A method for analyzing road passability based on remote sensing geological conditions

Abstract

The present invention discloses a method for analyzing road passability based on remote sensing geological conditions, including: detecting geological hazard points in high-resolution remote sensing images based on the YOLOv3 object detection framework; classifying the basic quality of rock masses according to the hardness and tectonic structure of different types of rock masses; calculating the theoretical bearing capacity of soil masses according to the basic physical and mechanical parameters of soil masses, the sedimentation age of soil masses and influencing factors such as groundwater, and then classifying the hardness of remote sensing soil works; dividing the initial passability of different roads according to OSM road classification; superimposing the disaster influence range, the basic quality classification of rock masses, the hardness of soil works and the initial passability of roads to obtain the final road passability classification.

Description

Technical Field

[0001] The present invention belongs to the field of road traffic, and specifically relates to a method for analyzing road passability based on remote sensing geological conditions. Background Art

[0002] When conducting field surveys, it is necessary to select the team's forward route and evaluate the road passability. Timely and effective road passability evaluation can bring many conveniences to the survey work and effectively avoid potential dangers. In a complex field environment, the road environment is complex and may be affected by various geological disasters. Existing passability analysis methods mostly start from the terrain, that is, evaluate the passability through factors such as slope, without considering the impact of factors such as the rock mass quality, soil hardness, and geological disasters in the road area on the road passing ability. Summary of the Invention

[0003] In order to solve the problems existing in the prior art, the present invention provides a method for analyzing road passability based on remote sensing geological conditions, and adopts the following technical solutions:

[0004] The method for analyzing road passability based on remote sensing geological conditions includes the following steps:

[0005] S1. Based on high-resolution remote sensing images, conduct geological disaster detection to obtain the disaster impact range;

[0006] S2. Based on remote sensing data, conduct classification of the basic quality grade of rock masses;

[0007] S3. Based on the basic physical parameters of the soil, the origin of the soil, and the sedimentation age factors, conduct division of the hardness of the remote sensing soil engineering construction;

[0008] S4. First, classify the roads, and based on the classification results, assign initial passability values to the roads and construct buffer zones;

[0009] S5. Perform raster overlay summation of the disaster impact range, the basic quality classification of rock masses, the hardness of the remote sensing soil engineering construction, and the initial road passability, and after reclassification according to easy to pass, passable, and difficult to pass, obtain the final road passability result.

[0010] Furthermore, the specific implementation method of S1 is as follows;

[0011] 1.1. Overlay the vector data of geological disasters such as landslides, debris flows, and collapses with high-resolution remote sensing image data, and after data annotation, cropping, and enhancement, obtain the geological disaster detection data set;

[0012] 1.2. Use the geological disaster detection data set in 1.1 to conduct model training and testing based on the YOLOv3 object detection framework;

[0013] 1.3. Use the trained geological disaster detection model to detect geological disasters in remote sensing images to obtain geological disaster points;

[0014] 1.4. Construct a buffer zone for the disaster points and rasterize them to obtain the disaster influence range.

[0015] Furthermore, the specific implementation method of S2 is as follows;

[0016] 2.1. According to the empirical values of the compressive strength of different types of rock masses and the degree of rock mass weathering, qualitatively classify the hardness of the rock, classify the rock mass into five categories: hard rock, relatively hard rock, relatively soft rock, soft rock and extremely soft rock, and assign values to the rock mass elements accordingly, and then perform rasterization processing;

[0017] 2.2. According to the formation principle of the structural plane of the rock mass structure, use the ArcGIS line density analysis tool to classify and assign values to the density of the rock mass structure to obtain the influencing factors of the structural plane of the rock mass structure;

[0018] 2.3. The primary structural plane formed during the diagenetic stage has a great influence on the characteristics of the rock mass. According to the diagenetic characteristics of various rock masses, assign values to the integrity of the primary structure of the rock mass to obtain the influencing factors of the primary structural plane of the rock mass;

[0019] 2.4. Perform raster overlay summation on the influencing factors of the structural plane of the rock mass structure and the influencing factors of the primary structural plane of the rock mass, and after reclassification, obtain the grading of the integrity degree of the rock mass;

[0020] 2.5. Use the hardness of the rock mass obtained in 2.1 and the grading of the integrity degree of the rock mass obtained in 2.4 to perform raster overlay summation, and after reclassification, obtain the grading of the basic quality grade of the rock mass.

[0021] Furthermore, the specific implementation method of S3 is as follows;

[0022] 3.1. Regard road construction as ordinary engineering construction, obtain the average physical and mechanical indexes of various soil bodies by referring to the "Engineering Geology Handbook", and calculate the theoretical bearing capacity of the remote sensing soil according to the physical and mechanical indexes of the soil. After assignment, obtain the assignment map of the theoretical bearing capacity of the remote sensing soil engineering construction;

[0023] 3.2. The factor of the soil deposition age directly affects the unit weight, cohesion and internal friction angle of the soil by affecting the soil density and porosity. Divide all soils into three categories according to the deposition time, corresponding to the soils deposited in the middle Pleistocene and before, the soils deposited in the late Pleistocene, and the soils deposited in the Holocene, and assign land bearing capacity adjustment coefficients respectively to obtain the assignment map of the influencing factors of the soil deposition age;

[0024] 3.3. The groundwater affects the theoretical bearing capacity of the soil by influencing the unit weight of the soil mass and the internal friction angle of the soil mass. Using the Euclidean distance analysis tool in ArcGIS, the soil mass range is divided into a saturated water area and a natural water-bearing area, and different adjustment coefficients are assigned respectively to obtain the groundwater influence factor assignment map of the soil mass theoretical bearing capacity;

[0025] 3.4. Perform raster quadrature calculations on the soil mass theoretical bearing capacity, sedimentary age influence factors, and groundwater influence factors. Specifically, multiply the adjustment coefficient of the influence factors on the basis of the soil mass theoretical bearing capacity. According to the obtained soil mass theoretical bearing capacity, the soil mass is divided into pebble soil, soft soil, medium soil, and hard soil, and the remote sensing soil engineering construction hardness is obtained after assignment classification.

[0026] Furthermore, the specific implementation method of S4 is as follows;

[0027] 4.1. According to the fclass field in the OSM road data, the roads are divided into nine categories: urban arterial roads, urban sub-arterial roads, urban branch roads, elevated and expressways, suburban rural roads, internal roads, pedestrian roads, bicycle lanes, and others;

[0028] 4.2. Define and assign the initial passability of the roads according to the road categories. Among them, urban arterial roads, urban sub-arterial roads, urban branch roads, elevated and expressways are assigned "easy to pass", suburban rural roads, internal roads, and other categories are assigned "relatively easy to pass", and the remaining categories are assigned "difficult to pass", and the corresponding values are assigned at the same time;

[0029] 4.3. Considering the influence of soil mass, rock mass, and geological disasters on road traffic, construct a buffer zone for the roads, rasterize the buffer zone, and assign the initial passability value of the roads.

[0030] Furthermore, in 2.1, YOLOv3 first scales the original image to a size of 416×416, divides the original image into S×S equal-sized cells according to the scale size of the feature map, and performs detections at three scales where the feature maps are 13×13, 26×26, and 52×52. Each cell has 3 anchor boxes to predict 3 bounding boxes;

[0031] For each cell, 4 values will be predicted for each bounding box, that is, the (x, y) coordinates of the target box and the width w and height h are respectively denoted as t x , t y , t w , t h , the center of the target has an offset (c x , c y ) relative to the upper left corner of the image in the cell, and the anchor box has a height and width p w , p h, then the corrected bounding box is:

[0032] b x = σ(t x ) + c x

[0033] b y = σ(t y ) + c y

[0034]

[0035]

[0036] During the training process, the sum of squared errors is used as the loss function. Assuming the true coordinates are t w , then the gradient can be obtained by minimizing the loss function, and the gradient is the true coordinate value minus the predicted coordinate value:

[0037] Furthermore, the theoretical bearing capacity calculation formula in 3.1 is as follows;

[0038] f a = M b γb + M d γ m d + M c c k

[0039] In the formula:

[0040] f a ——The characteristic value of the foundation bearing capacity determined by the shear strength index of the soil, with the unit of kPa;

[0041] M b , M d , M c ——Bearing capacity coefficients;

[0042] b——The width of the foundation bottom surface. When it is greater than 6m, it is taken as 6m. For sandy soil, when it is less than 3m, it is taken as 3m;

[0043] c k ——The standard value of the cohesion of the soil within the depth range of one times the short side width below the foundation;

[0044] γ——The unit weight of the soil below the foundation bottom surface. The effective unit weight is taken below the groundwater level;

[0045] γ m ——The weighted average unit weight of the soil above the foundation bottom surface. The effective unit weight is taken below the groundwater level.

[0046] Furthermore, in 3.2, the sedimentary soil deposited in the middle Pleistocene and earlier is sedimented and compacted, belonging to overconsolidation, and given an adjustment coefficient of 13; the sedimentary soil deposited in the late Pleistocene belongs to normal consolidation and is given an adjustment coefficient of 10; the sedimentary soil in the Holocene is sedimented relatively loosely, belonging to underconsolidation, and given an adjustment coefficient of 8.

[0047] Furthermore, in 3.4, the classification of the hardness of the soil mass is carried out according to the theoretical value F of the basic bearing capacity of the soil mass a The theoretical value F of the basic bearing capacity of the soil mass is obtained by multiplying the theoretical basic bearing capacity of the soil mass by the reduction coefficient of the relative influencing factors a ;

[0048] F a = 0 belongs to soil with intercalated stones;

[0049] F a ≥160 kPa belongs to hard soil;

[0050] 16000 kPa ≥ F a ≥80 kPa belongs to medium soil;

[0051] F a ≤80 kPa belongs to soft soil.

[0052] After classification according to the above principle, the hardness of the soil mass for construction of the soil works is obtained.

[0053] Furthermore, in S5, a buffer zone with a radius of 500 m is established along both sides of the road for rating the trafficability of the road.

[0054] Compared with the prior art, the present invention has the following advantages and remarkable effects:

[0055] (1) Evaluating and grading the trafficability of the road from the perspective of geological disasters and geological conditions can provide favorable assistance for tasks such as field geological surveys.

[0056] (2) Detecting geological disasters in a specific area based on remote sensing technology and combining rock mass and soil data can achieve large-scale and rapid evaluation of road trafficability. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a specific flow chart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] The technical solution of the present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0059] Step 1: Based on high-resolution remote sensing images, geological disaster detection is carried out.

[0060] In this step, first, the 0.8m remote sensing images are annotated and stored as XML files in Pascal format. Then, through a format conversion script, the XML files are converted into TXT files in the <label, x, y, w, h> format. Data samples with better quality are selected from the cut images. The network structure is based on the Pytorch library in Python. During the training process, to prevent overfitting, data augmentation processing is added. A self-written program is used to complete the data augmentation process. After data augmentation processing such as image cropping, contrast, brightness, and noise, the sample data for landslide detection is finally obtained. The pictures are imported into the Labelimg software, and combined with the position information of the vector spring points, a landslide image training data set is made.

[0061] The YOLO algorithm proposed by Redmon et al. in 2016 converts the object detection task into a regression problem, greatly accelerating the detection speed. YOLOv3 is proposed based on YOLOv2, maintaining the detection speed of YOLOv2 and greatly improving the detection accuracy, especially in the detection and recognition of small targets.

[0062] YOLOv3 first scales the original picture to a size of 416×416, divides the original image into S×S equally sized cells according to the scale of the feature map, and performs detections at three scales where the feature maps are 13×13, 26×26, and 52×52. Each cell has 3 anchor boxes to predict 3 bounding boxes.

[0063] The convolutional neural network predicts 4 values for each bounding box on each cell, that is, the coordinates (x, y) of the target box and the width w and height h are respectively denoted as t x ,t y ,t w ,t h . The center of the target has an offset (c x ,c y ) relative to the upper left corner of the image in the cell, and the anchor box has a height and width p w ,p h . Then the corrected bounding box is:

[0064] b x =σ(t x )+c x

[0065] b y =σ(t y )+c y

[0066]

[0067]

[0068] During the training process, the sum of squared errors is used as the loss function. Assume the true coordinates are t w , then the gradient can be obtained by minimizing the loss function, and the gradient is the true coordinate value minus the predicted coordinate value:

[0069] YOLOv3 uses a newly designed Darknet-53 in the feature extraction stage. A large number of residual connections are used in the network to increase the depth of the network; combined with the FPN network structure, upsampling is performed on the last two feature maps of the network, aggregated with the corresponding-sized feature maps in the early stage of the network, and then the prediction results are obtained through a convolutional network. In the object detection stage, YOLOv3 first performs convolution on the 13×13 feature map for prediction to obtain the first detection result; then the 13×13 feature map is upsampled to obtain a 26×26 feature map, which is fused with the 26×26 feature map in the previous network to form a new feature map, and after multiple convolutions, it is sent to the detection layer to obtain the second detection result; then the 26×26 feature map is upsampled and fused with the previous layer to obtain a 52×52 feature map, and after multiple convolutions, it is sent to the detection layer to obtain the third detection result. The three obtained detection results are subjected to non-maximum suppression processing to obtain the final recognition result.

[0070] Construct a buffer zone for the detected disaster points and convert them into a raster to obtain the result of the influence range of the disaster points.

[0071] Step 2: Classify the basic quality grade of rock masses based on remote sensing data.

[0072] The main idea of this step is that the hardness (without considering weathering factors) and the density of primary structural planes of the same lithology in adjacent areas are basically the same; there is a certain connection between the density of tectonic structural planes and tectonic phenomena (including folds and faults). Establish the connection between the density of tectonic structural planes and tectonic phenomena, combine the density of primary structural planes to infer the volumetric joint density of the rock mass, and qualitatively classify the integrity degree of the rock mass. According to the classification structure, jointly classify the basic quality grade of the rock mass with the rock hardness classification. The main approach is to retrieve and establish the physical and mechanical property parameters of various rock masses and their basic attributes such as lithology and structure based on existing data. Combining remote sensing technology means, adopting a human-computer interaction mode, comprehensively analyzing various parameters, and using the advanced calculation and analysis tools in ArcGIS software, the basic quality classification of the rock mass is obtained through raster calculation.

[0073] First, it is necessary to classify the hardness of rock masses. Since there is no in-situ saturated uniaxial compressive strength test, the hardness of rocks is mainly based on existing empirical values. Through data collection, the empirical values of the compressive strength of different rocks are determined. Under the fixed condition of slightly weathered rock masses, all rock masses are classified into five categories according to the content of the "Standard for Engineering Rock Mass Classification" (GB / T 50218-2014) of the People's Republic of China. The hardness of rocks under other weathering conditions should be adjusted. That is, the medium weathering is downgraded by one level, the strong weathering is downgraded by two to three levels, and the completely weathered rock is directly classified as extremely soft rock. For calculation and analysis, the graphic rows of rock mass elements are rasterized in the ArcGIS software and assigned values of "1, 2, 3, 4, 6" respectively, corresponding to hard rock, relatively hard rock, relatively soft rock, soft rock, and extremely soft rock. In the above steps, the assignment of "6" is mainly for the convenience of calculating the basic quality grade, which will be described in detail in the subsequent steps.

[0074] Secondly, it is necessary to classify the integrity degree of rock masses. Rock masses are composed of rocks and structural planes. Structural planes refer to the geological interfaces such as surfaces, seams, and bands with different directions, scales, shapes, and characteristics formed in rock masses. Structural planes are mainly divided into primary structural planes, tectonic structural planes, and secondary structural planes according to their genesis. Secondary structural planes mainly include weathering structural planes and unloading structural planes, etc. In the surveyed area, the weathering structural planes in the alpine and extremely alpine geomorphic environments are the dominant factors. It is difficult to delimit the area of weathering factors through principle analysis. Similar to the rock hardness classification, the integrity degree of rock masses does not consider weathering factors, and only analyzes and calculates from tectonic structural planes and primary structural planes.

[0075] Tectonic structural planes are the fracture surfaces generated in rock masses under the action of tectonic stress, such as cleavage, joints, faults, and interlayer dislocations. It is currently impossible to distinguish these planar structures by remote sensing means. In order to classify the tectonic structural planes of rock masses, a model needs to be constructed to connect them with the regional geological structure. Regional geological structures mainly include faults and folds. They have basically the same formation principle as tectonic structural planes, both caused by tectonic stress. Generally, the greater the density of tectonic movement, the stronger the tectonic stress, and the greater the density of tectonic structural planes in the rock mass. According to this principle, the influencing factors of tectonic structural planes in rock masses are classified.

[0076] According to the tectonic interpretation map, according to the density of tectonic movement in it, the tectonic density classification is carried out by using the line density analysis tool of ArcGIS, which is divided into five categories and assigned values of "0, 1, 2, 3, 4" respectively. Among them, the value of "0" corresponds to the block with zero tectonic density, and the other density values are divided into four categories. According to the density from small to large, they correspond to "1, 2, 3, 4" respectively.

[0077] The primary structural plane is the structural plane formed during the diagenetic stage, and its characteristics are closely related to the genesis of the rock mass, such as bedding planes, schistosity, cleavage, phyllitic cleavage, etc. According to the basic diagenetic characteristics of various rock masses and in accordance with the "Code for Geotechnical Investigation" (GB50021 - 2001) of the People's Republic of China National Standard, basic assignments are made to the integrity of the primary structure of the rock mass. The specific assignment information is shown in Table 1.

[0078] Table 1 Basic Assignment Table of Rock Mass

[0079]

[0080] After assignment according to the above method, the influencing factors of the primary structural plane of the rock mass are obtained.

[0081] The integrity degree of the rock mass is divided by comprehensively considering the primary and tectonic factors of the rock mass. The raster calculator in the Spatial Analyst tool of ArcGIS is used to perform raster calculations on the primary structure density assignment map and the tectonic structure density assignment map. The specific principle is to perform a summation operation on the assignments of their corresponding graphic elements to obtain the structural density sum value X. According to the corresponding relationship in Table 2, the map is reclassified and assigned, and the integrity of the rock mass is divided into five categories: intact, relatively intact, broken, relatively broken, and extremely broken.

[0082] Table 2 Corresponding Table of Rock Mass Integrity Degree and Assignment

[0083]

[0084] Note: Those marked with * are mainly for facilitating the calculation and classification of the basic quality grade

[0085] After assignment according to the above method, the classification of the integrity degree of the rock mass is obtained.

[0086] Finally, the basic quality of the rock mass is divided. The basic quality grade of the rock mass has a good reference role for the construction of rock mass works, anti - explosion strikes, and structural self - stability. Commonly used methods for classifying the basic quality grade of rock masses include classifying according to the qualitative characteristics of the basic quality of the rock mass and classifying according to the basic quality index BQ. Considering the particularity of remote sensing work, the qualitative classification method is adopted here.

[0087] Table 3 Classification of Rock Mass Basic Quality Grade

[0088]

[0089] In accordance with the specification requirements, the basic quality grade of the rock mass is classified according to Table 3. The raster calculator in the Spatial Analyst tool of ArcGIS is used to perform raster calculations on the rock hardness map and the rock mass integrity degree map. The principle is to add the assignment of the rock mass hardness to the assignment of the rock mass integrity degree to obtain the sum value Y. According to the distribution law of the rock mass quality grade in Table 3, classification is carried out according to the following method:

[0090] Table 4 Rock mass quality grade

[0091]

[0092] Among them, since the extremely soft rock and extremely fractured rock are assigned the value of "6", the special cases in the upper right corner and lower left corner of Table 3 exactly correspond to the calculation results, and thus the basic quality results of the rock mass are obtained.

[0093] Step 3: Based on factors such as the basic physical parameters of the soil, the origin of the soil, and the sedimentation age, conduct a division of the hardness of the remotely sensed soil engineering structures.

[0094] In this step, it is first necessary to calculate the theoretical bearing capacity of the soil. In the present invention, the hardness of the soil is divided into four categories: soil with stones, hard soil, medium soil, and soft soil. Among them, the physical properties of the soil with stones are relatively special, and its basic bearing capacity depends on the contact mode of the skeleton particles, the physical properties of the filler, and their proportion. The surveyed area belongs to the high mountain and extremely high mountain landforms. The gravel soil deposited in the valleys can all be classified as soil with stones. The perennially frozen soil in the special soil is located at a relatively high altitude and mostly in steep valleys, and can also be classified as soil with stones. The bearing capacity of the silty soil in the special soil is extremely weak and can be directly classified as soft soil.

[0095] The ultimate bearing capacity of the foundation soil is the minimum foundation bottom pressure corresponding to when the foundation soil undergoes shear failure and is about to lose overall stability. The bearing capacity of the foundation is generally determined through on-site load tests or other in-situ tests and formula calculations. According to the "Code for Design of Building Foundation" (GB5007 - 2002), except for Class I buildings, in-situ tests may not be carried out and theoretical formulas can be used for calculation. Since on-site verification in the surveyed area is difficult, theoretical formulas can be used for extrapolation. Through comprehensive comparison and determination, the critical pressure when a certain range of the plastic zone is entered is used for calculation, that is, a certain plastic zone is allowed for the foundation soil to develop, and this certain plastic zone is specified as having a maximum depth not greater than one-fourth of the foundation width. The detailed theoretical bearing capacity calculation formula is as follows:

[0096] f a =M b γb + M d γ m d + M c c k

[0097] In the formula:

[0098] f a ——The characteristic value of the bearing capacity of the foundation determined by the shear strength index of the soil (kPa);

[0099] M b 、M d 、M c ——Bearing capacity coefficients, which can be determined according to Table 5;

[0100] b—the width of the foundation bottom surface (m), when it is greater than 6 m, it is taken as 6 m, and for sandy soil, when it is less than 3 m, it is taken as 3 m;

[0101] c k —the standard value of the cohesion of the soil within the depth range of one time the short side width below the foundation;

[0102] γ—the unit weight of the soil below the foundation bottom surface, and the effective unit weight is taken below the groundwater level;

[0103] γ m —the weighted average unit weight of the soil above the foundation bottom surface, and the effective unit weight is taken below the groundwater level.

[0104] Table 5 Bearing capacity coefficients Mb, Md, Mc

[0105]

[0106] It can be seen from the theoretical formula that the main determining factors of soil bearing capacity are the unit weight γ and γ m , the internal friction angle φk of the soil and the soil cohesion c k . In the case where geotechnical tests cannot be carried out and only soil classification information is available, the average physical and mechanical indexes of various soils can be known by referring to the "Engineering Geology Handbook", and the theoretical bearing capacity of the soil can be calculated according to the formula.

[0107] Secondly, it is necessary to calculate the influencing factors of soil deposition age. The deposition time affects the soil density and porosity, and directly affects the unit weight, cohesion and internal friction angle of the soil. All soils are divided into three categories according to the deposition time, and are assigned values of "0, 1, 2" from early to late, corresponding to the soils deposited in the Middle Pleistocene and earlier, the soils deposited in the Late Pleistocene, and the soils deposited in the Holocene respectively. The soils deposited in the Middle Pleistocene and earlier are densely deposited, belonging to overconsolidated, and are assigned an adjustment coefficient of 13; the soils deposited in the Late Pleistocene belong to normal consolidation and are assigned an adjustment coefficient of 10; the soils deposited in the Holocene are loosely deposited, belonging to underconsolidated, and are assigned an adjustment coefficient of 8, so as to obtain the influencing factors of deposition age.

[0108] Then calculate the influencing factors of soil hydrology. The influence of groundwater on soil bearing capacity mainly includes the unit weight of the soil and the internal friction angle of the soil. First, the effective unit weight is taken below the groundwater level for the unit weight of the soil, that is, in the saturated state, the unit weight of the soil needs to be subtracted by 10; secondly, according to the "Engineering Geology Handbook", it can be known that the higher the water content of the soil, the smaller the internal friction angle of the soil.

[0109] To distinguish the influence range of groundwater on the bearing capacity of soil, based on the river network in the survey area, the Euclidean distance analysis tool in the Spatial Analyst tool of ArcGIS is used to divide the soil area into two parts: the saturated area and the natural water-bearing area. A "7" adjustment coefficient is assigned to the saturated area, and a "10" adjustment coefficient is assigned to the natural water-bearing area to obtain the groundwater influence factor.

[0110] Finally, the raster calculator in the Spatial Analyst tool of ArcGIS is used to perform raster quadrature calculations on the theoretical bearing capacity assignment map of soil works, the weight map of sedimentation age influence factors, and the weight map of groundwater influence factors. The specific principle is to multiply the basic theoretical bearing capacity of the soil by the reduction coefficient of the relative influence factor to obtain the theoretical value F of the basic bearing capacity of the soil. a Based on the fact that the Spatial Analyst tool of ArcGIS cannot use decimals, all reduction coefficients are increased by 10 times, and after the calculation result is obtained, it is divided by 100 to obtain the theoretical value of the basic bearing capacity of the soil.

[0111] According to the experience of works construction, the soil hardness is classified according to the theoretical value of the basic bearing capacity of the soil:

[0112] F a = 0 belongs to soil with intercalated stones ("0" value is only for convenient calculation and does not represent the specific actual bearing capacity);

[0113] F a ≥160 kPa belongs to hard soil;

[0114] 16000 kPa ≥ F a ≥ 80 kPa belongs to medium soil;

[0115] F a ≤ 80 kPa belongs to soft soil.

[0116] After classification according to the above principle, the hardness of soil works construction is obtained.

[0117] Step 4: Initial trafficability assignment and buffer construction of the road.

[0118] This step first requires classifying roads. The OSM road data is divided into 27 categories according to the fclass field, which can be summarized into 9 categories: urban arterial roads, urban secondary arterial roads, urban branch roads, elevated and expressways, suburban rural roads, internal roads, pedestrian roads, bicycle lanes, and others. In terms of initial passability, considering the highway construction situation, the initial passability of paved roads such as urban roads, elevated and expressways is set to easy to pass. Among the remaining road categories, considering that the geotechnical properties of unpaved roads have a greater impact on their passability, their initial passability is not set, and the final road passability level is obtained by assigning values based on the geotechnical properties of the roads later. The road width and initial passability assignment are shown in the following table.

[0119] Table 6 Road Width and Initial Passability Assignment Table

[0120]

[0121]

[0122] According to the road width and initial passability assignment table, use the field calculator function in the ArcGIS software to assign values to the road width and initial passability of the road data. Considering the impact of the rock mass, soil mass, and disaster occurrence near the road on the road passability, a buffer zone with a radius of 500m is established along both sides of the road for rating the road passability.

[0123] Step 5: Road Passability Analysis.

[0124] In this step, the disaster influence range, the basic quality classification of the rock mass, the hardness of the soil construction, and the initial passability of the road are subjected to raster overlay summation, and after reclassification according to easy to pass, relatively easy to pass, and difficult to pass, the final road passability result is obtained.

[0125] The specific embodiments described in this article are only examples to illustrate the spirit of the present invention. Those skilled in the technical field to which the present invention belongs can make various modifications or supplements to the described specific embodiments or use similar methods to replace them, but will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A method for analyzing road passability based on remote sensing geological conditions, characterized by including the following steps: S1. Based on high-resolution remote sensing images, conduct geological disaster detection to obtain the disaster influence range; The specific implementation method of S1 is as follows; 1.

1. Overlay the vector data of geological disasters such as landslides, debris flows, and collapses with high-resolution remote sensing image data, and after data annotation, cropping, and enhancement, obtain a geological disaster detection data set; 1.

2. Use the geological disaster detection data set in 1.1 to conduct model training and testing based on the YOLOv3 object detection framework; 1.

3. Use the trained geological disaster detection model to detect geological disasters in remote sensing images to obtain geological disaster points; 1.

4. Construct a buffer zone for the disaster points and rasterize them to obtain the disaster influence range; S2. Classify the basic quality grades of rock masses based on remote sensing data; The specific implementation method of S2 is as follows; 2.

1. According to the empirical values of the compressive strength of different types of rock masses and the degree of rock mass weathering, qualitatively divide the rock hardness, divide the rock masses into five categories: hard rock, relatively hard rock, relatively soft rock, soft rock and extremely soft rock, and assign corresponding values to the rock mass elements, and perform rasterization processing; 2.

2. According to the formation principle of the structural planes of rock masses, use the ArcGIS line density analysis tool to classify and assign values to the rock mass structure density to obtain the influencing factors of the structural planes of rock masses; 2.

3. The primary structural planes formed during the diagenetic stage have a great influence on the characteristics of rock masses. According to the diagenetic characteristics of various rock masses, assign values to the integrity of the primary structure of rock masses to obtain the influencing factors of the primary structural planes of rock masses; 2.

4. Perform raster overlay summation on the influencing factors of the structural planes of rock masses and the influencing factors of the primary structural planes, and reclassify to obtain the classification of the integrity degree of rock masses; 2.

5. Use the rock hardness obtained in 2.1 and the classification of the integrity degree of rock masses obtained in 2.4 to perform raster overlay summation, and reclassify to obtain the classification of the basic quality grades of rock masses; S3. Divide the hardness of remote sensing soil works based on the basic physical parameters of soil, soil genesis, and sedimentation age factors; S4. First classify the roads, and assign initial trafficability values and construct buffer zones for the roads according to the classification results; The specific implementation method of S4 is as follows; 4.

1. According to the fclass field in the OSM road data, divide the roads into nine categories: urban arterial roads, urban secondary arterial roads, urban branch roads, elevated and expressways, suburban rural roads, internal roads, pedestrian roads, bicycle lanes and others; 4.

2. Define and assign initial trafficability for the roads according to the road categories. Among them, urban arterial roads, urban secondary arterial roads, urban branch roads, elevated and expressways are assigned "easy to pass", suburban rural roads, internal roads and other categories are assigned "relatively easy to pass", and the remaining categories are assigned "difficult to pass", and corresponding values are assigned at the same time; 4.

3. Considering the influence of soil, rock mass and geological disasters on road trafficability, construct a buffer zone for the roads, rasterize the buffer zone, and assign the initial trafficability value to the roads; S5. Perform raster overlay summation on the disaster influence range, the basic quality classification of rock masses, the hardness of soil works construction, and the initial trafficability of roads, and reclassify according to easy to pass, passable, and difficult to pass to obtain the final road trafficability result.

2. The method for analyzing road passability based on remote sensing geological conditions according to claim 1, characterized in that: The specific implementation method of S3 is as follows; 3.

1. Regard road construction as ordinary works construction, obtain the average physical and mechanical indexes of various soils by referring to the "Engineering Geology Handbook", calculate the theoretical bearing capacity of remote sensing soil according to the physical and mechanical indexes of soil, and obtain the theoretical bearing capacity assignment map of remote sensing soil works construction after assignment; 3.

2. The factor of soil deposition age directly affects the unit weight, cohesion and internal friction angle of soil by influencing soil density and porosity. All soils are divided into three categories according to the deposition time, corresponding to the soils deposited in the Middle Pleistocene and before, the soils deposited in the Late Pleistocene, and the soils deposited in the Holocene, respectively. Adjustment coefficients for land bearing capacity are assigned to them to obtain the assignment map of the influencing factors of soil deposition age. 3.

3. Groundwater affects the theoretical bearing capacity of soil by influencing the unit weight of soil and the internal friction angle of soil. Using the Euclidean distance analysis tool in ArcGIS, the soil area is divided into a saturated zone and a natural water-bearing zone, and different adjustment coefficients are assigned to them to obtain the assignment map of the influencing factors of groundwater on the theoretical bearing capacity of soil. 3.

4. Grid integration calculations are performed on the theoretical bearing capacity of soil, the influencing factors of deposition age and the influencing factors of groundwater. Specifically, on the basis of the theoretical bearing capacity of soil, the adjustment coefficients of the influencing factors are multiplied. According to the obtained theoretical bearing capacity of soil, the soil is divided into rock-containing soil, soft soil, medium soil and hard soil, and the remote sensing soil construction hardness is obtained after assignment and classification.

3. The method for analyzing road passability based on remote sensing geological conditions according to claim 1, characterized in that: In 2.1, YOLOv3 first scales the original image to a size of 416×416, divides the original image into S×S equal-sized cells according to the scale of the feature map, and performs detections at three scales where the feature maps are 13×13, 26×26, and 52×52. Each cell has 3 anchor boxes to predict 3 bounding boxes. On each cell, four values are predicted for each bounding box. The (x, y) coordinates of the target box, width w, and height h are denoted as t x , t y , t w , t h . The center of the target has an offset (c x , c y ) relative to the upper left corner of the image in the cell. The anchor box has height and width p w , p h . Then the corrected bounding box is: b x = σ(t x ) + c x b y = σ(t y ) + c y During the training process, the sum of squared errors is used as the loss function. Assuming the true coordinate is t w , then the gradient is obtained by minimizing the loss function, and the gradient is the true coordinate value minus the predicted coordinate value:

4. The method for analyzing road passability based on remote sensing geological conditions according to claim 2, wherein: In 3.1, the formula for the theoretical bearing capacity is as follows; f a = M b γb + M d γ m d + M c c k In the formula: f a ——Characteristic value of subgrade bearing capacity determined by shear strength indexes of soil, with the unit of kPa; M b 、M d 、M c —— bearing capacity factor; b——The width of the foundation bottom surface, when it is greater than 6m, it is taken as 6m, and for sandy soil, when it is less than 3m, it is taken as 3m; c k —— Cohesion standard value of soil within the depth range of one-time short side width below the foundation γ——The unit weight of soil below the foundation bottom surface, and the effective unit weight is taken below the groundwater level; γ m ——The weighted average unit weight of the soil above the foundation base, and the effective unit weight is taken below the groundwater level.

5. The method for analyzing road passability based on remote sensing geological conditions according to claim 2, wherein: In 3.2, the soils deposited in the Middle Pleistocene and before are densely deposited and belong to overconsolidated, and an adjustment coefficient of 13 is assigned; the soils deposited in the Late Pleistocene belong to normal consolidation, and an adjustment coefficient of 10 is assigned; the soils deposited in the Holocene are relatively loosely deposited and belong to underconsolidated, and an adjustment coefficient of 8 is assigned.

6. The method for analyzing road passability based on remote sensing geological conditions according to claim 2, wherein: 3.4 Classify the soil hardness according to the theoretical value F of the basic bearing capacity of the soil mass a Multiply the theoretical value of the basic bearing capacity of the soil mass by the reduction coefficient of the relative influencing factors to obtain the theoretical value F of the basic bearing capacity of the soil mass a ; F a = 0 belongs to the intercalated stone soil; F a ≥160 kPa belongs to hard soil; 16000 kPa ≥ F a ≥ 80 kPa belongs to medium soil; F a ≤80 kPa belongs to soft soil; After classification according to the above principle, the hardness of soil construction is obtained.

7. The method for analyzing road passability based on remote sensing geological conditions according to claim 1, wherein: In S5, a buffer zone with a radius of 500m is established along both sides of the road for rating the road trafficability.

Images