle inspection, and the system comprises the following functional modules: a data fusion module which periodically collects visible light and multispectral images of a mining area, and generates a high-resolution data base map through fusion processing; the ground feature classification and diversity quantification module is used for completing ground feature classification by utilizing a random forest algorithm based on the data base map and calculating a Shannon diversity index through a space grid; the dynamic diagnosis and regulation and control strategy generation module is used for calculating a time gradient and a spatial heterogeneity change rate according to the multi-period Shannon diversity index, diagnosing an ecological state and generating a regulation and control strategy according to the time gradient and the spatial heterogeneity change rate; and the ecological toughness evaluation and early warning module is used for calculating an ecological toughness index by fusing the indexes and realizing graded evaluation and early warning of the ecological restoration situation. According to the invention, dynamic quantitative monitoring of the evolution process of the mine ecosystem is realized, the ecological change trend can be accurately identified, and graded early warning and regulation decision support is provided for restoration engineering.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ecological environment monitoring, in particular to a mine ecological restoration monitoring system based on unmanned aerial vehicle inspection. BACKGROUND

[0002] The exploitation of mineral resources not only supports the development of the national economy, but also causes serious ecological and environmental problems, such as vegetation destruction, soil erosion, and loss of biodiversity. Implementing scientific and effective ecological restoration and continuously monitoring the process is an important link to promote the sustainable development of mining areas. The restoration of the mine ecosystem is usually reflected in the composition, spatial structure, and dynamic process of the surface cover over time. Among them, the reconstruction of vegetation communities, the transformation of bare rock and slag soil, and the heterogeneity of the spatial distribution of various land objects are the key to evaluating the restoration effect. Therefore, comprehensive monitoring of the mine area at multiple time periods and multiple indicators is crucial to capture the subtle evolution trend of the ecological system and optimize the restoration strategy for precise control.

[0003] In the prior art, the patent number CN115752381A discloses a mine monitoring method based on unmanned aerial vehicle remote sensing technology. The technology includes an image acquisition module for collecting mine image information; a flight platform for carrying the image acquisition module for aerial photography; a control system for controlling the flight platform and the image acquisition module; a communication module for remote communication with the flight platform and the image acquisition module; a storage module for storing various information collected by the image acquisition module; and an information receiving and processing module for processing various information sent back by the image acquisition module. This method uses unmanned aerial vehicle remote sensing technology to monitor the mine, can obtain high-definition aerial images in low-altitude areas, and can obtain high-precision parameter information through integrated analysis of these parameters, generate a real scene three-dimensional model, and perform quantitative analysis such as volume calculation and change monitoring, with high precision, laying a foundation for mine ecological restoration monitoring.

[0004] However, the above-mentioned prior art relies on manual field investigation and single-time remote sensing interpretation, which has the problems of long monitoring period, high cost, and limited coverage, making it difficult to achieve continuous dynamic tracking of the restoration area. Its analysis is mostly limited to single indicators such as vegetation coverage, lacking quantitative description of the spatial heterogeneity of the ecological system and the composition of biodiversity, and unable to accurately reflect the evolution of the structure and function of the ecological system. At the same time, the existing methods are mostly post-description, lacking trend diagnosis and early warning ability based on multi-period data, leading to lagging management decisions and inability to make forward-looking and precise control interventions on the restoration process.

[0005] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The present application aims to provide a mine ecological restoration monitoring system based on unmanned aerial vehicle inspection to solve the problems raised in the background art. The present application can accurately capture the subtle evolution trend of the ecological system by fusing multi-period unmanned aerial remote sensing data and introducing multi-dimensional indexes such as Shannon diversity index, time gradient and spatial heterogeneity change rate. The system overcomes the drawbacks of traditional methods, such as relying on manual interpretation, low efficiency and difficulty in quantifying spatial heterogeneity. The core advantage is that the monitoring, diagnosis, early warning and decision support are integrated into one, and through the ecological resilience index and adaptive threshold mechanism, the scientific and forward situation awareness and hierarchical intervention of the restoration process are realized, which significantly improves the fine, intelligent and active regulation level of ecological restoration management.

[0007] To achieve the above purpose, the present application provides the following technical scheme: The mine ecological restoration monitoring system based on unmanned aerial vehicle inspection comprises the following modules: Data fusion module: unmanned aerial vehicle monitoring of the mine restoration area is carried out at a fixed time interval inspection cycle, image data of the mine restoration area is collected in each inspection cycle, the image data includes high-resolution visible light image and multispectral image, and the image data is preprocessed and fused to generate a data base map; Ground feature classification and diversity quantification module: based on the data base map, a supervised classification algorithm of random forest is used for ground feature classification interpretation to generate a land cover classification map, the mine restoration area in the land cover classification map is divided into regular spatial grids with a fixed side length based on the land cover classification map, and for each spatial grid, the area proportion of each ground feature category inside the spatial grid is quantified, and the Shannon diversity index of each spatial grid is calculated based on the quantified area proportion of each ground feature category inside each spatial grid; Dynamic diagnosis and control strategy generation module: based on the Shannon diversity index of the current inspection cycle and the historical inspection cycle, the time gradient and the spatial heterogeneity change rate are calculated respectively, and based on the numerical size relationship of the time gradient and the spatial heterogeneity change rate, the state diagnosis is performed and the control strategy is output; Ecological resilience assessment and early warning module: based on the time gradient and the spatial heterogeneity change rate, the ecological resilience index is calculated, and based on the numerical value of the ecological resilience index, the ecological restoration situation of the mine restoration area is assessed and early warned.

[0008] The image data is preprocessed and fused in the following way: The collected high-resolution visible light image and multispectral image are respectively subjected to radiation correction and geometric correction, the radiation correction includes radiation calibration and atmospheric correction, for the spectral image, the radiation calibration is used to convert the original pixel value of the spectral image into the apparent reflectance at the top of the atmosphere, for the high-resolution visible light image, the radiation calibration is used to convert the original pixel value of the high-resolution visible light image into the relative brightness value of the ground surface, and the atmospheric correction converts the apparent reflectance at the top of the atmosphere of the multispectral image into the true reflectance of the ground surface; The geometric correction includes geometric fine correction and orthographic correction, which are used to eliminate the geometric distortion of the high-resolution visible light image and the multispectral image, the high-resolution visible light image and the multispectral image subjected to the radiation correction and the geometric correction are subjected to pixel-level registration, and the high-resolution visible light image and the multispectral image subjected to the pixel-level registration are subjected to data fusion, thereby generating a data base map with high spatial resolution and multispectral information, which is used for subsequent ground feature classification and interpretation.

[0009] Based on the data base map, the execution logic of the supervised classification algorithm of random forest for ground feature classification and interpretation is as follows: According to the mine restoration target, a ground feature classification system including bare rock, slag, restored vegetation, natural vegetation, water body and artificial structure is established, and based on the data base map with high spatial resolution and multispectral information generated after preprocessing, a training sample set including pixel coordinates and category labels is created for each ground feature category through manual interpretation and drawing, feature extraction is performed for each pixel in the data base map to form a multi-dimensional feature for classification, and the multi-dimensional feature includes spectral feature, vegetation index feature and texture feature; The training sample set and the multi-dimensional feature corresponding to each sample in the training sample set are jointly input into the supervised classification algorithm of random forest, and the self-service sampling and node splitting of multiple decision trees are trained to construct a random forest ground feature classification model, and the random forest ground feature classification model is used to interpret the image data of the mine restoration area to obtain the preliminary ground feature category of each pixel, thereby forming a preliminary classification result, and the preliminary classification result is subjected to spot filtering and logic consistency checking to eliminate isolated noise pixels and merge small spots, thereby generating a land cover classification map.

[0010] The area proportion of each ground feature category in the internal space grid is quantified in the following manner: For each space grid, the total number of pixels of each ground feature category contained in the space grid is counted, and the ratio of the total number of pixels of each ground feature category to the total number of pixels in the space grid is calculated, thereby quantifying the area proportion of each ground feature category; The formula for calculating the Shannon diversity index of each space grid is as follows: ; wherein, is the Shannon diversity index of the spatial grid, used to quantify the richness of all land cover categories within the spatial grid; is the area proportion of the th land cover category within the spatial grid; is the cyclic variable of the land cover category, representing the th land cover category; is the natural logarithm of ; is the natural logarithm of is the total number of land cover categories within the spatial grid.

[0011] The calculation formula of the time gradient is: ; wherein, is the time gradient of the th inspection cycle, representing the change amount of the average value of the Shannon diversity index within a fixed time interval; is the average value of the Shannon diversity index of all spatial grids within the th inspection cycle; is the average value of the Shannon diversity index of all spatial grids within the th inspection cycle; is the average value of the Shannon diversity index of all spatial grids within the th inspection cycle;

[0012] The calculation formula of the spatial heterogeneity change rate is: ; wherein, is the spatial heterogeneity change rate of the th inspection cycle; is the standard deviation of the Shannon diversity index of all spatial grids within the th inspection cycle; is the standard deviation of the Shannon diversity index of all spatial grids within the th inspection cycle, as a fixed comparison benchmark for calculating the spatial heterogeneity change rate, used to measure the cumulative evolution degree of the spatial heterogeneity of the mine ecological system since the ecological restoration of the mine.

[0013] The logic for performing state diagnosis and outputting the control strategy is: when , it is determined that the ecological state is high-quality growth, and the control strategy is to maintain the existing management; when At that time, the ecological state was determined to be homogeneous growth, and the regulation strategy was artificial intervention to suppress dominant species and promote species diversity; when When the ecological state is determined to be in a state of systemic decline, the control strategy is to initiate engineering-level intervention and restoration. when At that time, the ecological state was determined to be structural decline, and the regulation strategy was to replant specific plant species. in, The ideal time gradient threshold is a preset threshold value. It is determined in the following ways: Get the The first inspection cycle to the [number]th The median of all time gradient values ​​calculated for the first inspection cycle is set as the median value for the first inspection cycle. The ideal time gradient threshold for each inspection cycle.

[0014] The formula used to calculate the ecological resilience index is as follows: ; in, For the first Ecological resilience index for each inspection cycle; This is the absolute value of the deviation between the time gradient and the ideal time gradient threshold.

[0015] The logic for assessing and issuing early warnings about the ecological restoration status of mine restoration areas is as follows: when At that time, the ecological restoration trend is determined to be healthy, and no early warning is triggered; when At that time, it was determined that there were potential risks to the ecological restoration situation, triggering a level-one warning; when At that time, the ecological restoration situation was deemed unsatisfactory, triggering an action-level early warning; in, As a preset risk threshold, The preset health threshold, and The value is not a fixed value. and The value is adaptively adjusted based on the restoration stage of the mine restoration area and meets the following requirements. > .

[0016] The execution logic for adaptive adjustment based on the restoration stage of the mine restoration area is as follows: The repair stage is divided according to the cumulative length of time from the beginning of the mine repair to the first inspection cycle When , it is defined as the early stage of repair, and a low threshold is used: , ; When , it is defined as the middle stage of repair, and a standard threshold is used: , ; When , it is defined as the stable stage of repair, and a high threshold is used: , ; wherein, is the reference threshold value determined by historical data and expert experience, and are preset time nodes.

[0017] Compared with the prior art, the beneficial effects of the present application are: The mine ecological restoration monitoring system based on unmanned aerial vehicle inspection of the present application realizes dynamic quantitative monitoring of the evolution process of the mine ecological system by periodically collecting visible light and multispectral images through the data fusion module and generating high-resolution data base maps; using the random forest algorithm for feature classification and Shannon diversity index calculation, the ecological change trend can be identified; through the analysis of time gradient and spatial heterogeneity change rate, the ecological state is dynamically diagnosed and targeted control strategies are generated; combined with the grading evaluation and early warning mechanism of the ecological resilience index, scientific grading early warning and decision support are provided for the repair project; the system also has the function of self-adaptive threshold adjustment, which can dynamically optimize the evaluation standard according to the repair stage, thereby significantly improving the accuracy, reliability and efficiency of the monitoring, and effectively promoting the continuous optimization and precise management of the mine ecological restoration. BRIEF DESCRIPTION OF DRAWINGS

[0018] Fig. 1 is a block diagram of the mine ecological restoration monitoring system based on unmanned aerial vehicle inspection; Fig. 2 is a schematic diagram of the operation process of the mine ecological restoration monitoring system based on unmanned aerial vehicle inspection. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail below in combination with specific embodiments.

[0020] ​It should be noted that unless otherwise defined, technical terms or scientific terms used in the present application shall have the common meaning understood by one of ordinary skill in the art to which the present application pertains. The terms "first", "second", and similar terms used in the present application do not denote any order, quantity, or importance, but are merely used to distinguish different components. The terms "include", "contain", and similar terms mean that the elements or objects before the terms encompass the elements or objects listed after the terms and their equivalents, and do not exclude other elements or objects. The terms "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "upper", "lower", "left", "right", and the like are merely used to indicate relative positional relationships, which can change accordingly when the absolute positions of the described objects change.

[0021] Embodiments Figs. 1-2 The present application provides a technical solution: The mine ecological restoration monitoring system based on unmanned aerial vehicle inspection includes the following modules: The data fusion module: the unmanned aerial vehicle remote sensing monitoring of the mine restoration area is carried out at a fixed time interval of the inspection cycle. Through periodic data collection, the dynamic change process of the restoration area is systematically captured, ensuring the continuity and timeliness of the monitoring. In each inspection cycle, the unmanned aerial vehicle platform is used to comprehensively collect remote sensing image data of the mine restoration area. The image data includes high-resolution visible light images and multispectral images. The image data is preprocessed and fused to generate data base maps with high spatial resolution and multispectral information, providing unified and reliable basic data support for subsequent ecological parameter extraction and spatial analysis. The image data is preprocessed and fused in the following way: The collected high-resolution visible light images and multispectral images are respectively radiometrically and geometrically corrected. Radiometric correction includes two key steps: radiometric calibration and atmospheric correction. Radiometric calibration is performed on multispectral images. By establishing the conversion relationship between pixel value and physical quantity, the original pixel value is converted into apparent reflectance at the top of the atmosphere, so as to eliminate the radiation deviation caused by the response characteristics of the sensor. For high-resolution visible light images, radiometric calibration converts the original pixel value into ground relative brightness value, ensuring that the image brightness information is consistent with the actual ground object performance. Atmospheric correction further processes multispectral images. By simulating the atmospheric transmission process, atmospheric scattering, absorption, and other influences are eliminated, and the apparent reflectance at the top of the atmosphere is converted into ground true reflectance closer to the actual ground situation, thereby improving the physical consistency and comparability of the image. The geometric correction includes geometric fine correction and orthorectification, aiming to eliminate the geometric distortion of the image caused by factors such as sensor posture and terrain undulation. The geometric fine correction corrects the image locally by ground control points or sensor parameters, reduces the position deviation; the orthorectification further corrects the projection error caused by the terrain in combination with the digital elevation model, and ensures that the image has accurate geographical coordinates and geometric shape. After the correction is completed, the high-resolution visible light image and the multispectral image are pixel-level registered, the two images are accurately aligned in space through feature matching, and the pixel position is consistent. The registered image is data fused, the fine spatial details of the high-resolution visible light image and the rich spectral information of the multispectral image are organically combined through the fusion algorithm, and a data base map with high spatial resolution and multispectral characteristics is generated. The data base map not only retains the shape and texture details of the ground objects, but also contains multi-band spectral data, providing multi-dimensional basic data support for subsequent ground object classification interpretation.

[0022] The ground object classification and diversity quantification module: based on the data base map, a supervised classification algorithm of random forest is used for ground object classification and interpretation to generate a land cover classification map. The land cover classification map not only provides an intuitive visualization of the spatial distribution of ground objects, but also lays a data foundation for subsequent ecological analysis. Based on the land cover classification map, the mine restoration area in the land cover classification map is divided into regular spatial grids with a fixed side length, and each grid is taken as an independent analysis unit to systematically evaluate local ecological characteristics. For each spatial grid, the area proportion of each ground object category inside the spatial grid is quantified, and the Shannon diversity index of each spatial grid is calculated based on the quantified area proportion of each ground object category inside each spatial grid. This index quantifies the ecological diversity level in the grid by comprehensively considering the richness and uniformity of the ground object categories. The execution logic of using a supervised classification algorithm of random forest to perform ground object classification based on the data base map is as follows: According to the mine restoration target, a ground object classification system including bare rock, slag, restored vegetation, natural vegetation, water body and artificial structure is established to ensure that the classification result can fully reflect the ecological composition and human intervention effect of the restoration area. Based on the data base map generated after preprocessing, which has high spatial resolution and multispectral information, a training sample set containing pixel coordinates and category labels is created for each ground object category through manual interpretation and drawing. Feature extraction is performed for each pixel from the data base map. The spectral feature utilizes the reflection characteristic difference of different ground objects in the visible light and multispectral bands to capture their basic spectral response law, thereby effectively distinguishing categories with significant texture differences such as bare rock and slag, to constitute multi-dimensional features for classification. The multi-dimensional features include spectral features, vegetation index features and texture features. After inputting the training sample set and its corresponding multi-dimensional features into the random forest supervised classification algorithm, the algorithm generates differentiated training subsets for each decision tree through bootstrap sampling, and evaluates the optimal split point from a randomly selected feature subset when splitting at each node of the tree. This mechanism effectively enhances the generalization ability of the model and suppresses overfitting. By training a large number of decision trees in parallel, a robust random forest land cover classification model is finally integrated. The model can integrate the prediction results of multiple trees and determine the optimal class of each pixel through a voting mechanism, thereby significantly improving the discrimination accuracy and stability of complex features such as bare rock, slag, and vegetation. The trained random forest land cover classification model is used to interpret the image data of the mine restoration area, assign a preliminary land cover class to each pixel, and form a preliminary classification result. The spot filtering algorithm is used to identify and eliminate isolated pixels caused by noise, improving the visual coherence of the classification map. At the same time, the logical consistency check rule is used to verify the rationality of the relationship between adjacent pixels, and the small patches caused by classification errors are merged, finally generating a land cover classification map with smooth boundaries, continuous land classes, and consistent with the actual landform distribution, laying a reliable foundation for subsequent ecological diversity quantification and dynamic monitoring.

[0023] The area proportion of each land cover class in the quantization space grid is calculated in the following manner: For each spatial grid, the total number of pixels of each land cover class contained therein is counted, and the ratio of the total number of pixels of each land cover class to the total number of pixels in the spatial grid is calculated to quantify the area proportion of each land cover class. The formula used to calculate the Shannon diversity index of each spatial grid is: ; Wherein, is the Shannon diversity index of the spatial grid, used to quantify the richness of all land cover classes in the spatial grid, The higher the value, the more diverse and evenly distributed the land cover classes in the spatial grid, and the better the ecological diversity. is the area proportion of the th land cover class in the spatial grid, reflecting the relative abundance of the land cover class, and the more balanced the proportion, the greater the contribution to diversity. is the loop variable of the land cover class, representing the th land cover class. is the natural logarithm of . is the total number of land cover classes in the spatial grid.

[0024] Dynamic diagnosis and regulation strategy generation module: Based on the Shannon diversity index of the current inspection cycle and the historical inspection cycle, calculate its temporal gradient and spatial heterogeneity change rate to assess the stability and evolution of the ecosystem structure in the spatial dimension, and perform state diagnosis and output regulation strategy based on the numerical relationship between the temporal gradient and the spatial heterogeneity change rate. The formula for calculating the time gradient is: ; in, For the first The time gradient of each inspection cycle represents the change in the average value of the Shannon diversity index over a fixed time interval. For the first The average Shannon diversity index of all spatial grids within a single inspection cycle represents the current overall diversity level. For the first The average Shannon diversity index of all spatial grids within a single inspection cycle is used as the benchmark. A fixed time interval is used to standardize the amount of change, making the gradients comparable in the time dimension; and It is an inverse proportional relationship, when the time interval is fixed. The larger, the more Time gradient of each inspection cycle The smaller.

[0025] The formula for calculating the rate of change of spatial heterogeneity is as follows: ; This formula assesses the degree of evolution of ecosystem spatial patterns by measuring the cumulative change in spatial heterogeneity by comparing the spatial variation of diversity in the current cycle with that in the initial cycle. in, For the first The rate of change of spatial heterogeneity over each inspection cycle represents the proportion of current spatial heterogeneity relative to the initial state. For the first The standard deviation of the Shannon diversity index of all spatial grids within a single inspection cycle measures the degree of spatial variability; the larger the standard deviation, the higher the spatial heterogeneity. For the first The standard deviation of the Shannon diversity index of all spatial grids within each inspection cycle is used as a fixed benchmark for calculating the rate of change of spatial heterogeneity, in order to measure the cumulative degree of evolution of spatial heterogeneity of the mine ecosystem since the mine ecological restoration. and It is positively correlated. and It is an inverse correlation.

[0026] The logic for diagnosing the execution status and outputting the control strategy is as follows: when This indicates that the ecosystem is not only showing an overall increasing diversity trend, but also that the spatial pattern is becoming richer and more heterogeneous, reflecting that the ecological restoration is on a positive development track. The ecological state is judged to be of high quality growth, and the regulation strategy should focus on maintaining the existing management and avoiding unnecessary intervention that could disrupt the natural restoration process. when This indicates that while the overall diversity of the system is increasing, the spatial distribution is becoming more homogeneous, which may pose a risk of overexpansion of a single species or land type, leading to a simplification of the ecological structure. The ecological state is determined to be homogeneous growth, and the regulation strategy emphasizes suppressing dominant species and introducing or promoting other species through artificial intervention to enhance species diversity and ecological complexity. when This indicates that not only has the spatial heterogeneity of the system declined, but the overall diversity has also deteriorated. The ecological state is judged to be in a state of systemic decline, and the regulation strategy needs to initiate comprehensive intervention and restoration measures at the engineering level to reverse the deteriorating trend. when This indicates that the spatial structure of the system is still heterogeneous, but the overall diversity is declining, manifested as local degradation or the absence of key species. The ecological state is determined to be structural decline, and the regulation strategy focuses on targeted replanting of plant species to restore key ecological functions and maintain the stability of the spatial pattern. in, The ideal time gradient threshold is a preset threshold value. It is determined in the following ways: Get the The first inspection cycle to the [number]th The median of all time gradient values ​​calculated for the first inspection cycle is set as the median value for the first inspection cycle. The ideal time gradient threshold for each inspection cycle.

[0027] Ecological resilience assessment and early warning module: Based on the time gradient and spatial heterogeneity change rate, calculate the ecological resilience index, and based on the value of the ecological resilience index, assess and provide early warning of the ecological restoration status of the mine restoration area; The formula used to calculate the ecological resilience index is as follows: ; The formula calculates the ecological resilience index of the first inspection cycle, which is used to evaluate the ability of the ecosystem to resist disturbance and maintain stability. The higher the value, the stronger the ecological resilience, which combines the spatial heterogeneity change rate and the deviation of the time gradient from the ideal value; is the ecological resilience index of the first inspection cycle, which represents the recovery ability and stability of the ecosystem, and the larger the value, the stronger the resilience; is the absolute value of the deviation of the time gradient from the ideal time gradient threshold, which is used to quantify the degree of deviation of the actual change trend from the expected value, and the larger the absolute value, the lower the stability of the system.

[0028] The logic for evaluating and warning the ecological restoration situation of the mine restoration area is as follows: When , it indicates that the recovery situation of the ecosystem is in an ideal state, the biodiversity and spatial heterogeneity are well maintained, and the system has strong anti-disturbance ability. It is determined that the ecological restoration situation is healthy, and no warning is triggered; When , it indicates that the ecological restoration has not deteriorated significantly, but there are potential instability factors, which may be manifested as local biodiversity decline or insufficient recovery power. It is determined that the ecological restoration situation has potential risks, and a warning of attention level is triggered, prompting the management personnel to increase the monitoring frequency, in-depth analysis of the influence of environmental changes or human intervention, and timely identification of potential problems and the implementation of preventive measures; When , it indicates that the resilience of the ecosystem is severely insufficient, and the recovery situation is not good, which may cause problems such as species degradation, spatial structure disorder or functional disorder. It is determined that the ecological restoration situation is not good, and the system resilience is insufficient, triggering an action-level warning, and immediately starting targeted intervention, such as adjusting the restoration strategy, strengthening ecological management or implementing emergency engineering, to prevent further degradation of the system; is a preset risk threshold, is a preset health threshold, and are not fixed values, and are self-adaptively adjusted according to the restoration stage of the mine restoration area, and satisfy > .

[0029] The execution logic of self-adaptive adjustment according to the restoration stage of the mine restoration area is as follows: ​​The aforementioned restoration phase is based on the period from the start of mine restoration to the [number missing]. The cumulative time of each inspection cycle This division, along with the time dimension, reflects the continuity and dynamic evolution of the repair process: when This period is defined as the initial stage of repair, using a low threshold: , To adapt to the relative fragility of the ecosystem structure and its high sensitivity to disturbances during the initial stage of restoration; when The time frame is defined as the mid-repair period, using a standard threshold: , To adapt to the characteristics and requirements of different restoration stages and to match the gradual establishment and restoration of ecological functions; when This period is defined as the stabilization phase of repair, using a high threshold: , To meet the requirements of higher stability and resistance to disturbance that the restored ecosystem should possess during the stable period; in, The benchmark threshold is determined through historical data and expert experience. and The preset time points serve to mark key turning points in the repair process, thereby supporting the phased evaluation and adjustment of the repair effect.

[0030] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0031] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0032] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0033] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

1. A mine ecological restoration monitoring system based on unmanned aerial vehicle (UAV) inspection, characterized in that, Includes the following modules: Data fusion module: The module monitors the mine restoration area by drone at fixed time intervals. During each inspection cycle, it collects image data of the mine restoration area, including high-resolution visible light images and multispectral images. The module preprocesses and fuses the image data to generate a data base map. Land cover classification and diversity quantification module: Based on the base map, a supervised classification algorithm of random forest is used to classify and interpret land cover to generate a land cover classification map. Based on the land cover classification map, the mine restoration area in the land cover classification map is divided into regular spatial grids with fixed side lengths. For each spatial grid, the area ratio of each land cover category within the spatial grid is quantified. Based on the area ratio of the land cover categories quantified within each spatial grid, the Shannon diversity index of each spatial grid is calculated. Dynamic diagnosis and control strategy generation module: Based on the Shannon diversity index of the current inspection cycle and the historical inspection cycle, calculate the temporal gradient and spatial heterogeneity change rate respectively, and based on the numerical relationship between the temporal gradient and the spatial heterogeneity change rate, perform state diagnosis and output control strategy. Ecological resilience assessment and early warning module: Based on the time gradient and spatial heterogeneity change rate, calculate the ecological resilience index, and based on the value of the ecological resilience index, assess and provide early warning of the ecological restoration status of the mine restoration area.

2. The mine ecological restoration monitoring system based on UAV inspection according to claim 1, characterized in that: The image data is preprocessed and fused in the following manner: Radiometric correction and geometric correction are performed on the acquired high-resolution visible light images and multispectral images, respectively. The radiometric correction includes radiometric calibration and atmospheric correction. For the spectral images, the radiometric calibration is used to convert the original pixel values ​​of the spectral images into the apparent reflectance of the top atmosphere. For the high-resolution visible light images, the radiometric calibration is used to convert the original pixel values ​​of the high-resolution visible light images into the relative brightness values ​​of the ground surface. The atmospheric correction converts the apparent reflectance of the top atmosphere of the multispectral images into the true reflectance of the ground surface. The geometric correction includes geometric fine correction and orthorectification, which are used to eliminate geometric distortions in high-resolution visible light images and multispectral images. The high-resolution visible light images and multispectral images that have undergone radiometric and geometric correction are then registered at the pixel level. The high-resolution visible light images and multispectral images that have completed pixel-level registration are then fused to generate a data base map with both high spatial resolution and multispectral information for subsequent land cover classification and interpretation.

3. The mine ecological restoration monitoring system based on UAV inspection according to claim 1, characterized in that: Based on the aforementioned base map, the execution logic for land cover classification and interpretation using the supervised classification algorithm of random forest is as follows: Based on the mine restoration goals, a land cover classification system was established, including bare rock, slag, restored vegetation, natural vegetation, water bodies, and artificial structures. Based on the pre-processed base map with high spatial resolution and multispectral information, a training sample set containing pixel coordinates and category labels was created for each land cover category through manual interpretation and delineation. Features were extracted from each pixel in the base map to form multidimensional features for classification. The multidimensional features include spectral features, vegetation index features, and texture features. The training sample set and the multi-dimensional features corresponding to each sample in the training sample set are input into the supervised classification algorithm of random forest. The algorithm is trained by bootstrap sampling and node splitting of multiple decision trees to construct a random forest land cover classification model. The random forest land cover classification model is then used to interpret the image data of the mine restoration area to obtain the preliminary land cover category of each pixel to form a preliminary classification result. The preliminary classification result is then subjected to spot filtering and logical consistency checks to eliminate isolated noise pixels and merge small patches to generate a land cover classification map.

4. The mine ecological restoration monitoring system based on UAV inspection according to claim 3, characterized in that: The method for implementing the quantification of the area proportion of each land cover category within the spatial grid is as follows: For each spatial grid, the total number of pixels of each land cover category contained within the spatial grid is counted, and the ratio of the total number of pixels of each land cover category to the total number of pixels within the spatial grid is calculated to quantify the area proportion of each land cover category. The formula used to calculate the Shannon diversity index for each spatial grid is as follows: ； in, The Shannon diversity index for spatial grids is used to quantify the richness of all land cover categories within a spatial grid. For the first The proportion of area occupied by each type of crop within the spatial grid; Let be the loop variable for land cover categories, representing the th Categories of crops; for The natural logarithm; This represents the total number of land cover categories within the spatial grid.

5. The mine ecological restoration monitoring system based on UAV inspection according to claim 1, characterized in that: The formula for calculating the time gradient is: ； in, For the first The time gradient of each inspection cycle represents the change in the average value of the Shannon diversity index over a fixed time interval. For the first The average Shannon diversity index of all spatial grids within one inspection cycle; For the first The average Shannon diversity index of all spatial grids within one inspection cycle; It is a fixed time interval.

6. The mine ecological restoration monitoring system based on UAV inspection according to claim 5, characterized in that: The formula for calculating the rate of change of spatial heterogeneity is as follows: ； in, For the first The rate of spatial heterogeneity change in each inspection cycle; For the first Standard deviation of Shannon diversity index for all spatial grids within one inspection cycle; For the first The standard deviation of the Shannon diversity index of all spatial grids within a single inspection cycle is used as a fixed benchmark for calculating the rate of change of spatial heterogeneity, and is used to measure the cumulative degree of evolution of spatial heterogeneity of the mine ecosystem since mine ecological restoration.

7. The mine ecological restoration monitoring system based on UAV inspection according to claim 6, characterized in that: The logic for diagnosing the execution status and outputting the control strategy is as follows: when At that time, the ecological state was determined to be of high quality and growth, and the regulation strategy was to maintain the existing management. when At that time, the ecological state was determined to be homogeneous growth, and the regulation strategy was artificial intervention to suppress dominant species and promote species diversity; when When the ecological state is determined to be in a state of systemic decline, the control strategy is to initiate engineering-level intervention and restoration. when At that time, the ecological state was determined to be structural decline, and the regulation strategy was to replant specific plant species. in, The ideal time gradient threshold is a preset threshold value. It is determined in the following ways: Get the The first inspection cycle to the [number]th The median of all time gradient values ​​calculated for the first inspection cycle is set as the median value for the first inspection cycle. The ideal time gradient threshold for each inspection cycle.

8. The mine ecological restoration monitoring system based on UAV inspection according to claim 1, characterized in that: The formula used to calculate the ecological resilience index is as follows: ； in, For the first Ecological resilience index for each inspection cycle; This is the absolute value of the deviation between the time gradient and the ideal time gradient threshold.

9. The mine ecological restoration monitoring system based on UAV inspection according to claim 8, characterized in that: The logic for assessing and issuing early warnings about the ecological restoration status of mine restoration areas is as follows: when At that time, the ecological restoration trend is determined to be healthy, and no early warning is triggered; when At that time, it was determined that there were potential risks to the ecological restoration situation, triggering a level-one warning; when At that time, the ecological restoration situation was deemed unsatisfactory, triggering an action-level early warning; in, As a preset risk threshold, The preset health threshold, and The value is not a fixed value. and The value is adaptively adjusted based on the restoration stage of the mine restoration area and meets the following requirements. > .

10. The mine ecological restoration monitoring system based on UAV inspection according to claim 9, characterized in that: The execution logic for adaptive adjustment based on the restoration stage of the mine restoration area is as follows: The aforementioned restoration phase is based on the period from the start of mine restoration to the [number missing]. The cumulative time length of each inspection cycle Divide into: when This period is defined as the initial stage of repair, using a low threshold: , ; when The time frame is defined as the mid-repair period, using a standard threshold: , ; when This period is defined as the stabilization phase of repair, using a high threshold: , ; in, The benchmark threshold is determined through historical data and expert experience. and This is a preset time point.

Images