A rock high slope collapse target discrimination method and system

Abstract

The present application relates to the field of rock high slope safety monitoring, in particular to a rock high slope collapse target discrimination method and system. The center end constructs the reference environment data according to the spatial monitoring unit and issues based on high-resolution images and three-dimensional point clouds. The slope monitoring terminal periodically collects the current image and point cloud and registers, calculates the image change index and the terrain change index to obtain the comprehensive change index, and selects the candidate collapse monitoring unit. The multi-temporal data of the candidate unit is stored in the preset time window, the multi-temporal feature sequence is constructed, and the collapse risk level and the collapse target discrimination result are output according to the collapse discrimination parameter. The terminal adaptively adjusts the sampling period, the change threshold and the time window according to the resource state to control the calculation amount, uploads the multi-temporal features and monitoring data of the high-risk candidate unit to the center end for review, and updates the related parameters by counting the false alarm rate and the missed alarm rate, solving the real-time collapse identification problem of high-resolution image point cloud under low-power consumption conditions.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring of high rock slopes, specifically a method and system for identifying collapse targets on high rock slopes. Background Technology

[0002] With the large-scale construction of highways, railways, and water conservancy projects in mountainous areas, the number of high rock slopes has increased rapidly, and incidents of traffic disruption, facility damage, and casualties caused by slope collapses occur frequently. In order to reduce disaster risks, long-term monitoring and collapse risk assessment of high rock slopes are usually required.

[0003] In the existing technology, common slope monitoring methods include: using point monitoring instruments such as total station, GNSS, and crack gauge. These methods can only obtain displacement information of a small number of discrete monitoring points, and the spatial coverage is limited, making it difficult to fully reflect the overall stability of high rock slopes.

[0004] High-resolution images and 3D point clouds are acquired by using UAV imagery, aerial photography and LiDAR scanning. Modeling and multi-temporal deformation analysis are performed on central equipment or back-end servers. Although this can obtain relatively fine geometric change information, the computational load is large and it is difficult to run in real time on low-power devices in the field.

[0005] Some video intelligent analysis or image recognition methods are deployed on field terminals. They are generally based on two-dimensional images for single-frame or short-sequence target detection, lacking fusion with three-dimensional point clouds and analysis of deformation evolution over long periods. Their ability to identify slow-change + abrupt-change processes such as rock slope collapse is limited.

[0006] Furthermore, high rock slopes are often located in environments with limited power supply and unstable communication conditions. The computing resources of on-site monitoring terminals are limited, making it difficult to run complex algorithms for high-resolution images and 3D point clouds directly on the terminals. If all data is uploaded to the central terminal for processing, it will lead to excessive communication burden, insufficient timeliness, and difficulty in timely adjusting the algorithm parameters of the on-site terminals. In the long run, the false alarm rate and false negative rate will be difficult to control.

[0007] Therefore, the existing technology still has the following problems:

[0008] High-resolution images and 3D point cloud data are large in volume. Traditional processing mainly relies on central terminal equipment, which makes it difficult to achieve real-time collapse identification on low-power terminals at the slope site.

[0009] On-site terminals are mostly based on a single data source or simple thresholds, failing to fully utilize the fusion of image change information and 3D point cloud geometric change information. The selection of candidate collapse areas is not accurate enough, and the computational burden is heavy.

[0010] The lack of a multi-temporal feature sequence construction and comprehensive discrimination mechanism for candidate collapse monitoring units makes it impossible to fully utilize the temporal evolution patterns of morphological and displacement changes;

[0011] The lack of an adaptive adjustment mechanism based on terminal resource status, as well as a cloud-based closed-loop parameter update mechanism based on false alarm rate and false negative rate, makes it difficult for the monitoring system to balance real-time performance and discrimination accuracy in long-term operation. Summary of the Invention

[0012] The purpose of this invention is to provide a solution to the aforementioned problems existing in the prior art. Specifically, this invention is achieved through the following technical solution:

[0013] A method for identifying collapse targets on high rock slopes includes the following steps:

[0014] Step 1: The central terminal device acquires initial high-resolution image data and initial 3D point cloud data of the rocky high slope, generates benchmark environmental data containing static morphological features of multiple spatial monitoring units, and sends it to the slope monitoring terminal.

[0015] Step 2: The slope monitoring terminal collects the current monitoring image data and the current monitoring 3D point cloud data according to the preset sampling cycle. The current monitoring image data and the current monitoring 3D point cloud data are then spatially registered with the reference environmental data to obtain aligned data.

[0016] Step 3: The slope monitoring terminal calculates the image change index and topographic change index of each spatial monitoring unit based on the aligned data and the baseline environmental data to obtain the comprehensive change index. Spatial monitoring units with comprehensive change index greater than the preset change threshold are identified as candidate collapse monitoring units.

[0017] Step 4: The slope monitoring terminal stores multi-temporal relative alignment data of candidate collapse monitoring units within a preset time window, and obtains multi-temporal feature sequences based on the stored multi-temporal relative alignment data of candidate collapse monitoring units;

[0018] Step 5: The slope monitoring terminal outputs the collapse risk level and collapse target identification result of the candidate collapse monitoring unit based on the collapse discrimination parameters and multi-temporal characteristic sequence. When the collapse risk level reaches the preset high risk level, proceed to step 7; otherwise, proceed to step 6.

[0019] Step 6: Obtain the resource status parameters of the slope monitoring terminal, and control the number of candidate collapse monitoring units and the length of multi-temporal feature sequences according to the resource status parameters, so that the calculation amount of a single collapse target identification does not exceed the preset calculation limit, and return to Step 2 to perform the next cycle of identification;

[0020] Step 7: The slope monitoring terminal uploads the multi-temporal characteristic sequence, alignment data, and collapse target discrimination results of the candidate collapse monitoring units to the central device. The central device performs a review and discrimination based on the baseline environmental data and the uploaded data within a preset statistical time period, outputs the review results, calculates the false alarm rate and the missed alarm rate, and updates one or more of the collapse discrimination parameters, preset sampling period, preset change threshold, and preset time window according to the false alarm rate and the missed alarm rate, and sends them to the slope monitoring terminal.

[0021] Furthermore, the central terminal device acquires initial high-resolution image data and initial three-dimensional point cloud data of the rocky high slope, generates benchmark environmental data containing static morphological features of multiple spatial monitoring units, and distributes it to the slope monitoring terminal, including:

[0022] The central equipment acquires initial high-resolution image data through remote sensing photography equipment and initial 3D point cloud data through lidar equipment; it constructs a multi-level resolution image dataset from the initial high-resolution image data, dividing the rocky high slope area into multiple spatial monitoring units; it performs rasterization processing on the initial 3D point cloud data to obtain topographic raster data corresponding to the spatial monitoring units; based on the multi-level resolution image dataset and topographic raster data, it calculates static morphological features according to preset static morphological parameters, which include at least two of the following: average elevation, slope, aspect, and surface roughness, to form baseline environmental data.

[0023] Furthermore, the slope monitoring terminal collects current monitoring image data and current monitoring 3D point cloud data according to a preset sampling cycle, and spatially registers the current monitoring image data and current monitoring 3D point cloud data with the reference environmental data in sequence to obtain aligned data, including:

[0024] The slope monitoring terminal collects current monitoring image data and current monitoring 3D point cloud data according to a preset sampling cycle. It maps the current monitoring image data to the target resolution image level in the reference environmental data. Based on feature point matching, it performs translational and rotational alignment on the current monitoring image data and the target resolution image data to obtain registered image data. Based on the spatial reference of the registered image data, it projects the current monitoring 3D point cloud data to a coordinate system consistent with the terrain raster data in the reference environmental data and performs rasterization processing to obtain aligned data.

[0025] Furthermore, the slope monitoring terminal calculates image change indicators and topographic change indicators for each spatial monitoring unit based on aligned data and baseline environmental data to obtain a comprehensive change index. Spatial monitoring units with a comprehensive change index greater than a preset change threshold are identified as candidate collapse monitoring units, including:

[0026] For each spatial monitoring unit, based on the aligned current monitoring image data and the corresponding image data in the benchmark environmental data, an image change index is obtained according to the image change parameters. The image change parameters include at least one of grayscale difference, texture change amount, and edge direction change amount.

[0027] Based on the aligned current terrain raster data and the corresponding terrain raster data in the baseline environmental data, terrain change indicators are obtained according to terrain change parameters. The terrain change parameters include at least one of elevation difference, slope difference, and surface roughness difference.

[0028] The image change index and the terrain change index are normalized and weighted and summed according to preset weights to obtain the comprehensive change index. Based on the comprehensive change index sorted from high to low, spatial monitoring units with comprehensive change index greater than the preset change threshold and number not greater than the preset upper limit are selected as candidate collapse monitoring units.

[0029] Furthermore, the slope monitoring terminal stores multi-temporal relative alignment data of candidate landslide monitoring units within a preset time window, and obtains a multi-temporal feature sequence based on the stored multi-temporal relative alignment data of candidate landslide monitoring units, including:

[0030] The preset time window covers the most recent M preset sampling periods, where M is a preset positive integer. Within the preset time window, the slope monitoring terminal stores the aligned current monitoring image data and current monitoring 3D point cloud data of the candidate collapse monitoring unit corresponding to each preset sampling period.

[0031] Image change features are extracted from the aligned current monitoring image data to form an image change feature sequence; terrain change features are extracted from the aligned current monitoring 3D point cloud data to form a terrain change feature sequence; and a multi-temporal feature sequence is obtained based on the image change feature sequence and the terrain change feature sequence.

[0032] Furthermore, the slope monitoring terminal performs comprehensive analysis of multi-temporal characteristic sequences based on collapse discrimination parameters to obtain the collapse risk level and collapse target discrimination results of candidate collapse monitoring units, including:

[0033] The collapse discrimination parameters include weighting coefficients for each feature dimension of the multi-temporal feature sequence and thresholds for classifying multi-level collapse risk levels. The slope monitoring terminal weights each feature dimension of the multi-temporal feature sequence according to the weighting coefficients to obtain a comprehensive discrimination value. The comprehensive discrimination value is compared with the thresholds for classifying multi-level collapse risk levels. When the comprehensive discrimination value is less than the lowest threshold for classifying collapse risk level, the candidate collapse monitoring unit is judged as being in a normal state. When the comprehensive discrimination value is between the adjacent thresholds for classifying collapse risk level, the candidate collapse monitoring unit is judged as being at the corresponding collapse risk level. When the comprehensive discrimination value is greater than the highest threshold for classifying collapse risk level, the candidate collapse monitoring unit is judged as having a collapse target and being at a high risk level.

[0034] Furthermore, the acquisition of resource status parameters from the slope monitoring terminal, and the control of the number of candidate landslide monitoring units and the length of multi-temporal feature sequences based on these parameters, ensuring that the computational load for a single landslide target identification does not exceed a preset computational limit, includes:

[0035] The slope monitoring terminal periodically reads processor occupancy, power consumption, and equipment temperature to form resource status parameters. These parameters are compared with preset resource thresholds. When processor occupancy exceeds the processor resource threshold and / or power consumption is less than the power resource threshold and / or equipment temperature exceeds the temperature resource threshold, the slope monitoring terminal is determined to be in a high-load state. In this high-load state, the preset sampling period is increased and / or the preset change threshold is raised and / or the preset time window is shortened to ensure that the computational load for a single collapse target identification does not exceed the preset computational limit.

[0036] Furthermore, when the landslide risk level reaches a preset high-risk level, the slope monitoring terminal uploads the multi-temporal characteristic sequences, alignment data, and landslide target discrimination results of the candidate landslide monitoring units to the central terminal device, including:

[0037] When the collapse risk level of a candidate collapse monitoring unit reaches the preset high risk level, the slope monitoring terminal combines the multi-temporal feature sequence corresponding to the candidate collapse monitoring unit, the aligned current monitoring image data, the aligned current monitoring 3D point cloud data, and the collapse target discrimination result to form the uploaded data and sends it to the central device. The central device performs a review and discrimination based on the baseline environmental data and the uploaded data, generates the central review and discrimination result, and outputs the review result.

[0038] Furthermore, the statistical calculation of false alarm rate and false negative rate, and the updating of one or more of the collapse discrimination parameters, preset sampling period, preset change threshold, and preset time window based on the false alarm rate and false negative rate, and the data are then sent to the slope monitoring terminal, including:

[0039] Within a preset statistical time period, the central terminal equipment compares the collapse target identification results of the slope monitoring terminal with the central terminal verification identification results, calculates the false alarm rate and the missed alarm rate, and compares the false alarm rate and the missed alarm rate with preset thresholds. When the false alarm rate or the missed alarm rate is greater than the preset threshold, it updates one or more of the collapse identification parameters, preset sampling period, preset change threshold and preset time window according to the preset adjustment strategy, forms the updated parameters and sends them to the slope monitoring terminal.

[0040] A rock slope collapse target identification system, applying the aforementioned rock slope collapse target identification method, includes a central terminal device and a slope monitoring terminal; the slope monitoring terminal is communicatively connected to the central terminal device;

[0041] The central terminal device includes a benchmark environment construction unit, a verification and parameter update unit, and a communication device, wherein the benchmark environment construction unit and the verification and parameter update unit are respectively connected to the communication device;

[0042] The slope monitoring terminal includes a data acquisition and registration unit, a multi-source change calculation and candidate screening unit, a multi-temporal feature construction unit, a collapse discrimination unit, a resource status perception and adjustment unit, and a result output and uploading unit.

[0043] The data acquisition and registration unit, the multi-source change calculation and candidate screening unit, the multi-temporal feature construction unit, the collapse discrimination unit, and the resource status perception and adjustment unit are respectively connected to the result output and upload unit; the result output and upload unit is communicatively connected to the communication device.

[0044] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0045] By constructing benchmark environmental data containing the static morphological characteristics of spatial monitoring units at the central terminal, the complex high-resolution data modeling process is completed at the central terminal with sufficient computing power, reducing the computational burden on the slope monitoring terminal.

[0046] The slope monitoring terminal uses both image change indicators and topographic change indicators to calculate comprehensive change indicators and screen candidate collapse monitoring units. While ensuring sensitivity, it significantly reduces the number of spatial units that need in-depth analysis and improves computational efficiency.

[0047] By constructing a multi-temporal feature sequence within a preset time window and comprehensively considering the temporal evolution of image change features and terrain change features, it is possible to more accurately identify the collapse targets and risk levels of high rock slopes and improve the reliability of the judgment.

[0048] The slope monitoring terminal dynamically adjusts the preset sampling period, preset change threshold, and preset time window based on processor occupancy, power consumption, and equipment temperature, thereby achieving adaptive control over the number of candidate collapse monitoring units and the length of multi-temporal characteristic sequences, enabling the algorithm to operate stably under low power consumption conditions.

[0049] The central terminal equipment performs verification and judgment based on the baseline environmental data and uploaded data, calculates the false alarm rate and the missed alarm rate, and updates the collapse judgment parameters and preset parameters according to the statistical results. This achieves closed-loop optimization between the cloud and the slope monitoring terminal, improving the overall judgment accuracy and stability of the system in long-term operation. Attached Figure Description

[0050] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0051] Figure 1 This is a flowchart illustrating a method for identifying collapse targets on a high rock slope. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0053] Example 1

[0054] A method for identifying collapse targets on high rock slopes includes the following steps:

[0055] Step 1: The central terminal device acquires initial high-resolution image data and initial 3D point cloud data of the rocky high slope, generates benchmark environmental data containing static morphological features of multiple spatial monitoring units, and sends it to the slope monitoring terminal.

[0056] Step 2: The slope monitoring terminal collects the current monitoring image data and the current monitoring 3D point cloud data according to the preset sampling cycle. The current monitoring image data and the current monitoring 3D point cloud data are then spatially registered with the reference environmental data to obtain aligned data.

[0057] Step 3: The slope monitoring terminal calculates the image change index and topographic change index of each spatial monitoring unit based on the aligned data and the baseline environmental data to obtain the comprehensive change index. Spatial monitoring units with comprehensive change index greater than the preset change threshold are identified as candidate collapse monitoring units.

[0058] Step 4: The slope monitoring terminal stores multi-temporal relative alignment data of candidate collapse monitoring units within a preset time window, and obtains multi-temporal feature sequences based on the stored multi-temporal relative alignment data of candidate collapse monitoring units;

[0059] Step 5: The slope monitoring terminal outputs the collapse risk level and collapse target identification result of the candidate collapse monitoring unit based on the collapse discrimination parameters and multi-temporal characteristic sequence. When the collapse risk level reaches the preset high risk level, proceed to step 7; otherwise, proceed to step 6.

[0060] Step 6: Obtain the resource status parameters of the slope monitoring terminal, and control the number of candidate collapse monitoring units and the length of multi-temporal feature sequences according to the resource status parameters, so that the calculation amount of a single collapse target identification does not exceed the preset calculation limit, and return to Step 2 to perform the next cycle of identification;

[0061] Step 7: The slope monitoring terminal uploads the multi-temporal characteristic sequence, alignment data, and collapse target discrimination results of the candidate collapse monitoring units to the central device. The central device performs a review and discrimination based on the baseline environmental data and the uploaded data within a preset statistical time period, outputs the review results, calculates the false alarm rate and the missed alarm rate, and updates one or more of the collapse discrimination parameters, preset sampling period, preset change threshold, and preset time window according to the false alarm rate and the missed alarm rate, and sends them to the slope monitoring terminal.

[0062] Specifically, the central equipment acquires initial high-resolution image data of the target rocky high slope using remote sensing photography equipment, and acquires corresponding initial three-dimensional point cloud data using lidar equipment. Preferably, the spatial resolution of the initial high-resolution image data is between centimeters and decimeters, and the point cloud density of the initial three-dimensional point cloud data is preset according to monitoring requirements.

[0063] The central equipment preprocesses the initial high-resolution image data, including radiometric correction and geometric correction, and constructs multi-level resolution image datasets, such as original resolution, 2x downsampling resolution, 4x downsampling resolution, and other levels.

[0064] The central terminal equipment divides the rocky high slope area into multiple spatial monitoring units based on the monitoring area. The spatial monitoring units can be regular grid units, such as rectangular or square grids of fixed size, or they can be irregularly partitioned according to the slope geometry.

[0065] For the initial 3D point cloud data, the central device projects the point cloud onto a uniform coordinate system of elevation grids to obtain terrain grid data corresponding to the spatial monitoring unit. For each spatial monitoring unit, the central device calculates static morphological features based on multi-level resolution image datasets and terrain grid data. Static morphological features include at least two or more of the following: average elevation, slope, aspect, and surface roughness, used to characterize the geometric shape and surface features of the spatial monitoring unit in its initial state.

[0066] The central terminal device uses the spatial location index of the spatial monitoring unit as the key to organize the corresponding static morphological features, multi-level resolution image data, and terrain raster data into baseline environmental data, which is then stored in the central terminal device. Subsequently, the central terminal device transmits the portion of the baseline environmental data corresponding to the monitoring range of the slope monitoring terminal to the slope monitoring terminal via a communication device.

[0067] During the monitoring operation period, the slope monitoring terminal collects current monitoring image data and current monitoring 3D point cloud data according to the preset sampling cycle. The preset sampling cycle is set according to the on-site risk level and terminal resources, and can range from several minutes to tens of minutes.

[0068] The slope monitoring terminal maps the current monitoring image data to the target resolution image level in the baseline environmental data. The target resolution can preferably be an intermediate level that balances detail and computational burden. The slope monitoring terminal extracts feature points between the current monitoring image data and the target resolution image data, uses a feature matching algorithm to estimate translation and rotation parameters, and performs translational and rotational alignment on the current monitoring image data to obtain registered image data.

[0069] Based on the spatial reference of the registered image data, the slope monitoring terminal projects the current monitored 3D point cloud data onto a coordinate system consistent with the terrain raster data in the baseline environmental data, and performs rasterization processing to obtain the current terrain raster data. The registered image data and the current terrain raster data together constitute the aligned data.

[0070] For each spatial monitoring unit, the slope monitoring terminal calculates image change parameters, such as grayscale difference, texture change, and edge direction change, based on the aligned current monitoring image data and the corresponding image data in the baseline environmental data, thereby obtaining image change indicators.

[0071] Meanwhile, the slope monitoring terminal calculates terrain change parameters, such as elevation difference, slope difference, and surface roughness difference, based on the current terrain raster data and the corresponding terrain raster data in the baseline environmental data, thereby obtaining terrain change indicators.

[0072] To integrate image change indicators and topographic change indicators at a unified scale, the slope monitoring terminal normalizes the two types of indicators and then sums them according to preset weights to obtain a comprehensive change index. The larger the comprehensive change index, the higher the degree of comprehensive change in the spatial monitoring unit.

[0073] The slope monitoring terminal compares the comprehensive change index with the preset change threshold, and identifies the spatial monitoring units with the comprehensive change index greater than the preset change threshold as candidate collapse monitoring units. In order to control the amount of calculation, the comprehensive change index can be further sorted from high to low, and the number of candidate collapse monitoring units can be limited to no more than the preset upper limit.

[0074] The slope monitoring terminal is set with a preset time window, which covers the most recent M preset sampling periods, where M is a preset positive integer. For each candidate collapse monitoring unit, the slope monitoring terminal stores the aligned current monitoring image data and the current monitoring 3D point cloud data corresponding to each sampling period within the preset time window.

[0075] In each sampling period, the slope monitoring terminal extracts image change features from the aligned current monitoring image data, such as brightness change features, texture change features, edge morphology change features, etc., to form an image change feature sequence; and extracts terrain change features from the aligned current monitoring 3D point cloud data, such as elevation change features, slope change features, volume change features, surface roughness change features, etc., to form a terrain change feature sequence.

[0076] The slope monitoring terminal constructs a multi-temporal feature sequence by combining the image change feature sequence and the terrain change feature sequence in chronological order. That is, for each sampling period within a preset time window, the image change feature and terrain change feature corresponding to that sampling period are spliced ​​together in a preset order to obtain the temporal feature vector of that sampling period. The slope monitoring terminal arranges the temporal feature vectors of each sampling period in the order of sampling time to construct a multi-temporal feature sequence, which is used to reflect the morphological and displacement changes of the candidate collapse monitoring unit within the preset time window.

[0077] The slope monitoring terminal pre-stores collapse discrimination parameters, which include weight coefficients for each feature dimension of the multi-temporal feature sequence and thresholds for classifying multi-level collapse risk levels.

[0078] For each candidate landslide monitoring unit, the slope monitoring terminal weights each feature dimension of the multi-temporal feature sequence according to weighting coefficients to obtain a comprehensive discrimination value. This comprehensive discrimination value is then compared with multi-level landslide risk classification thresholds: when the comprehensive discrimination value is less than the lowest landslide risk classification threshold, the candidate landslide monitoring unit is classified as normal; when the comprehensive discrimination value is between adjacent landslide risk classification thresholds, the candidate landslide monitoring unit is classified as belonging to the corresponding landslide risk level; and when the comprehensive discrimination value is greater than the highest landslide risk classification threshold, the candidate landslide monitoring unit is classified as having a landslide target and being at a high risk level.

[0079] The slope monitoring terminal outputs the collapse risk level and collapse target identification results of the candidate collapse monitoring units. When the collapse risk level reaches the preset high risk level, proceed to step seven; otherwise, proceed to step six.

[0080] The slope monitoring terminal periodically reads processor utilization, power consumption, and equipment temperature to form resource status parameters. These parameters are compared with preset resource thresholds. When processor utilization exceeds the processor resource threshold and / or power consumption is less than the power resource threshold and / or equipment temperature exceeds the temperature resource threshold, the slope monitoring terminal is determined to be in a high-load state.

[0081] Under high load conditions, the slope monitoring terminal reduces the number of candidate collapse monitoring units and the length of multi-temporal feature sequences by increasing the preset sampling period and / or raising the preset change threshold and / or shortening the preset time window, ensuring that the computational load for a single collapse target identification does not exceed the preset computational limit. Then, it returns to step two to perform the next cycle of data acquisition and identification.

[0082] When the collapse risk level of a candidate collapse monitoring unit reaches the preset high risk level, the slope monitoring terminal combines the multi-temporal feature sequence corresponding to the candidate collapse monitoring unit, the aligned current monitoring image data, the aligned current monitoring three-dimensional point cloud data, and the collapse target discrimination result to form the uploaded data, which is then sent to the central device through the communication device.

[0083] The central-end equipment performs verification and discrimination based on baseline environmental data and uploaded data, generates central-end verification and discrimination results, and outputs the results. Within a preset statistical time period, the central-end equipment compares the collapse target discrimination results of the slope monitoring terminal with the central-end verification and discrimination results, calculates the false alarm rate and the missed alarm rate, and compares the false alarm rate and the missed alarm rate with preset thresholds. When the false alarm rate or the missed alarm rate exceeds the preset threshold, one or more of the collapse discrimination parameters, preset sampling period, preset change threshold, and preset time window are updated according to the preset adjustment strategy. The updated parameters are then sent to the slope monitoring terminal through the communication device, realizing closed-loop optimization between the cloud and the slope monitoring terminal.

[0084] Example 2

[0085] A rock slope collapse target identification system, applying the aforementioned rock slope collapse target identification method, includes a central terminal device and a slope monitoring terminal; the slope monitoring terminal is communicatively connected to the central terminal device;

[0086] The central terminal device includes a benchmark environment construction unit, a verification and parameter update unit, and a communication device, wherein the benchmark environment construction unit and the verification and parameter update unit are respectively connected to the communication device;

[0087] The reference environment construction unit is used to acquire initial high-resolution image data and initial three-dimensional point cloud data and generate reference environment data.

[0088] The verification and parameter update unit is used to verify and judge based on the baseline environmental data and the data uploaded by the slope monitoring terminal, to calculate the false alarm rate and the missed alarm rate within the preset statistical time period, and to update one or more of the collapse judgment parameters, preset sampling period, preset change threshold and preset time window according to the false alarm rate and the missed alarm rate and send them to the slope monitoring terminal.

[0089] The slope monitoring terminal includes a data acquisition and registration unit, a multi-source change calculation and candidate screening unit, a multi-temporal feature construction unit, a landslide discrimination unit, a resource status perception and adjustment unit, and a result output and uploading unit. The data acquisition and registration unit, the multi-source change calculation and candidate screening unit, the multi-temporal feature construction unit, the landslide discrimination unit, and the resource status perception and adjustment unit are respectively connected to the result output and uploading unit; the result output and uploading unit is communicatively connected to the communication device.

[0090] The data acquisition and registration unit is used to acquire the current monitoring image data and the current monitoring 3D point cloud data according to a preset sampling period and generate alignment data.

[0091] The multi-source change calculation and candidate screening unit is used to calculate image change index and terrain change index based on aligned data and benchmark environmental data to obtain a comprehensive change index. Spatial monitoring units with a comprehensive change index greater than a preset change threshold are identified as candidate collapse monitoring units.

[0092] The multi-temporal feature construction unit is used to construct a multi-temporal feature sequence of candidate collapse monitoring units within a preset time window;

[0093] The collapse discrimination unit is used to obtain the collapse risk level and collapse target discrimination result of the candidate collapse monitoring unit based on the collapse discrimination parameters;

[0094] The resource status sensing and adjustment unit is used to acquire resource status parameters and control the number of candidate collapse monitoring units and the length of multi-temporal feature sequences according to the resource status parameters, so that the amount of computation for a single collapse target discrimination does not exceed the preset computation limit.

[0095] The result output and upload unit is used to output the collapse risk level and collapse target discrimination result of the candidate collapse monitoring unit, and upload the multi-temporal feature sequence, aligned data and collapse target discrimination result when the collapse risk level reaches the preset high risk level.

[0096] Example 3

[0097] This embodiment, based on the method described in Embodiment 1, provides a specific application method on high rock slopes along mountain highways, illustrating the parameter settings and operation process of the method in actual engineering scenarios.

[0098] There is a high rock slope on the right side of a mountain highway, with a height of approximately 65 meters. The slope surface consists of multiple terraces and gently sloping layers. The design unit assessed that this slope has a risk of collapse and requires long-term monitoring. Therefore, a rock slope collapse target identification system of this invention was installed on this slope, including a central terminal device and a slope monitoring terminal.

[0099] The central equipment utilizes a high-performance industrial server and employs a drone equipped with an oblique photography system to conduct multi-view aerial photography of the target slope, acquiring initial high-resolution image data with a ground resolution of approximately 0.05 m. Simultaneously, a solid-state LiDAR is used to perform a 3D scan of the slope, obtaining initial 3D point cloud data with a point cloud density of approximately 800 points / square meter.

[0100] The central equipment performs geometric correction and orthorectification on the image data, dividing the monitoring area into regular spatial monitoring units of 2 m × 2 m, resulting in several spatial monitoring units. After coordinate unification and filtering of the initial 3D point cloud data, it is projected onto a 2 m × 2 m grid and the average elevation and standard deviation of each grid are calculated to form terrain raster data.

[0101] For each spatial monitoring unit, the central terminal device calculates static morphological parameters—average elevation, slope, aspect, and surface roughness—based on the corresponding area's imagery and topographic raster data, and obtains static morphological characteristics accordingly, forming baseline environmental data. Considering the storage and computing capabilities of the slope monitoring terminal, the central terminal device only packages and distributes the spatial monitoring units corresponding to the terminal's observation field of view, along with their static morphological characteristics, image thumbnails, and topographic raster data, to the slope monitoring terminal.

[0102] The slope monitoring terminal is installed on the outside of the guardrail at the edge of the highway. It uses an industrial camera and a medium-short range lidar as front-end sensors. The terminal contains a low-power embedded processor board (such as an ARM architecture processor) and is powered by a solar panel and a battery.

[0103] During initial setup, the preset sampling period is set to 5 minutes, meaning one frame of current monitoring image data and one frame of current monitoring 3D point cloud data are acquired every 5 minutes. The slope monitoring terminal scales the current monitoring image data to match the target resolution image layer in the baseline environmental data (e.g., 0.2 m resolution), and uses a feature point matching algorithm to complete translation and rotation alignment to obtain registered image data. Then, based on the spatial reference of the registered image data, the current monitoring 3D point cloud data is projected onto a coordinate system consistent with the terrain raster data in the baseline environmental data, completing rasterization processing to obtain aligned data.

[0104] In each sampling period, the slope monitoring terminal calculates image change parameters for each spatial monitoring unit, including the average grayscale difference between the current image and the reference image, the local texture energy difference, and the edge direction change, and combines them to obtain image change indicators; at the same time, it calculates terrain change parameters, including the elevation difference, slope difference, and surface roughness difference between the current terrain raster and the reference terrain raster, and combines them to obtain terrain change indicators.

[0105] To facilitate the integration of indicators with different dimensions, the slope monitoring terminal normalizes the image change indicators and topographic change indicators separately, and then performs a weighted summation according to preset weights to form a comprehensive change indicator. Based on engineering experience, a preset change threshold is set as a certain statistical quantile value (e.g., the 95th percentile) of the comprehensive change indicator, while limiting the number of candidate landslide monitoring units to no more than 10% of the total number of spatial monitoring units. Spatial monitoring units with a comprehensive change indicator greater than the preset change threshold and ranked high are identified as candidate landslide monitoring units.

[0106] In this embodiment, the preset time window is set to the most recent 12 preset sampling periods, corresponding to a time length of 1 hour. Within this time window, the slope monitoring terminal stores 12 consecutive sets of aligned current monitoring image data and current monitoring 3D point cloud data for each candidate collapse monitoring unit, extracts image change features and terrain change features respectively, and combines them in chronological order to form a multi-temporal feature sequence.

[0107] Before system deployment, landslide discrimination parameters are preset by the central terminal equipment using historical data. These parameters mainly include weighting coefficients for each feature dimension and thresholds for multi-level landslide risk classification. For example, four risk levels are set: "normal," "low risk," "medium risk," and "high risk." The slope monitoring terminal uses the landslide discrimination parameters to perform weighted calculations on multi-temporal feature sequences, obtaining a comprehensive discrimination value. This value is then compared with the thresholds for each risk level to obtain the landslide risk level of the candidate landslide monitoring unit and the landslide target discrimination result.

[0108] When a candidate landslide monitoring unit is determined to be "high risk" and there is a landslide target, the slope monitoring terminal immediately issues an audible and visual alarm through the result output and uploading unit, and uploads the multi-temporal characteristic sequence and aligned data to the central terminal device.

[0109] Because this high slope monitoring terminal is powered by solar energy, its power supply may decrease at night. The terminal periodically reads processor utilization, power level, and equipment temperature to form resource status parameters. When the power level falls below a set threshold (e.g., battery level below 30%) or the processor utilization exceeds a set value, the terminal automatically adjusts the preset sampling period from 5 minutes to 15 minutes, appropriately increases the preset change threshold, and shortens the preset time window. This reduces the number of candidate collapse monitoring units and the length of multi-temporal feature sequences, ensuring that the computational load for a single judgment does not exceed the preset computational limit, and guaranteeing continuous operation of the system under low power consumption conditions.

[0110] After receiving high-risk uploaded data, the central terminal equipment verifies and judges the candidate landslide monitoring unit, and calculates the false alarm rate and missed alarm rate within a preset statistical period (e.g., the last 7 days). When the false alarm rate of a certain slope monitoring terminal is found to be high, the preset change threshold of the comprehensive change index can be appropriately increased or the weight coefficient in the landslide judgment parameters can be adjusted; when the missed alarm rate is high, the preset change threshold can be decreased or the preset time window length can be appropriately increased, and the updated parameters can be sent to the corresponding slope monitoring terminal to achieve personalized optimization for the high rock slope of the highway.

[0111] Example 4

[0112] This embodiment illustrates the application of the method of the present invention in an open-pit mining scenario, mainly demonstrating the parameter division and functional division when multiple slope monitoring terminals and a central terminal device work together.

[0113] A certain open-pit mine has several high rock slopes, ranging in height from 40 m to 90 m, with significant differences in orientation and geological structure among the different slopes. To achieve unified monitoring and zoning, a central terminal device is configured at the mine control center, and multiple slope monitoring terminals are deployed on each key slope, with each terminal responsible for several spatial monitoring units.

[0114] The central equipment uses existing aerial photogrammetry data and ground laser scanning data from the mining area to create a unified model of the entire open-pit mine. First, high-resolution images of the entire mining area are stitched together and orthorectified, dividing the mining area into multiple monitoring zones, each corresponding to one or more high rock slopes.

[0115] The central equipment establishes an independent baseline environmental dataset for each monitoring zone. When dividing spatial monitoring units, the scale and importance of each slope are considered. For example, a 1 m × 1 m spatial monitoring unit is used for slopes with a high safety level, while a 3 m × 3 m spatial monitoring unit can be used for slopes with a low safety level. The 3D point cloud data is rasterized to form corresponding terrain raster data; static morphological features are calculated for each spatial monitoring unit to form zone-level baseline environmental data.

[0116] During the deployment phase, the central terminal equipment distributes the baseline environmental data within the corresponding monitoring zone to different slope monitoring terminals as needed, based on the monitoring field of view of each slope monitoring terminal. This enables multiple terminals to share the same baseline environment, but each terminal only holds the required portion.

[0117] For high slopes located deep in the mine, with frequent mining operations and large mining equipment, the slope monitoring terminal uses a shorter preset sampling period, such as 2 minutes, and sets the preset time window to the most recent 20 preset sampling periods in order to capture signs of rapid collapse.

[0118] For high slopes located on the periphery of the mining area with less operational interference, the slope monitoring terminal can use a longer preset sampling period, such as 10 minutes, and set the preset time window to the most recent 12 preset sampling periods to reduce communication and computing burden.

[0119] For different slope monitoring terminals, the central equipment sets different preset change thresholds and preset upper limits based on the slope grade and historical data, making it easier for important slopes to be selected as candidate collapse monitoring units, while the candidate range for general slopes is relatively narrowed.

[0120] Each slope monitoring terminal performs spatial registration of the current monitoring image data and the current monitoring three-dimensional point cloud data according to the methods described in Example 1 and Example 3, calculates image change indicators and terrain change indicators, and forms a comprehensive change indicator.

[0121] For slopes subject to frequent mechanical disturbances, the weight of topographic change index can be set higher in the calculation of comprehensive change index to reduce the impact of local lighting changes and mechanical shading on image change index; for slopes not significantly affected by mining, the weight of image change index can be increased to make full use of information on crack development and rock loosening in the image.

[0122] Each slope monitoring terminal independently completes the screening of candidate collapse monitoring units and records the spatial location of the candidate collapse monitoring units locally for subsequent multi-temporal feature construction and collapse identification.

[0123] In this embodiment, the central terminal device can issue different sets of collapse discrimination parameters to different monitoring zones. For example, for some high slopes with complex rock structures and well-developed joints, the feature weights related to changes in surface roughness are increased in the collapse discrimination parameters; for high slopes dominated by overall displacement, the feature weights related to changes in elevation and slope are increased.

[0124] Each slope monitoring terminal constructs a multi-temporal characteristic sequence within a preset time window and determines the collapse risk level and collapse target based on the collapse discrimination parameters it receives. For candidate collapse monitoring units identified as "medium risk" or above, the slope monitoring terminal can generate detailed log records locally, including comprehensive change indicators, characteristic sequence summaries, and risk level change processes, facilitating subsequent analysis.

[0125] Because power supply conditions in open-pit mines vary from region to region, the slope monitoring terminal in this embodiment supports a tiered operation mode. The resource status sensing and adjustment unit divides the terminal operation mode into "normal mode," "energy-saving mode," and "protection mode" based on resource status parameters.

[0126] In normal mode, the terminal operates according to the preset sampling period and preset time window, performing a complete candidate screening and multi-temporal discrimination process;

[0127] In energy-saving mode (e.g., when the power supply is less than 40%), the terminal appropriately increases the preset sampling period, raises the preset change threshold, and shortens the preset time window. At the same time, it only constructs multi-temporal features for the candidate collapse monitoring units that rank high in the comprehensive change index.

[0128] In protection mode (e.g., when the power supply is below 20% or the device temperature exceeds the high temperature threshold), the terminal retains monitoring of only a very small number of high-risk areas, shuts down some sensors or reduces the sampling frequency, and prioritizes resources for data caching and alarm functions, waiting for external maintenance.

[0129] Through the above-mentioned hierarchical operation mode, the method of the present invention can still maintain the ability to continuously monitor key high slopes in the complex environment of open-pit mines.

[0130] Within a preset statistical period, the central equipment performs statistical analysis on the collapse target identification results uploaded by multiple slope monitoring terminals and the central verification identification results, obtaining the false alarm rate and false negative rate for different terminals. The central equipment can update parameters for individual slope monitoring terminals, or uniformly adjust collapse identification parameters and preset parameters for a certain geological type or a certain monitoring zone.

[0131] For example, when it is found that the overall false alarm rate of a certain type of high slope with well-developed joints is too high, the central terminal equipment can uniformly reduce the weight of the image change index of this type of slope, appropriately increase the weight of the topographic change index, and adjust the threshold for the classification of collapse risk level accordingly. Then, the updated parameters are sent to all slope monitoring terminals under this type of geological condition to achieve cross-terminal parameter collaborative optimization.

[0132] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for identifying targets in rocky high slope collapses, characterized in that, Includes the following steps: Step 1: The central terminal device acquires initial high-resolution image data and initial 3D point cloud data of the rocky high slope, generates benchmark environmental data containing static morphological features of multiple spatial monitoring units, and sends it to the slope monitoring terminal. Step 2: The slope monitoring terminal collects the current monitoring image data and the current monitoring 3D point cloud data according to the preset sampling cycle. The current monitoring image data and the current monitoring 3D point cloud data are then spatially registered with the reference environmental data to obtain aligned data. Step 3: The slope monitoring terminal calculates the image change index and topographic change index of each spatial monitoring unit based on the aligned data and the baseline environmental data to obtain the comprehensive change index. Spatial monitoring units with comprehensive change index greater than the preset change threshold are identified as candidate collapse monitoring units. Step 4: The slope monitoring terminal stores multi-temporal relative alignment data of candidate collapse monitoring units within a preset time window, and obtains multi-temporal feature sequences based on the stored multi-temporal relative alignment data of candidate collapse monitoring units; Step 5: The slope monitoring terminal outputs the collapse risk level and collapse target identification result of the candidate collapse monitoring unit based on the collapse discrimination parameters and multi-temporal characteristic sequence. When the collapse risk level reaches the preset high risk level, proceed to step 7; otherwise, proceed to step 6. Step 6: Obtain the resource status parameters of the slope monitoring terminal, and control the number of candidate collapse monitoring units and the length of multi-temporal feature sequences according to the resource status parameters, so that the calculation amount of a single collapse target identification does not exceed the preset calculation limit, and return to Step 2 to perform the next cycle of identification; Step 7: The slope monitoring terminal uploads the multi-temporal characteristic sequence, alignment data and collapse target discrimination results of the candidate collapse monitoring units to the central device. The central device performs a review and discrimination based on the baseline environmental data and the uploaded data within a preset statistical time period and outputs the review results. The system calculates the false alarm rate and the missed alarm rate, and updates one or more of the collapse discrimination parameters, preset sampling period, preset change threshold and preset time window based on the false alarm rate and the data and sends them to the slope monitoring terminal.

2. The method for identifying landslide targets on a high rock slope according to claim 1, characterized in that, The central terminal device acquires initial high-resolution image data and initial three-dimensional point cloud data of the rocky high slope, generates benchmark environmental data containing static morphological features of multiple spatial monitoring units, and distributes it to the slope monitoring terminal, including: The central equipment acquires initial high-resolution image data through remote sensing photography equipment and initial 3D point cloud data through lidar equipment; it constructs a multi-level resolution image dataset from the initial high-resolution image data, dividing the rocky high slope area into multiple spatial monitoring units; it performs rasterization processing on the initial 3D point cloud data to obtain topographic raster data corresponding to the spatial monitoring units; based on the multi-level resolution image dataset and topographic raster data, it calculates static morphological features according to preset static morphological parameters, which include at least two of the following: average elevation, slope, aspect, and surface roughness, to form baseline environmental data.

3. The method for identifying collapse targets on a high rock slope according to claim 2, characterized in that, The slope monitoring terminal collects current monitoring image data and current monitoring 3D point cloud data according to a preset sampling period, and spatially registers the current monitoring image data and current monitoring 3D point cloud data with the reference environmental data in sequence to obtain aligned data, including: The slope monitoring terminal collects current monitoring image data and current monitoring 3D point cloud data according to a preset sampling cycle. It maps the current monitoring image data to the target resolution image level in the reference environmental data. Based on feature point matching, it performs translational and rotational alignment on the current monitoring image data and the target resolution image data to obtain registered image data. Based on the spatial reference of the registered image data, it projects the current monitoring 3D point cloud data to a coordinate system consistent with the terrain raster data in the reference environmental data and performs rasterization processing to obtain aligned data.

4. The method for identifying collapse targets on a high rock slope according to claim 3, characterized in that, The slope monitoring terminal calculates image change indicators and topographic change indicators for each spatial monitoring unit based on aligned data and baseline environmental data to obtain a comprehensive change index. Spatial monitoring units with comprehensive change indices greater than a preset change threshold are identified as candidate collapse monitoring units, including: For each spatial monitoring unit, based on the aligned current monitoring image data and the corresponding image data in the benchmark environmental data, an image change index is obtained according to the image change parameters. The image change parameters include at least one of grayscale difference, texture change amount, and edge direction change amount. Based on the aligned current terrain raster data and the corresponding terrain raster data in the baseline environmental data, terrain change indicators are obtained according to terrain change parameters. The terrain change parameters include at least one of elevation difference, slope difference, and surface roughness difference. The image change index and the terrain change index are normalized and weighted and summed according to preset weights to obtain the comprehensive change index. Based on the comprehensive change index sorted from high to low, spatial monitoring units with comprehensive change index greater than the preset change threshold and number not greater than the preset upper limit are selected as candidate collapse monitoring units.

5. The method for identifying collapse targets on a high rock slope according to claim 1, characterized in that, The slope monitoring terminal stores multi-temporal relative alignment data of candidate landslide monitoring units within a preset time window, and obtains a multi-temporal feature sequence based on the stored multi-temporal relative alignment data of candidate landslide monitoring units, including: The preset time window covers the most recent M preset sampling periods, where M is a preset positive integer. Within the preset time window, the slope monitoring terminal stores the aligned current monitoring image data and current monitoring 3D point cloud data of the candidate collapse monitoring unit corresponding to each preset sampling period. Image change features are extracted from the aligned current monitoring image data to form an image change feature sequence; terrain change features are extracted from the aligned current monitoring 3D point cloud data to form a terrain change feature sequence; and a multi-temporal feature sequence is obtained based on the image change feature sequence and the terrain change feature sequence.

6. The method for identifying collapse targets on a high rock slope according to claim 1, characterized in that, The slope monitoring terminal outputs the collapse risk level and collapse target identification results of candidate collapse monitoring units based on collapse discrimination parameters and multi-temporal characteristic sequences, including: The collapse discrimination parameters include weighting coefficients for each feature dimension of the multi-temporal feature sequence and thresholds for classifying multi-level collapse risk levels. The slope monitoring terminal weights each feature dimension of the multi-temporal feature sequence according to the weighting coefficients to obtain a comprehensive discrimination value. The comprehensive discrimination value is compared with the thresholds for classifying multi-level collapse risk levels. When the comprehensive discrimination value is less than the lowest threshold for classifying collapse risk level, the candidate collapse monitoring unit is judged as being in a normal state. When the comprehensive discrimination value is between the adjacent thresholds for classifying collapse risk level, the candidate collapse monitoring unit is judged as being at the corresponding collapse risk level. When the comprehensive discrimination value is greater than the highest threshold for classifying collapse risk level, the candidate collapse monitoring unit is judged as having a collapse target and being at a high risk level.

7. The method for identifying collapse targets on a high rock slope according to claim 1, characterized in that, The process of acquiring resource status parameters of slope monitoring terminals and controlling the number of candidate landslide monitoring units and the length of multi-temporal feature sequences based on these parameters, ensuring that the computational load for a single landslide target identification does not exceed a preset computational limit, includes: The slope monitoring terminal periodically reads processor occupancy, power consumption, and equipment temperature to form resource status parameters. These parameters are compared with preset resource thresholds. When processor occupancy exceeds the processor resource threshold and / or power consumption is less than the power resource threshold and / or equipment temperature exceeds the temperature resource threshold, the slope monitoring terminal is determined to be in a high-load state. In this high-load state, the preset sampling period is increased and / or the preset change threshold is raised and / or the preset time window is shortened to ensure that the computational load for a single collapse target identification does not exceed the preset computational limit.

8. The method for identifying collapse targets on a high rock slope according to claim 1, characterized in that, The slope monitoring terminal uploads the multi-temporal feature sequences, alignment data, and collapse target discrimination results of candidate collapse monitoring units to the central terminal device, including: When the collapse risk level of a candidate collapse monitoring unit reaches the preset high risk level, the slope monitoring terminal combines the multi-temporal feature sequence corresponding to the candidate collapse monitoring unit, the aligned current monitoring image data, the aligned current monitoring 3D point cloud data, and the collapse target discrimination result to form the uploaded data and sends it to the central device. The central device performs a review and discrimination based on the baseline environmental data and the uploaded data, generates the central review and discrimination result, and outputs the review result.

9. The method for identifying collapse targets on a high rock slope according to claim 1, characterized in that, The aforementioned statistical false alarm rate and false negative rate are used to update one or more of the collapse discrimination parameters, preset sampling period, preset change threshold, and preset time window based on the false alarm rate and false negative rate, and the data is then sent to the slope monitoring terminal, including: Within a preset statistical time period, the central terminal equipment compares the collapse target identification results of the slope monitoring terminal with the central terminal verification identification results, calculates the false alarm rate and the missed alarm rate, and compares the false alarm rate and the missed alarm rate with preset thresholds. When the false alarm rate or the missed alarm rate is greater than the preset threshold, it updates one or more of the collapse identification parameters, preset sampling period, preset change threshold and preset time window according to the preset adjustment strategy, forms the updated parameters and sends them to the slope monitoring terminal.

10. A rock slope collapse target identification system, employing the rock slope collapse target identification method according to any one of claims 1-9, characterized in that, It includes a central terminal device and a slope monitoring terminal; the slope monitoring terminal is communicatively connected to the central terminal device; The central terminal device includes a benchmark environment construction unit, a verification and parameter update unit, and a communication device, wherein the benchmark environment construction unit and the verification and parameter update unit are respectively connected to the communication device; The slope monitoring terminal includes a data acquisition and registration unit, a multi-source change calculation and candidate screening unit, a multi-temporal feature construction unit, a collapse discrimination unit, a resource status perception and adjustment unit, and a result output and uploading unit. The data acquisition and registration unit, the multi-source change calculation and candidate screening unit, the multi-temporal feature construction unit, the collapse discrimination unit, and the resource status perception and adjustment unit are respectively connected to the result output and upload unit; the result output and upload unit is communicatively connected to the communication device.

Images