Construction site environment state monitoring method based on satellite image and unmanned aerial vehicle data

Abstract

The present application belongs to the technical field of environmental state monitoring, and specifically relates to a construction site environmental state monitoring method based on satellite images and unmanned aerial vehicle data, steps of which include: collecting satellite image monitoring information and unmanned aerial vehicle monitoring information of the construction site environmental state; performing construction site environmental surface coverage change analysis to output construction site environmental change characteristics; obtaining PM2.5 concentration characteristics based on a portable PM2.5 monitor arranged at the construction site; performing construction site equipment YOLO target detection to output equipment distribution heat maps and determine construction site construction progress characteristics; constructing a construction site safety risk framework to determine the construction site safety risk level; and determining the construction site environmental state monitoring level. The present application realizes comprehensive and dynamic monitoring of the construction site environmental state by fusing satellite images and unmanned aerial vehicle data, improves monitoring accuracy and efficiency, and provides strong support for scientific management of the construction site.

Description

Technical Field

[0001] The present invention belongs to the technical field of environmental status monitoring, and in particular relates to a construction site environmental status monitoring method based on satellite images and drone data. Background Art

[0002] Currently, construction site environmental status relies heavily on manual inspections, which require significant manpower, have limited coverage, and are difficult to conduct frequently. This makes it difficult to fully and timely assess environmental conditions at large construction sites with complex terrain. Conventional environmental monitoring equipment (such as simple dust monitors) is limited in its functionality and lacks the accuracy and comprehensiveness of environmental information it collects. This makes it difficult to accurately capture subtle changes in layout.

[0003] Satellite imagery offers wide coverage and can periodically acquire data over large areas. This can be used to monitor changes in land occupation at construction sites and the impact on the surrounding ecological environment. With the continuous advancement of satellite remote sensing technology, the spatial resolution of high-resolution satellite imagery has significantly increased (reaching sub-meter levels), and the spectral resolution has also continued to improve. This makes it possible to identify different land features at construction sites (such as construction material storage areas and exposed land), providing a higher-quality data foundation for environmental status monitoring. Compared to satellite imagery, which is limited by its revisit cycle, drones can launch on demand and quickly acquire real-time data, enabling timely response and monitoring of sudden environmental changes at construction sites (such as sudden increases in dust caused by temporary earthwork excavation).

[0004] Therefore, it is possible to consider monitoring the environmental status of the construction site based on satellite images and drones. Summary of the Invention

[0005] In response to the shortcomings of existing technologies, the present invention provides a construction site environmental status monitoring method based on satellite imagery and drone data. By integrating satellite imagery and drone data, comprehensive and dynamic monitoring of the construction site environmental status is achieved, improving monitoring accuracy and efficiency, and providing strong support for the scientific management of construction sites.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0007] The construction site environmental status monitoring method based on satellite imagery and drone data includes the following steps:

[0008] S1. Collect satellite imagery and drone monitoring information on the construction site's environmental status. Satellite imagery includes multispectral images, thermal infrared data, high-resolution optical images, and radar satellite data. Drone monitoring includes high-resolution RGB images and LiDAR point cloud data.

[0009] S2. Based on the multispectral imagery and high-resolution RGB imagery of the construction site, analyze the surface cover changes in the construction site environment and output the characteristics of the construction site environment changes;

[0010] S3. Obtain PM2.5 concentration characteristics based on a portable PM2.5 monitor set up at the construction site;

[0011] S4. Based on the high-resolution optical image and high-resolution RGB image of the construction site, perform YOLO target detection on the construction site equipment, output the equipment distribution heat map, and determine the construction progress characteristics of the construction site;

[0012] S5. Based on the radar satellite data and LiDAR point cloud data of the construction site, a construction site safety risk framework is established to determine the construction site safety risk level;

[0013] S6. Determine the environmental status monitoring level of the construction site based on the environmental change characteristics of the construction site, PM2.5 concentration characteristics, construction progress characteristics of the construction site and the safety risk level of the construction site.

[0014] Preferably, in said S2, the process of outputting the construction site environment change characteristics is:

[0015] The multispectral images and high-resolution RGB images of the two construction sites were atmospherically corrected using the QUAC module of the remote sensing image processing platform ENVI. The two images were aligned to the WGS84 coordinate system using the gdalwarp tool of the raster spatial data conversion library GDAL, and geometric registration was performed to obtain the pre-processed multispectral images and high-resolution RGB images of the two construction sites.

[0016] Each band of the pre-processed multispectral images of the two construction sites is stacked into a multidimensional array;

[0017] Use the PCA function of scikit-learn to reduce the dimension, retain the first three principal component values, and obtain the principal component values ​​of two periods;

[0018] Based on the pre-processed high-resolution RGB images of the two construction sites, the average color values ​​of the two phases were extracted;

[0019] The principal component values ​​of the two periods, the original band reflectance, and the average color values ​​of the two periods were input into the trained random forest model to obtain the probability map of vegetation reduction in the construction site environment;

[0020] Based on the vegetation reduction probability map of the construction site environment, the characteristics of the construction site environment changes are output.

[0021] Preferably, the process of outputting the construction site environment change characteristics based on the construction site environment vegetation reduction probability map is:

[0022] Obtaining the construction site environment vegetation reduction probability threshold stored in the database;

[0023] Compare the construction site environment vegetation reduction probability threshold with each construction site environment vegetation reduction probability in the construction site environment vegetation reduction probability map;

[0024] Count the proportion of construction site vegetation reduction probabilities that exceed the construction site vegetation reduction probability threshold, and mark them as construction site vegetation reduction risk signals;

[0025] Output the characteristics of construction site environmental changes, specifically the risk signal of vegetation reduction in the construction site environment.

[0026] Preferably, in said S4, the process of outputting the equipment distribution heat map and determining the construction progress characteristics of the construction site is:

[0027] Use the labeled data set stored in the database to train the YOLO model to obtain a trained YOLO model;

[0028] Apply the trained YOLO model to high-resolution optical images and high-resolution RGB images of the construction site;

[0029] The YOLO model detects construction site equipment, including excavators, cranes, and trucks, and draws bounding boxes around the equipment locations.

[0030] Apply non-maxima suppression to merge overlapping bounding boxes:

[0031] Sort all bounding boxes by classification confidence from high to low, select the bounding box with the highest classification confidence as the retained box, calculate the intersection-over-union (IoU) of the retained box with all remaining bounding boxes, remove the bounding boxes with an IoU greater than the set IoU threshold, and continue to select the highest classification confidence box from the unprocessed bounding boxes until all bounding boxes are processed;

[0032] Filter out detection results with low classification confidence according to the classification confidence threshold stored in the database;

[0033] Output device distribution heat map;

[0034] Determine construction progress characteristics at the construction site based on equipment distribution heat maps.

[0035] Preferably, the process of determining the construction progress characteristics of the construction site based on the equipment distribution heat map is:

[0036] Determine the number of excavators, cranes, and trucks based on the equipment distribution heat map;

[0037] Store the number of excavators, cranes, and trucks as specified labels;

[0038] Obtaining a mapping set of specified tags and construction site construction progress ratios pre-stored in the database;

[0039] Based on the currently specified label, the construction progress ratio of the matching construction site is determined and marked as the construction progress feature of the construction site.

[0040] Preferably, in said S5, the process of determining the safety risk level of the construction site is:

[0041] Based on radar satellite data and LiDAR point cloud data of the construction site, a construction site safety risk framework is constructed to extract construction site safety risk data, including the distribution density of equipment on the construction site. , Total number of construction site equipment , Construction site operation area , Maximum slope of the construction site slope ;

[0042] Based on the construction site safety risk data, a comprehensive analysis is conducted to obtain the construction site safety risk factors;

[0043] Compare the construction site safety risk factor with the construction site safety moderate risk range stored in the database;

[0044] If the construction site safety risk factor falls within the medium risk range of construction site safety, the construction site safety risk level is medium risk;

[0045] If the construction site safety risk factor does not fall within the moderate construction site safety risk range, and the construction site safety risk factor is higher than the maximum value of the moderate construction site safety risk range, the construction site safety risk level is high risk;

[0046] If the construction site safety risk factor does not fall within the moderate risk range of construction site safety, and the construction site safety risk factor is lower than the minimum value of the moderate risk range of construction site safety, the construction site safety risk level is low risk.

[0047] Preferably, the construction site safety risk factors are obtained in the following manner:

[0048] ;

[0049] Where, is a safety risk factor at the construction site. Define density for equipment distribution on construction sites, Define the total number of construction site equipment, The area of ​​the construction site operation area is defined, e is a natural constant, and when calculating, 、 、 、 、 、 、 Perform dimensionless processing.

[0050] Preferably, in said S6, the process of determining the construction site environmental status monitoring level is:

[0051] Store construction site environmental change characteristics, PM2.5 concentration characteristics, construction site construction progress characteristics, and construction site safety risk levels as designated tags;

[0052] Obtaining a mapping set of designated tags and construction site environmental status monitoring levels pre-stored in the database;

[0053] Based on the current specified tag, determine the matching construction site environmental status monitoring level;

[0054] If the construction site environmental status monitoring level is safe, construction will continue;

[0055] If the construction site environmental status monitoring level is dangerous, a danger warning will be issued to the monitoring center and an emergency control plan will be obtained.

[0056] Preferably, the process of obtaining the emergency control plan is:

[0057] Integrate the construction site environmental change characteristics, PM2.5 concentration characteristics, and construction site progress characteristics into comprehensive data on the construction site environmental status;

[0058] Obtain the comprehensive data-emergency plan mapping set stored in the database, perform similarity analysis on the comprehensive data of the construction site environmental status and the comprehensive data in the comprehensive data-emergency plan mapping set, determine the comprehensive data that is most similar to the comprehensive data of the construction site environmental status, and map the corresponding emergency plan, which is recorded as the emergency control plan.

[0059] Preferably, the process of acquiring the comprehensive data-contingency plan mapping set stored in the database is:

[0060] Collect environmental status data of historically similar construction sites to obtain a set of environmental status data of historically similar construction sites, including historical environmental status data at different construction stages, where the historical environmental status data includes historical construction site environmental change characteristics, historical PM2.5 concentration characteristics, and historical construction site construction progress characteristics;

[0061] Based on the K-means clustering algorithm, cluster analysis is performed on the historical environmental status data of different construction stages to determine multiple cluster centers, each of which corresponds to at least one historical environmental status data;

[0062] Perform mean processing on the historical environmental status data corresponding to each cluster center to obtain the averaged historical environmental status data, which is recorded as comprehensive data;

[0063] Obtain the historical control plan corresponding to each historical environmental status data and group them according to the cluster center;

[0064] Obtain the control effect data of each historical control scheme, including the growth rate of the proportion of construction site vegetation reduction probability not exceeding the construction site vegetation reduction probability threshold , Construction site construction progress ratio growth rate and PM2.5 concentration reduction rate ;

[0065] Determine the control effect coefficient Kx based on the control effect data:

[0066] ;

[0067] in, , and All are weight factors and dimensionless;

[0068] Determine the historical control plan corresponding to the largest control effect coefficient in each group and record it as the emergency plan;

[0069] The comprehensive data corresponding to each cluster center is associated with the emergency plan to obtain a comprehensive data-emergency plan mapping set and store it in the database.

[0070] The present invention has the following beneficial effects:

[0071] The present invention comprehensively utilizes multispectral satellite imagery, thermal infrared data, high-resolution optical imagery, radar satellite data, and high-resolution RGB imagery and LiDAR point cloud data from drones to obtain information about the construction site from multiple dimensions and at different scales, comprehensively reflecting the environmental status of the construction site. Drones can take off at any time as needed to obtain real-time data. Combined with periodic monitoring of satellite imagery, they can achieve real-time dynamic monitoring of the environmental status of the construction site, promptly identifying and resolving problems. By combining environmental change characteristics, PM2.5 concentration characteristics, construction progress characteristics, and safety risk levels to determine the environmental status monitoring level of the construction site, scientific decision-making and management can be made.

[0072] By analyzing multispectral images and high-resolution RGB images, the present invention can accurately detect changes in the surface coverage of the construction site environment and output accurate environmental change characteristics, which helps to timely grasp the usage of the site and environmental evolution trends. Dust pollution analysis is carried out in combination with thermal infrared data and meteorological station data to obtain PM2.5 concentration characteristics, which can scientifically evaluate the dust pollution situation and provide a basis for taking targeted environmental protection measures. High-resolution optical images and high-resolution RGB images are used for equipment YOLO target detection, and equipment distribution heat maps are output. The construction progress characteristics can be effectively determined, which makes it easier for managers to understand the progress of the project in a timely manner and arrange resources reasonably. A safety risk framework is constructed based on radar satellite data and LiDAR point cloud data to determine the safety risk level, providing a reliable basis for safety management of the construction site and helping to prevent and respond to safety accidents in advance. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 Schematic diagram of the process of the present invention;

[0074] Figure 2 Flowchart of the steps for determining the environmental condition monitoring level of a construction site. DETAILED DESCRIPTION

[0075] The technical solutions in the embodiments of the present invention are described clearly and completely below with reference to the accompanying drawings.

[0076] Example 1: Figure 1 As shown in FIG, the construction site environmental status monitoring method based on satellite imagery and drone data includes the following steps:

[0077] S1. Collect satellite imagery and drone monitoring information on the construction site's environmental status. Satellite imagery includes multispectral images, thermal infrared data, high-resolution optical images, and radar satellite data. Drone monitoring includes high-resolution RGB images and LiDAR point cloud data.

[0078] S2. Based on the multispectral imagery and high-resolution RGB imagery of the construction site, analyze the surface cover changes in the construction site environment and output the characteristics of the construction site environment changes;

[0079] S3. Obtain PM2.5 concentration characteristics based on a portable PM2.5 monitor set up at the construction site;

[0080] S4. Based on the high-resolution optical image and high-resolution RGB image of the construction site, perform YOLO target detection on the construction site equipment, output the equipment distribution heat map, and determine the construction progress characteristics of the construction site;

[0081] S5. Based on the radar satellite data and LiDAR point cloud data of the construction site, a construction site safety risk framework is established to determine the construction site safety risk level;

[0082] S6. Determine the environmental status monitoring level of the construction site based on the environmental change characteristics of the construction site, PM2.5 concentration characteristics, construction progress characteristics of the construction site and the safety risk level of the construction site.

[0083] In S2, the process of outputting the characteristics of construction site environment changes is as follows:

[0084] The multispectral images and high-resolution RGB images of the two construction sites were atmospherically corrected using the QUAC module of the remote sensing image processing platform ENVI. The two images were aligned to the WGS84 coordinate system using the gdalwarp tool of the raster spatial data conversion library GDAL, and geometric registration was performed to obtain the pre-processed multispectral images and high-resolution RGB images of the two construction sites.

[0085] Each band of the pre-processed multispectral images of the two construction sites is stacked into a multidimensional array;

[0086] Use the PCA function of scikit-learn (an open-source Python machine learning library) to reduce the dimension, retain the first three principal component values, and obtain the principal component values ​​for two periods (retain the first three principal component values ​​for each period);

[0087] Based on the pre-processed high-resolution RGB images of the two construction sites, the average color values ​​of the two phases were extracted;

[0088] The principal component values ​​of the two periods, the original band reflectance, and the average color values ​​of the two periods were input into the trained random forest model to obtain the probability map of vegetation reduction in the construction site environment;

[0089] Based on the vegetation reduction probability map of the construction site environment, the characteristics of the construction site environment change are output. The process is as follows:

[0090] Obtaining the construction site environment vegetation reduction probability threshold stored in the database;

[0091] Compare the construction site environment vegetation reduction probability threshold with each construction site environment vegetation reduction probability in the construction site environment vegetation reduction probability map;

[0092] Count the proportion of construction site vegetation reduction probabilities that exceed the construction site vegetation reduction probability threshold, and mark them as construction site vegetation reduction risk signals;

[0093] Output the characteristics of construction site environmental changes, specifically the risk signal of vegetation reduction in the construction site environment.

[0094] Integrating satellite imagery (multispectral, thermal infrared, etc.) and drone data (high-resolution RGB and LiDAR point clouds) provides comprehensive coverage of construction site information, capturing everything from macroscopic site layout to microscopic ground feature details. This provides a rich and comprehensive data foundation for subsequent analysis, improving monitoring accuracy and reliability. Using ENVI's QUAC module for atmospheric correction and GDAL's gdalwarp tool for geometric registration, we effectively eliminate atmospheric interference and image geometric distortion, ensuring data quality and making analysis results more reliable.

[0095] By stacking multispectral image bands and performing PCA dimensionality reduction, data dimensions are reduced while retaining key information, improving processing efficiency. Combining multiple features, such as the average color value of RGB images, with input into a random forest model, we can uncover potential relationships between complex data, accurately generate a vegetation loss probability map, and thus identify environmental change characteristics. By setting a vegetation loss probability threshold, comparing the probability map with the threshold, and calculating the percentage, we can quantitatively output a vegetation loss risk signal, providing clear and measurable indicators for construction site environmental risk assessments, allowing managers to intuitively understand the degree of environmental change risk.

[0096] In S4, the process of outputting the equipment distribution heat map and determining the construction progress characteristics of the construction site is as follows:

[0097] Use the labeled data set stored in the database to train the YOLO model to obtain a trained YOLO model;

[0098] Apply the trained YOLO model to high-resolution optical images and high-resolution RGB images of the construction site;

[0099] The YOLO model detects construction site equipment, including excavators, cranes, and trucks, and draws bounding boxes around the equipment locations.

[0100] Apply non-maxima suppression to merge overlapping bounding boxes:

[0101] Sort all bounding boxes by classification confidence from high to low, select the bounding box with the highest classification confidence as the retained box, calculate the intersection-over-union (IoU) of the retained box with all remaining bounding boxes, remove the bounding boxes with an IoU greater than the set IoU threshold, and continue to select the highest classification confidence box from the unprocessed bounding boxes until all bounding boxes are processed;

[0102] Filter out detection results with low classification confidence according to the classification confidence threshold stored in the database;

[0103] Output device distribution heat map;

[0104] Based on the equipment distribution heat map, the construction progress characteristics of the construction site are determined. The process is as follows:

[0105] Determine the number of excavators, cranes, and trucks based on the equipment distribution heat map;

[0106] Store the number of excavators, cranes, and trucks as specified labels;

[0107] Obtaining a mapping set of specified tags and construction site construction progress ratios pre-stored in the database;

[0108] Based on the currently specified label, the construction progress ratio of the matching construction site is determined and marked as the construction progress feature of the construction site.

[0109] The YOLO model is used to detect objects on construction site equipment. YOLO is a fast and accurate object detection algorithm that can quickly identify equipment such as excavators, cranes, and trucks in high-resolution optical and RGB images. This significantly improves the efficiency of equipment detection and saves significant time and labor costs compared to traditional detection methods.

[0110] Non-maximum suppression (NMS) is used to merge overlapping bounding boxes, effectively avoiding duplicate detection of the same device. By sorting bounding boxes by classification confidence, calculating the intersection over union (IoU), and removing bounding boxes with high overlap, the location and number of devices can be accurately determined, improving the accuracy of detection results.

[0111] The detection results with low classification confidence are filtered out according to the classification confidence threshold stored in the database to further eliminate false detections, making the final detection results more reliable and providing a high-quality data foundation for subsequent analysis.

[0112] Output a heat map of equipment distribution to visually display the distribution of equipment on the construction site. This allows managers to quickly understand equipment concentration areas and usage frequency, providing a clear overview of the overall construction site layout and resource allocation.

[0113] The periodic collection of satellite imagery and drone data enables real-time dynamic monitoring of construction site equipment and progress. Compared to traditional manual counting of equipment and progress, this allows for more timely identification of potential construction issues (such as idle equipment and delayed progress) and prompt implementation of corrective measures.

[0114] In S5, the process of determining the safety risk level of the construction site is as follows:

[0115] Based on radar satellite data and LiDAR point cloud data of the construction site, a construction site safety risk framework is constructed to extract construction site safety risk data, including the distribution density of equipment on the construction site. , Total number of construction site equipment , Construction site operation area , Maximum slope of the construction site slope ;

[0116] Based on the construction site safety risk data, a comprehensive analysis is conducted to obtain the construction site safety risk factors;

[0117] Compare the construction site safety risk factor with the construction site safety moderate risk range stored in the database;

[0118] If the construction site safety risk factor falls within the medium risk range of construction site safety, the construction site safety risk level is medium risk;

[0119] If the construction site safety risk factor does not fall within the moderate construction site safety risk range, and the construction site safety risk factor is higher than the maximum value of the moderate construction site safety risk range, the construction site safety risk level is high risk;

[0120] If the construction site safety risk factor does not fall within the moderate construction site safety risk range and is lower than the minimum value of the moderate construction site safety risk range, the construction site safety risk level is low risk. The moderate construction site safety risk range is a closed interval.

[0121] The construction site safety risk factors are obtained as follows:

[0122] ;

[0123] Where, is a safety risk factor at the construction site. Define density for equipment distribution on construction sites, Define the total number of construction site equipment, The area of ​​the construction site operation area is defined, e is a natural constant, and when calculating, 、 、 、 、 、 、 Perform dimensionless processing.

[0124] Comprehensive utilization of radar satellite data and LiDAR point cloud data. The former can obtain data under different weather conditions and is not affected by cloud cover. The latter can accurately obtain three-dimensional spatial information. The combination of the two can fully capture the spatial layout, equipment distribution and other conditions of the construction site, providing comprehensive and accurate data support for safety risk assessment.

[0125] Key risk data such as equipment distribution density, total number of equipment, operating area, and maximum slope gradient are extracted from the data. These data cover important aspects such as equipment layout, site utilization, and terrain conditions at the construction site. They are core elements for assessing safety risks, making risk assessment more targeted and scientific.

[0126] By constructing a safety risk factor calculation formula, various risk data are quantified and integrated. This mathematical model comprehensively considers the impact of factors such as equipment distribution, quantity, operating area, and terrain slope on safety risks, making risk assessment results more objective and accurate, and avoiding errors caused by subjective judgment.

[0127] By comparing safety risk factors with a pre-defined medium-risk range, we clearly categorize low, medium, and high risk levels. This standardized risk classification allows managers to quickly understand the safety status of construction sites and provides clear guidance for implementing appropriate safety management measures.

[0128] In S6, the process of determining the construction site environmental status monitoring level is as follows:

[0129] like Figure 2 As shown, the construction site environmental change characteristics, PM2.5 concentration characteristics, construction site construction progress characteristics and construction site safety risk level are stored as specified tags;

[0130] Obtaining a mapping set of designated tags and construction site environmental status monitoring levels pre-stored in the database;

[0131] Based on the current specified tag, determine the matching construction site environmental status monitoring level;

[0132] If the construction site environmental status monitoring level is safe, construction will continue;

[0133] If the construction site environmental status monitoring level is dangerous, a danger warning will be issued to the monitoring center and an emergency control plan will be obtained.

[0134] The integration of multiple features such as environmental changes, PM2.5 concentration, construction progress, and safety risks changes the one-sidedness of single-indicator evaluation, can fully depict the environmental status of the construction site, and provide managers with complete and accurate on-site information.

[0135] Clearly defined monitoring levels correspond to different response methods (continuing construction if safe, issuing an early warning if dangerous), providing clear guidance for construction decisions, avoiding blind decisions, and ensuring that construction activities proceed within a safe framework. When the monitoring level is determined to be dangerous, an early warning is quickly issued to the monitoring center, allowing relevant personnel to be promptly informed of potential risks, quickly respond, and take countermeasures, reducing the probability of accidents and the severity of the damage.

[0136] The process of obtaining an emergency control plan is as follows:

[0137] Integrate the construction site environmental change characteristics, PM2.5 concentration characteristics, and construction site progress characteristics into comprehensive data on the construction site environmental status;

[0138] Obtain the comprehensive data-emergency plan mapping set stored in the database, perform similarity analysis on the comprehensive data of the construction site environmental status and the comprehensive data in the comprehensive data-emergency plan mapping set, determine the comprehensive data that is most similar to the comprehensive data of the construction site environmental status, and map the corresponding emergency plan, which is recorded as the emergency control plan.

[0139] Based on the similarity analysis between the comprehensive data of the construction site environmental status and the comprehensive data-emergency plan mapping set stored in the database, the emergency plan corresponding to the comprehensive data most similar to the current construction site situation is accurately determined, making the acquisition of emergency control plans highly targeted and scientific, effectively improving the accuracy and effectiveness of emergency response, providing strong support for safety management of the construction site, and reducing potential risks.

[0140] The process of obtaining the comprehensive data-contingency plan mapping set stored in the database is as follows:

[0141] Collect environmental status data of historically similar construction sites to obtain a set of environmental status data of historically similar construction sites, including historical environmental status data at different construction stages, where the historical environmental status data includes historical construction site environmental change characteristics, historical PM2.5 concentration characteristics, and historical construction site construction progress characteristics;

[0142] Based on the K-means clustering algorithm, cluster analysis is performed on the historical environmental status data of different construction stages to determine multiple cluster centers, each of which corresponds to at least one historical environmental status data;

[0143] Perform mean processing on the historical environmental status data corresponding to each cluster center to obtain the averaged historical environmental status data, which is recorded as comprehensive data;

[0144] Obtain the historical control plan corresponding to each historical environmental status data and group them according to the cluster center;

[0145] Obtain the control effect data of each historical control scheme, including the growth rate of the proportion of construction site vegetation reduction probability not exceeding the construction site vegetation reduction probability threshold , Construction site construction progress ratio growth rate and PM2.5 concentration reduction rate ;

[0146] Determine the control effect coefficient Kx based on the control effect data:

[0147] ;

[0148] in, , and All are weight factors and dimensionless;

[0149] Determine the historical control plan corresponding to the largest control effect coefficient in each group and record it as the emergency plan;

[0150] The comprehensive data corresponding to each cluster center is associated with the emergency plan to obtain a comprehensive data-emergency plan mapping set and store it in the database.

[0151] By making full use of the environmental status data of historically similar construction sites and the relevant control plans and effect data, and through cluster analysis, mean processing and other methods, a comprehensive data-emergency plan mapping set is formed, which integrates a large amount of historical experience, so that the formulation of emergency control plans is no longer based on guesswork, but has a basis to rely on. It can better adapt to the complex and changing environment of the construction site, improve the intelligent level of construction management and emergency response capabilities, and ensure the smooth progress, safety and stability of the construction process.

[0152] Example 2: Based on Example 1, when constructing the construction site safety risk framework, the construction site slope stability factor is also considered:

[0153] Generate a digital elevation model of the construction site based on LiDAR point cloud data and extract slope information;

[0154] Using slope gradient information, combined with soil type and rainfall data, slope stability indices are calculated;

[0155] The slope stability index is added as new safety risk data to the construction site safety risk framework to more comprehensively determine the construction site safety risk level. If the slope stability index is lower than the set slope stability threshold and the construction site safety risk factor is in the medium risk or high risk range, the construction site safety risk level will be increased by one level on the original basis.

[0156] Example 3: Based on Example 1, after outputting the construction site environmental change characteristics, the construction site environmental vegetation restoration potential assessment is also performed. The process is as follows:

[0157] Construct a vegetation restoration potential model based on the construction site environmental vegetation reduction probability map and soil moisture data;

[0158] The soil moisture, slope information of the vegetation loss area and the distance to the nearest surviving vegetation area were used as input features, and the trained support vector machine model was used to predict the vegetation restoration potential level.

[0159] If the vegetation restoration potential level is high, a recommended area for natural vegetation restoration will be generated in the environmental status monitoring report;

[0160] If the vegetation restoration potential level is medium or low, manual intervention measures will be recommended in combination with the applicability analysis of drone spraying and grass planting technology, and specific intervention locations and priorities will be provided in the monitoring report.

Claims

1. A construction site environmental status monitoring method based on satellite images and drone data, characterized by The following steps are involved: S1. Collect satellite imagery and drone monitoring information on the construction site's environmental status. Satellite imagery includes multispectral images, thermal infrared data, high-resolution optical images, and radar satellite data. Drone monitoring includes high-resolution RGB images and LiDAR point cloud data. S2. Based on the multispectral imagery and high-resolution RGB imagery of the construction site, analyze the surface cover changes in the construction site environment and output the characteristics of the construction site environment changes; S3. Obtain PM2.5 concentration characteristics based on a portable PM2.5 monitor set up at the construction site; S4. Based on the high-resolution optical image and high-resolution RGB image of the construction site, perform YOLO target detection on the construction site equipment, output the equipment distribution heat map, and determine the construction progress characteristics of the construction site; S5. Based on the radar satellite data and LiDAR point cloud data of the construction site, a construction site safety risk framework is constructed to determine the construction site safety risk level. The process is as follows: Based on radar satellite data and LiDAR point cloud data of the construction site, a construction site safety risk framework is constructed to extract construction site safety risk data, including the distribution density of equipment on the construction site. , Total number of construction site equipment , Construction site operation area , Maximum slope of the construction site slope ; Based on the construction site safety risk data, a comprehensive analysis was conducted to obtain the construction site safety risk factors: ； Where, is a safety risk factor at the construction site. Define density for equipment distribution on construction sites, Define the total number of construction site equipment, The area of ​​the construction site operation area is defined, e is a natural constant, and when calculating, 、 、 、 、 、 、 Perform dimensionless processing; Compare the construction site safety risk factor with the construction site safety moderate risk range stored in the database; If the construction site safety risk factor falls within the medium risk range of construction site safety, the construction site safety risk level is medium risk; If the construction site safety risk factor does not fall within the moderate construction site safety risk range, and the construction site safety risk factor is higher than the maximum value of the moderate construction site safety risk range, the construction site safety risk level is high risk; If the construction site safety risk factor does not fall within the moderate construction site safety risk range, and the construction site safety risk factor is lower than the minimum value of the moderate construction site safety risk range, the construction site safety risk level is low risk; S6. Determine the construction site environmental status monitoring level based on the construction site environmental change characteristics, PM2.5 concentration characteristics, construction site construction progress characteristics, and construction site safety risk level. The process is as follows: Store construction site environmental change characteristics, PM2.5 concentration characteristics, construction site construction progress characteristics, and construction site safety risk levels as designated tags; Obtaining a mapping set of designated tags and construction site environmental status monitoring levels pre-stored in the database; Based on the current specified tag, determine the matching construction site environmental status monitoring level; If the construction site environmental status monitoring level is safe, construction will continue; If the construction site environmental status monitoring level is dangerous, a danger warning will be issued to the monitoring center and an emergency control plan will be obtained.

2. The method for monitoring construction site environmental conditions based on satellite images and drone data according to claim 1, characterized in that: In S2, the process of outputting the construction site environment change characteristics is as follows: The multispectral images and high-resolution RGB images of the two construction sites were atmospherically corrected using the QUAC module of the remote sensing image processing platform ENVI. The two images were aligned to the WGS84 coordinate system using the gdalwarp tool of the raster spatial data conversion library GDAL, and geometric registration was performed to obtain the pre-processed multispectral images and high-resolution RGB images of the two construction sites. Each band of the pre-processed multispectral images of the two construction sites is stacked into a multidimensional array; Use the PCA function of scikit-learn to reduce the dimension, retain the first three principal component values, and obtain the principal component values ​​of two periods; Based on the pre-processed high-resolution RGB images of the two construction sites, the average color values ​​of the two phases were extracted; The principal component values ​​of the two periods, the original band reflectance, and the average color values ​​of the two periods were input into the trained random forest model to obtain the probability map of vegetation reduction in the construction site environment; Based on the vegetation reduction probability map of the construction site environment, the characteristics of the construction site environment changes are output.

3. The method for monitoring construction site environmental conditions based on satellite images and drone data according to claim 2, characterized in that: The process of outputting the construction site environment change characteristics based on the construction site environment vegetation reduction probability map is as follows: Obtaining the construction site environment vegetation reduction probability threshold stored in the database; Compare the construction site environment vegetation reduction probability threshold with each construction site environment vegetation reduction probability in the construction site environment vegetation reduction probability map; Count the proportion of construction site vegetation reduction probabilities that exceed the construction site vegetation reduction probability threshold, and mark them as construction site vegetation reduction risk signals; Output the characteristics of construction site environmental changes, specifically the risk signal of vegetation reduction in the construction site environment.

4. The method for monitoring construction site environmental conditions based on satellite images and drone data according to claim 1, characterized in that: In the aforementioned S4, the process of outputting the equipment distribution heat map and determining the construction progress characteristics of the construction site is as follows: Use the labeled data set stored in the database to train the YOLO model to obtain a trained YOLO model; Apply the trained YOLO model to high-resolution optical images and high-resolution RGB images of the construction site; The YOLO model detects construction site equipment, including excavators, cranes, and trucks, and draws bounding boxes around the equipment locations. Apply non-maxima suppression to merge overlapping bounding boxes: Sort all bounding boxes by classification confidence from high to low, select the bounding box with the highest classification confidence as the retained box, calculate the intersection-over-union (IoU) of the retained box with all remaining bounding boxes, remove the bounding boxes with an IoU greater than the set IoU threshold, and continue to select the highest classification confidence box from the unprocessed bounding boxes until all bounding boxes are processed; Filter out detection results with low classification confidence according to the classification confidence threshold stored in the database; Output device distribution heat map; Determine construction progress characteristics at the construction site based on equipment distribution heat maps.

5. The method for monitoring construction site environmental conditions based on satellite images and drone data according to claim 4 is characterized in that: The process of determining the construction progress characteristics of the construction site based on the equipment distribution heat map is as follows: Determine the number of excavators, cranes, and trucks based on the equipment distribution heat map; Store the number of excavators, cranes, and trucks as specified labels; Obtaining a mapping set of specified tags and construction site construction progress ratios pre-stored in the database; Based on the currently specified label, the construction progress ratio of the matching construction site is determined and marked as the construction progress feature of the construction site.

6. The method for monitoring construction site environmental conditions based on satellite images and drone data according to claim 1, characterized in that: The process of obtaining an emergency control plan is as follows: Integrate the construction site environmental change characteristics, PM2.5 concentration characteristics, and construction site progress characteristics into comprehensive data on the construction site environmental status; Obtain the comprehensive data-emergency plan mapping set stored in the database, perform similarity analysis on the comprehensive data of the construction site environmental status and the comprehensive data in the comprehensive data-emergency plan mapping set, determine the comprehensive data that is most similar to the comprehensive data of the construction site environmental status, and map the corresponding emergency plan, which is recorded as the emergency control plan.

7. The method for monitoring construction site environmental conditions based on satellite images and drone data according to claim 6, characterized in that: The process of obtaining the comprehensive data-contingency plan mapping set stored in the database is as follows: Collect environmental status data of historically similar construction sites to obtain a set of environmental status data of historically similar construction sites, including historical environmental status data at different construction stages, where the historical environmental status data includes historical construction site environmental change characteristics, historical PM2.5 concentration characteristics, and historical construction site construction progress characteristics; Based on the K-means clustering algorithm, cluster analysis is performed on the historical environmental status data of different construction stages to determine multiple cluster centers, each of which corresponds to at least one historical environmental status data; Perform mean processing on the historical environmental status data corresponding to each cluster center to obtain the averaged historical environmental status data, which is recorded as comprehensive data; Obtain the historical control plan corresponding to each historical environmental status data and group them according to the cluster center; Obtain the control effect data of each historical control scheme, including the growth rate of the proportion of construction site vegetation reduction probability not exceeding the construction site vegetation reduction probability threshold , Construction site construction progress ratio growth rate and PM2.5 concentration reduction rate ; Determine the control effect coefficient Kx based on the control effect data: ； in, , and All are weight factors and dimensionless; Determine the historical control plan corresponding to the largest control effect coefficient in each group and record it as the emergency plan; The comprehensive data corresponding to each cluster center is associated with the emergency plan to obtain a comprehensive data-emergency plan mapping set and store it in the database.

Images