Mountain area railway drainage facility site selection method based on hydrological space site selection analysis model

Abstract

The invention provides a mountainous area railway drainage facility site selection method based on a hydrological space site selection analysis model, and relates to the technical field of runoff pollution control, and the method comprises the steps: carrying out the preprocessing of collected data, and obtaining the environment data of a mountainous area railway; a terrain factor analysis layer, a runoff calculation layer, a dumping point screening layer, a spatial clustering analysis layer, a geometric recognition classification layer and a drainage facility recommendation layer are utilized to construct a three-dimensional hydrological space site selection analysis model; and carrying out spatial clustering dimension reduction analysis and surface runoff calculation on the mountainous area railway environment data by utilizing the three-dimensional hydrological space site selection analysis model to obtain a site selection result of the mountainous area railway drainage facility so as to finish site selection of the mountainous area railway drainage facility. The method solves the problem that accurate site selection of drainage facilities of mountainous railways is difficult.

Description

Technical Field

[0001] This specification relates to the technical field of runoff pollution control, and in particular to a site selection method for railway drainage facilities in mountainous areas based on a hydrological spatial site selection analysis model. Background Art

[0002] Mountainous areas are characterized by complex terrain and variable climates. Drainage areas are generally steep and runoff is rapid. Heavy rains often lead to the rapid concentration of large amounts of rainwater in narrow mountain catchments, preventing the timely drainage of accumulated water. This in turn causes water damage such as roadbed erosion and landslides. Current mountain railway drainage design suffers from a lack of adequate hydrological analysis, improper drainage system planning, and inadequate facilities. These issues directly lead to poor drainage efficiency and frequent water damage. The mechanisms of water damage on mountain railways are highly complex, and their spatiotemporal distribution is influenced by multiple factors, including topography, geology, hydrology, and meteorology. Traditional methods for selecting drainage measures rely primarily on manual topographic map interpretation and on-site surveys. Mountainous terrain is often complex and varied, and manual surveys are often used to determine facility locations. However, these methods struggle to fully and accurately capture the complex terrain and hidden water paths, resulting in low efficiency and limited accuracy. This makes them difficult to apply in large-scale projects and hinders storage and interpretation. Summary of the Invention

[0003] In view of the above-mentioned deficiencies in the prior art, the site selection method for railway drainage facilities in mountainous areas based on the hydrological spatial site selection analysis model provided by the present invention solves the problem of difficulty in accurately selecting the site for railway drainage facilities in mountainous areas.

[0004] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a method for site selection of drainage facilities for railways in mountainous areas based on a hydrological spatial site selection analysis model, comprising: S1: Preprocess the collected data to obtain mountain railway environmental data; S2: Construct a three-dimensional hydrological spatial site selection analysis model using the terrain factor analysis layer, runoff calculation layer, pour point screening layer, spatial cluster analysis layer, geometric identification classification layer, and drainage facility recommendation layer; S3: Using the three-dimensional hydrological spatial site selection analysis model, perform spatial clustering dimensionality reduction analysis and surface runoff calculation on the mountain railway environmental data to obtain the site selection results of the mountain railway drainage facilities, and complete the site selection of the mountain railway drainage facilities.

[0005] The beneficial effects of the present invention are as follows: the present invention provides a method for site selection of drainage facilities for mountain railways based on a hydrological spatial site selection analysis model, which obtains mountain railway environmental data by preprocessing the collected data, and uses the three-dimensional hydrological spatial site selection analysis model to perform spatial clustering dimensionality reduction analysis and surface runoff calculation on the mountain railway environmental data to obtain the site selection results of drainage facilities for mountain railways, thereby improving the accuracy of site selection of drainage facilities for mountain railways and their applicability under complex terrain conditions; reducing the probability of water accumulation in railway roadbeds, and significantly enhancing the railway's ability to resist water damage.

[0006] Furthermore, the S1 includes: The digital elevation model data, land use type data, soil hydrological condition information, and meteorological rainfall data are converted to the same coordinate system and spatial resolution, and boundary gaps and overlapping areas are repaired through topological consistency checking to obtain the processed digital elevation model data, land use type data, soil hydrological condition information, and meteorological rainfall data; The local window minimum filtering algorithm was used to filter the areas blocked by buildings under the forest in the digital elevation model data, and the exposed slope areas were smoothed and optimized. The elevation difference threshold method was used to eliminate abnormal elevation disturbances to obtain optimized digital elevation model data. The spatial distribution of soil use type data was checked and joint classification was performed using soil texture information to obtain joint classification data; Based on meteorological rainfall data, an image is constructed using the inverse distance weighted method to obtain a regional rainfall spatial distribution map; wherein the digital elevation model data, land use type data, soil hydrological condition information and meteorological rainfall data belong to the collected data, and the optimized digital elevation model data, the joint classification data and the regional rainfall spatial distribution map belong to the mountain railway environment data.

[0007] By unifying coordinates, spatially correcting, filtering, and jointly classifying multi-source heterogeneous data, we achieved the construction of high-precision mountain railway environmental data, providing a unified and accurate data foundation for subsequent hydrological analysis. In particular, the optimized DEM and jointly classified data improved the accuracy of identifying microtopography and land types, significantly enhancing the reliability and adaptability of model inputs.

[0008] Furthermore, the S3 includes: S310: Using the terrain factor analysis layer, extracting terrain features from the optimized digital elevation model data to obtain a terrain basic factor map; S320: Using the runoff calculation layer, based on the terrain basic factor map, the joint classification data, the soil hydrological condition information, and the regional rainfall spatial distribution map, a slope correction is introduced to perform calculations to obtain the water catchment unit, runoff data, and initial pour point information; S330: Using the pour point screening layer, performing flood safety buffer analysis and feature elimination on the initial pour point information and railway line vector data to obtain a set of threatening pour points; S340: using a spatial clustering analysis layer, clustering and adaptively optimizing the threatening pouring point set to obtain cluster spatial characteristics; S350: using the geometric recognition classification layer, performing principal component analysis on the cluster spatial features, and obtaining cluster morphology classification results by setting a variance explanation rate threshold; S360: Using the drainage facility recommendation layer, based on the cluster morphology classification results, the runoff data and the roadbed data, the standard parameters are reversed through the Manning equation to obtain the site selection results of the mountain railway drainage facilities, and complete the site selection of the mountain railway drainage facilities; among them, the terrain factor analysis layer, the runoff calculation layer, the pour point screening layer, the spatial clustering analysis layer, the geometric identification classification layer and the drainage facility recommendation layer belong to the three-dimensional hydrological space site selection analysis model.

[0009] The introduction of a multi-layered 3D hydrological spatial analysis model automates the entire process, from terrain extraction and runoff simulation to facility recommendation. Multi-scale analysis methods, such as slope correction, spatial clustering, and principal component identification, effectively enhance the scientific nature and applicability of drainage facility layout in complex mountainous areas.

[0010] Furthermore, the expression of the runoff data is: ; ; ; ; in, represents runoff data, represents the total rainfall, Indicates the dynamic initial loss coefficient, that is, at position The initial rainfall loss ratio at Represents the horizontal coordinate of the grid cell, Represents the vertical coordinate of the grid cell, represents the potential water storage capacity, represents the modified general condition, Indicates wet conditions, Indicates drought conditions, represents a natural constant, represents the local slope, represents the normalized difference vegetation index, It is used to perform interval restriction processing on the dynamic initial loss ratio expression to ensure the rationality of its physical meaning and value range.

[0011] By constructing an improved SCS-CN runoff model that includes slope and vegetation factors, dynamic and accurate estimation of runoff under different landforms and surface conditions is achieved, effectively overcoming the limitations of traditional fixed CN value models in mountainous applications and providing more responsive flow input for facility design.

[0012] Furthermore, the expression of the actual buffer distance of the threatening pouring point set is: ; in, Indicates the actual buffer distance, represents the railway flood prevention safety distance threshold, represents the maximum slope in the study area, Indicates the local slope.

[0013] The terrain slope factor is introduced into the pour point buffer calculation and a dynamic buffer distance model is constructed, so that the selected threatening pour points are more consistent with the water flow risk distribution under real terrain conditions, thereby improving the physical rationality of the screening results and the safety boundary guarantee capability of engineering applications.

[0014] Furthermore, the S340 includes: Using the spatial clustering analysis layer, the distance between each point and its minimum number of neighboring points is calculated, and the K-distance curve is formed after arranging them in ascending order. The mutation gradient at the inflection point is selected as the neighborhood radius threshold for adaptive optimization. Clustering the threatening pouring point set using the adaptively optimized neighborhood radius threshold to obtain a clustering result; The clustering results are screened twice to eliminate invalid clusters located in tunnel and bridge structure areas, and the spatial characteristics of clusters are obtained.

[0015] The K-distance curve mutation point is used as the adaptive clustering radius threshold to automatically determine the optimal clustering parameters. Combined with the structural shielding screening mechanism of the engineering section, the rationality of the spatial distribution of the clustering results is effectively improved, ensuring that the facility layout focuses on the real risk catchment area and reducing the misidentification rate.

[0016] Furthermore, the S350 includes: Using the geometric recognition classification layer, principal component analysis is performed on the cluster spatial characteristics to obtain the variance contribution rate of each principal component; Based on the variance explanation rate threshold, the variance contribution rate of the principal components is analyzed to obtain linear clusters and nonlinear clusters; The least squares method is used to fit the points in the linear cluster to extract the main axis direction angle, and the layout position and direction of the linear drainage ditch are determined by combining the railway direction and the drainage slope layout requirements. For nonlinear clusters, the lowest elevation point is selected as the recommended facility location by traversing the point set within the cluster, and the preferentially matched centralized drainage facilities are obtained; wherein, the layout position and direction of the linear drainage ditch and the centralized drainage facilities belong to the cluster morphology classification results.

[0017] By performing principal component analysis (PCA) on clusters to distinguish linear and nonlinear morphologies, and then matching different types of drainage facilities, a "cluster geometry identification-facility matching" mechanism was established, achieving an automated linkage from morphological identification to facility recommendation. This strategy can accurately guide the layout and form of drainage ditches and culverts, improving the scientific nature and operability of engineering design. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein: Figure 1 is an exemplary flow chart of a method for site selection of drainage facilities for mountain railways based on a hydrological spatial site selection analysis model according to some embodiments of this specification; Figure 2 is an exemplary schematic diagram of a digital elevation model of a study area according to some embodiments of this specification; FIG3( a ) is an exemplary schematic diagram of surface runoff extraction results according to some embodiments of this specification; FIG3( b ) is an exemplary schematic diagram of pour point extraction results according to some embodiments of the present specification; Figure 4 is an exemplary schematic diagram of the water catchment unit division results of the study area according to some embodiments of this specification; Figure 5 is an exemplary schematic diagram of pour point range screening according to some embodiments of this specification; Figure 6 is an exemplary schematic diagram of a K-distance curve according to some embodiments of this specification; FIG7( a ) is an exemplary schematic diagram of a DBSACN clustering principle according to some embodiments of this specification; FIG7( b ) is an exemplary schematic diagram of the DBSACN clustering principle according to some embodiments of this specification; Figure 8 is an exemplary schematic diagram of the principle of principal component analysis according to some embodiments of this specification; Figure 9 is an exemplary schematic diagram of railway line elevation according to some embodiments of this specification; Figure 10is an exemplary schematic diagram of pour point screening results according to some embodiments of this specification; Figure 11 is an exemplary schematic diagram of pouring point clustering K-Distance according to some embodiments of this specification; Figure 12 is an exemplary schematic diagram of pouring point clustering-PCA result statistics according to some embodiments of this specification; Figure 13 is an exemplary schematic diagram of the pouring point clustering-PCA results shown in some embodiments of this specification; Figure 14 is an exemplary schematic diagram of drainage facility site selection prediction results according to some embodiments of this specification; Figure 15 This is an exemplary schematic diagram comparing the predicted results and the actual drainage facility locations according to some embodiments of this specification. DETAILED DESCRIPTION

[0019] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0020] Example Figure 1 This is an exemplary flow chart of a method for selecting a site for drainage facilities on a mountain railway based on a hydrological spatial site selection analysis model according to some embodiments of this specification. Figure 1 As shown, the process includes the following steps. In some embodiments, the process can be executed by a processor.

[0021] S1: Preprocess the collected data to obtain mountain railway environment data.

[0022] The collected data refers to the environmental data collected in the study area. For example, the collected data may include digital elevation model data, land use type data, soil and hydrological condition information, railway line vector data, roadbed data, and meteorological rainfall data. The specific collected data are shown in Table 1.

[0023] Digital elevation model data is the digital elevation model (DEM) related data of the study area.

[0024] In some embodiments, the processor may utilize airborne laser (LiDAR) to collect data on the study area environment and obtain digital elevation model data.

[0025] Land use type data is data on the classification of land use conditions.

[0026] In some embodiments, the processor may utilize remote sensing images and multi-period land use remote sensing to obtain a monitoring dataset as land use type data.

[0027] Soil hydrological condition information is data related to the soil quality and hydrological quality of the study area.

[0028] In some embodiments, the processor may collect digital results of the soil survey as soil hydrological condition information.

[0029] Railway line position vector data is the spatial position of the railway line and its geometric attribute data expressed in vector form.

[0030] In some embodiments, the processor may collect data provided by the engineering design department to obtain railway line position vector data.

[0031] Roadbed data is data that characterizes the spatial position and morphological characteristics of the foundation structure along the railway line, including the plane boundary, elevation information, slope gradient, width, fill and cut properties (such as fill area, cut area), reinforcement type and its structural characteristics of the roadbed.

[0032] Meteorological rainfall data are data related to the climate and rainfall conditions in the study area.

[0033] In some embodiments, the processor can collect recorded information from meteorological stations in the study area and surrounding areas to obtain meteorological rainfall data.

[0034]

[0035] Mountain railway environmental data is pre-processed collected data. For example, mountain railway environmental data may include optimized digital elevation model data, joint classification data, regional rainfall spatial distribution maps, soil hydrological condition information, railway line vector data, and roadbed data.

[0036] In some embodiments, the processor can convert digital elevation model data, land use type data, soil hydrological condition information and meteorological rainfall data into the same coordinate system and spatial resolution, repair boundary gaps and overlapping areas through topological consistency check, and obtain processed digital elevation model data, land use type data, soil hydrological condition information and meteorological rainfall data; use the local window minimum filtering algorithm to filter the forest building occlusion area in the digital elevation model data, perform smoothing optimization on the exposed slope area, and use the elevation difference threshold method to eliminate abnormal elevation disturbances to obtain optimized digital elevation model data; perform spatial distribution verification on soil use type data, use soil texture information for joint classification, and obtain joint classification data; based on meteorological rainfall data, use the inverse distance weighted method to construct an image to obtain a regional rainfall spatial distribution map.

[0037] In some embodiments, the processor can uniformly convert DEM, land use, soil hydrological properties and meteorological rainfall data into the same coordinate system (CGCS2000) and spatial resolution (0.2 m), and repair boundary gaps and overlapping areas through topological consistency checks to ensure that there are no dislocations and gaps between model input layers, thereby achieving high-precision coupling of multi-source data.

[0038] The optimized digital elevation model data is the digital elevation model data after filtering and smoothing.

[0039] In some embodiments, the processor can perform slope filtering and anomaly removal processing on the digital elevation model data. For the abnormal points that may be generated by the airborne LiDAR point cloud under the forest and in the areas blocked by buildings, a local window minimum filtering algorithm is introduced to smooth and optimize the exposed slope areas, and the elevation difference threshold method is used to remove abnormal elevation disturbances, thereby improving the DEM's restoration accuracy of the micro-topography structure and obtaining optimized digital elevation model data.

[0040] Joint classification data is the joint classification data of soil use type data and soil texture information.

[0041] In some embodiments, the processor can verify the spatial distribution of land use type data such as forest land, bare land, and cultivated land based on the remote sensing interpretation results and combined with the field survey sample data of the study area, and introduce soil texture information from the soil census database to jointly classify it with the land use type to improve the accuracy of the CN value assignment of the subsequent SCS-CN model.

[0042] The regional rainfall spatial distribution map is the image data of the rainfall distribution in each sub-region of the target area.

[0043] In some embodiments, the processor can introduce daily rainfall data from multiple meteorological stations in and around the study area based on meteorological rainfall data, and use the inverse distance weighted (IDW) method to construct a regional rainfall spatial distribution map, thereby realizing the transition from point observation to surface data and providing more continuous and detailed rainfall input for surface runoff simulation.

[0044] In some embodiments, high-precision DEM data with a resolution of 0.2 m generated by point clouds acquired by airborne laser LiDAR is used, such as Figure 2 As shown, the DEM data covers a projected area of ​​approximately 6.1 km 2 , which is a small watershed.

[0045] Based on high-precision DEM and machine learning algorithms, combined with DBSCAN density clustering and PCA principal component analysis, the locations of 14 culverts and 2 drainage ditches along the mountain railway were successfully predicted, with a 92% consistency between the culvert locations and the actual project layout, verifying the applicability of this method in complex terrain conditions.

[0046] S2: A three-dimensional hydrological spatial site selection analysis model is constructed using the terrain factor analysis layer, runoff calculation layer, pour point screening layer, spatial clustering analysis layer, geometric identification classification layer, and drainage facility recommendation layer.

[0047] The three-dimensional hydrological spatial site selection analysis model is. For example, the three-dimensional hydrological spatial site selection analysis model may include a terrain factor analysis layer, a runoff calculation layer, a pour point screening layer, a spatial clustering analysis layer, a geometric identification and classification layer, and a drainage facility recommendation layer; wherein the output of the terrain factor analysis layer serves as the input of the runoff calculation layer, the output of the runoff calculation layer serves as the input of the pour point screening layer and the drainage facility recommendation layer, the output of the pour point screening layer serves as the input of the spatial clustering analysis layer, the output of the spatial clustering analysis layer serves as the input of the geometric identification and classification layer, the output of the geometric identification and classification layer serves as the input of the drainage facility recommendation layer, and the output of the drainage facility recommendation layer serves as the final output of the three-dimensional hydrological spatial site selection analysis model.

[0048] The terrain factor analysis layer is used to extract terrain features from the optimized digital elevation model data and obtain the terrain basic factor map.

[0049] The runoff calculation layer is used to calculate based on the terrain basic factor map, joint classification data, soil hydrological condition information and regional rainfall spatial distribution map, introduce slope correction, and obtain water collection units, runoff data and initial pouring point information.

[0050] In some embodiments, the processor can implement the extraction of basic hydrological features of high-precision DEM data in ArcGIS software, which mainly includes eight steps: filling analysis, flow direction extraction, flow statistics, flow division, river network classification, grid river network vectorization, pour point drawing and catchment area division. This process is based on the D8 flow direction algorithm. By analyzing the elevation relationship between the elevation value of each grid cell and the elevation of the eight surrounding neighboring grids, the flow direction of the water flow is determined and the convergence capacity is calculated. The surface runoff distribution is further extracted in combination with flow statistics. After the surface runoff is obtained, the end point of the runoff, that is, the pour point, can be extracted. These points are often the convergence area of ​​water flow and the main direction of precipitation flow to low-lying areas or drainage systems. They can reflect the relationship between water flow and parameters such as slope and surface type. By determining one or more pour points (the outlet location of the basin), the boundary of the entire basin and its sub-basins can be automatically divided, and the catchment units in the study area can be obtained.

[0051] In some embodiments, the processor can use the ArcGIS hydrological analysis process to comprehensively consider the influence of terrain factors such as slope, elevation, ridge and valley lines on the direction and convergence path of water flow, and extract surface runoff and pouring point information as well as the division of catchment areas. As shown in Figure 3 (a), the surface runoff in the study area presents a network distribution along both sides of the middle black railway line, converging from higher elevation areas to low-lying areas; the runoff distribution is mostly parallel or perpendicular to the railway line, indicating that the surface slope mainly changes in the direction perpendicular to the railway line; the runoff network on the north side is dense and has more branches, while the runoff network on the south side is relatively sparse and has a small number of branches, which is consistent with the overall terrain of the study area, which is low in the northeast and high in the southwest. The pouring point extraction results are shown in Figure 3 (b), which are usually located at the end of the tributary of the runoff or in the area where it intersects with the mainstream. In the area with dense tributaries, the distribution density of pouring points is higher, indicating that these areas are concentrated points of water flow or low-lying terrain. The catchment units are divided according to the surface runoff and pouring point extraction results. A total of 23 catchment units are divided in the study area, such as Figure 4 As shown, the maximum area is 0.749 km 2 , with a minimum area of ​​0.046 km 2 , with an average area of ​​0.161 km 2 , the maximum slope is 39°, the minimum slope is 12°, and the average slope is 27°.

[0052] The pour point screening layer is used to perform flood safety buffer analysis and feature elimination on the initial pour point information and railway line position vector data to obtain a set of threatening pour points.

[0053] The spatial clustering analysis layer is used to perform clustering and adaptive optimization processing on the threatening pouring point set to obtain cluster spatial characteristics.

[0054] The geometric recognition classification layer is used to perform principal component analysis on the spatial characteristics of the cluster clusters and obtain cluster morphology classification results by setting a variance explanation rate threshold.

[0055] The drainage facility recommendation layer is used to perform reverse deduction of specification parameters through the Manning equation based on the cluster morphology classification results, the runoff data and the roadbed data, to obtain the site selection results of the mountain railway drainage facilities, and complete the site selection of the mountain railway drainage facilities.

[0056] S3: Using the three-dimensional hydrological spatial site selection analysis model, perform spatial clustering dimensionality reduction analysis and surface runoff calculation on the mountain railway environmental data to obtain the site selection results of the mountain railway drainage facilities, and complete the site selection of the mountain railway drainage facilities.

[0057] The site selection results of mountain railway drainage facilities are the results of selecting the location for setting up mountain railway drainage facilities.

[0058] In some embodiments, the processor may implement S3 based on the following steps.

[0059] S310: Using the terrain factor analysis layer, extract terrain features from the optimized digital elevation model data to obtain a terrain basic factor map.

[0060] The terrain basic factor map is a spatial distribution layer that reflects the basic terrain characteristics of the study area, extracted based on the optimized digital elevation model (DEM) data. It mainly includes slope map, aspect map, flow direction map, ridge valley line map, etc., which are used to depict surface water flow paths, confluence trends and landform structure characteristics.

[0061] In some embodiments, the processor can extract terrain features such as slope, flow direction, and ridges from the optimized digital elevation model data to generate a terrain basic factor map. For example, based on high-precision DEM data, the D8 (Deterministic Eight-Neighbor) algorithm analyzes the elevation difference between each pixel and its eight surrounding pixels, calculating the primary flow direction for each grid cell. Based on this, the upstream flow received by each pixel is further counted to generate a flow accumulation map, and surface runoff paths are extracted based on a set threshold.

[0062] S320: Utilizing the runoff calculation layer, based on the terrain basic factor map, joint classification data, soil hydrological condition information, and regional rainfall spatial distribution map, slope correction is introduced for calculation to obtain water catchment units, runoff data, and initial pour point information.

[0063] A watershed is an area unit that contains the point where water flows converge.

[0064] Runoff data is data that reflects the magnitude of water runoff.

[0065] In some embodiments, surface runoff estimation is based on a three-dimensional hydrological spatial site selection analysis model (an improved adaptive enhanced SCS-CN hydrological model). This model, building on the traditional SCS-CN framework, incorporates multiple optimization strategies, including slope correction, soil texture adjustment, and a dynamic initial loss mechanism, to improve its accuracy and adaptability to complex mountainous terrain. This model comprehensively considers multiple spatial factors, including soil permeability, land cover type, slope factor, and rainfall intensity. Integrating regional geological survey results and recommended parameter tables, it constructs a multi-factor-driven runoff response calculation method.

[0066] In some embodiments, the expression for runoff data may be: ; ; ; ; in, represents runoff data, represents the total rainfall, Indicates the dynamic initial loss coefficient, that is, at position The initial rainfall loss ratio at Represents the horizontal coordinate of the grid cell, Represents the vertical coordinate of the grid cell, represents the potential water storage capacity, represents the modified general condition, Indicates wet conditions, Indicates drought conditions, represents a natural constant, represents the local slope, represents the normalized difference vegetation index, It is used to perform interval restriction processing on the dynamic initial loss ratio expression to ensure the rationality of its physical meaning and value range.

[0067] clip(…, 0, 0.5) is a truncation function that limits the result to the interval [0, 0.5]: if the value of η calculated by the linear function is less than 0, it is forced to 0; if it is greater than 0.5, it is set to 0.5. This avoids extreme outliers and ensures that η is always within a reasonable range.

[0068] In some embodiments, the processor establishes a dynamic CN value correction model to characterize the combined effects of different land types, vegetation cover, and slope on runoff generation. This model uses the Normalized Difference Vegetation Index (NDVI) as a measure of land cover, constructs a dual-factor adjustment mechanism, and incorporates the influence of slope to form an adaptive CN value estimation formula. Given that the fixed initial loss ratio (generally set at 0.2) in the traditional SCS-CN model fails to reflect the varying initial rainfall interception capacities of different land types (e.g., forestland, bare land, and cultivated land), this paper introduces a dynamic initial loss coefficient mechanism to differentially regulate runoff response and construct an adaptive runoff model. This model creates a natural transition between forest, grassland, bare land, and hardened land, effectively improving the accuracy of runoff estimation in response to land type changes.

[0069] Using this method, the model can calculate surface runoff depth in complex terrain, providing key input parameters for subsequent drainage facility flow design. Combined with high-precision DEM data extracted from the study area's catchment unit boundaries and the spatiotemporal distribution of rainfall intensity, it ultimately enables detailed calculation of runoff in each sub-basin, ensuring that the results closely align with the hydrological characteristics along the railway line.

[0070] In some embodiments, the processor can set the soil thickness in the study area to 0.5-1m; the bedrock is dolomite interbedded with shale, with developed fissures and smooth drainage; the bedrock in the study area is mainly dolomite. According to the given empirical value

[20] , combined with data and field survey information, the rock layer permeability coefficient can be obtained to be 4e-7cm / s, about 4μm / s. In summary, the soil hydrological classification in the study area is Class C. Combined with the land use data interpreted from remote sensing images (forest land accounts for 75%, bare land 15%, and cultivated land 10%), soil texture categories, and slope and NDVI values ​​extracted from high-resolution DEM, the CN values ​​corresponding to the dry and wet conditions of each type are first set according to the SCS-CN parameter recommendation table. Subsequently, the slope + NDVI joint correction model constructed by the present invention is used to perform spatial correction on the CN values ​​to improve the perception accuracy of the CN values ​​for different surface runoff production capacities. The corrected CN values ​​are obtained according to formula (3). The surface runoff is calculated using the improved SCS-CN model (formula (4)). Results show that across the 23 catchment units in the study area, runoff ranged from 2.1×10³m³ to 3.4×10³m³, with an average runoff of 7.4×10³m³. The highest runoff occurred in the northern catchment unit (number 1), which has a catchment area of ​​0.749 km², an average slope of 35°, and a slope-corrected CN value of 75. The lowest runoff occurred in the southern catchment unit (number 13), which has a catchment area of ​​0.046 km², an average slope of 28°, and a corrected CN value of 75. The spatial distribution of runoff showed a trend of higher runoff in the north and lower runoff in the south, with runoff depths ranging from 45 mm to 46 mm.

[0071] The initial pour point information is the pour point location information of the basin outlet.

[0072] S330: Using the pour point screening layer, perform flood safety buffer analysis and feature elimination on the initial pour point information and railway line position vector data to obtain a set of threatening pour points.

[0073] The threatening pour point set is the set of pour points within the buffer distance.

[0074] In some embodiments, the processor can extract adjacent pouring points through spatial topology screening, and perform distance correction in combination with the terrain slope factor to retain potential threat points with significant hydraulic connectivity, such as Figure 5 As shown in the figure, the spatial distribution of the selected pour points accurately represents the dynamic interaction between mountain runoff and railway alignment. Meanwhile, the railway construction site design map is overlaid, and pour points in special sections such as tunnel entrances and exits and bridge piers are manually removed. This filtered point set will serve as the input data for DBSCAN density clustering, providing spatial constraints for subsequent density clustering.

[0075] In some embodiments, the expression for the actual buffer distance of the threatening pour point set may be: ; in, Indicates the actual buffer distance, represents the railway flood prevention safety distance threshold, represents the maximum slope in the study area, Indicates the local slope.

[0076] S340: Using the spatial clustering analysis layer, clustering and adaptively optimizing the threatening pouring point set is performed to obtain cluster spatial characteristics.

[0077] Cluster spatial features are features that reflect the spatial location information of clusters.

[0078] In some embodiments, the processor can use the spatial clustering analysis layer to calculate the distance between each point and its minimum number of neighboring points, form a K-distance curve after arranging in ascending order, and select the mutation gradient at the inflection point as the adaptively optimized neighborhood radius threshold; use the adaptively optimized neighborhood radius threshold to cluster the threatening pouring point set to obtain a clustering result; perform a secondary screening on the clustering result to eliminate invalid clusters located in the tunnel and bridge structure areas to obtain the cluster cluster spatial characteristics.

[0079] The adaptively optimized neighborhood radius threshold is the threshold of the cluster neighborhood radius after adaptive optimization.

[0080] In some embodiments, the processor can implement adaptive estimation of the neighborhood radius threshold (EPS value) by introducing the K-distance curve inflection point method: Specifically, the distance from each point to its minimum number of neighboring points (min_samples) is first calculated, and the K-distance curve is formed after arranging them in ascending order. The sudden change gradient at the inflection point is selected as the optimal EPS threshold. This method can effectively distinguish high-density water flow convergence clusters from sparse isolated points, improving the physical rationality of the clustering results, such as Figure 6 shown.

[0081] The clustering results reflect the clustering of threatening pouring points.

[0082] In some embodiments, the processor can utilize the DBSCAN clustering method to cluster and adaptively optimize the set of threatening pour points, thereby obtaining cluster spatial characteristics. During the clustering process, once a point's neighborhood contains a sufficient number of other pour points, it is marked as a core point, triggering a cluster expansion process that gradually incorporates boundary points to construct a cluster structure, as shown in Figures 7(a) and 7(b). To enhance project adaptability, the clustering results are secondary screened using DEM elevation profile data and design alignment layers along the railway line. Invalid clusters located in structural areas such as tunnels and bridges are automatically eliminated, ensuring that drainage facility locations are concentrated only in areas of actual catchment threat.

[0083] S350: Using the geometric recognition classification layer, principal component analysis is performed on the cluster spatial features, and a cluster morphology classification result is obtained by setting a variance explanation rate threshold.

[0084] The variance explanation rate threshold is the threshold used to divide linear clusters and nonlinear clusters.

[0085] The cluster morphology classification results reflect the spatial characteristics of the clusters and the drainage layout. For example, the cluster morphology classification results can include the layout and direction of linear drains and centralized drainage facilities.

[0086] In some embodiments, the processor may use a geometric recognition classification layer to perform principal component analysis on the cluster spatial features to obtain a variance contribution rate of each principal component; Based on the variance explanation rate threshold, the variance contribution rate of the principal components is analyzed to obtain linear clusters and nonlinear clusters; The least squares method is used to fit the points in the linear cluster to extract the main axis direction angle, and the layout position and direction of the linear drainage ditch are determined by combining the railway direction and the drainage slope layout requirements. For nonlinear clusters, the lowest elevation point is selected as the recommended facility location by traversing the point set within the cluster, and the preferentially matched centralized drainage facilities are obtained; wherein, the layout position and direction of the linear drainage ditch and the centralized drainage facilities belong to the cluster morphology classification results.

[0087] The variance contribution rate of the principal component is the ratio of the contribution of each principal component to the variance.

[0088] In some embodiments, as Figure 9 and Figure 10 As shown, to analyze the geometric morphology of each cluster, the processor uses principal component analysis (PCA) to reduce the dimensionality of the clusters. PCA projects the data onto the direction of maximum variance, extracts key features, and calculates the proportion of variance explained by the principal components. Unlike traditional PCA, which is typically used for global dimensionality reduction, this paper restricts PCA to the local scope of each cluster, constructing a point set covariance matrix. If the variance explained by the first principal component exceeds a certain percentage, it indicates that most of the variation in the cluster is concentrated in that principal component direction, and the cluster structure approaches a linear structure. Conversely, if the percentage is low, it indicates that the cluster has significant variation in multiple directions, indicating a nonlinear structure. Using this principle, a correlation is further established between geometric cluster types and drainage facility layout methods: For linear clusters, the least squares method is used to fit a straight line to the cluster point set, extracting the principal axis angle. The location and orientation of the linear drainage ditch are then determined based on the railway alignment and drainage slope layout requirements. Nonlinear cluster: Considering its multi-directional concentrated confluence characteristics, priority is given to matching centralized drainage facilities, such as culverts or drop wells, and by traversing the point set within the cluster, the lowest elevation point is selected as the recommended facility location to ensure that water flow can be preferentially diverted to the lowest point to reduce the risk of accumulation. Figure 8 This method, through a dual-discrimination mechanism of "geometric principal direction identification + drainage structure matching," not only improves the accuracy of cluster morphology determination but also directly converts spatial analysis results into facility layout parameters, paving the way for engineering implementation from spatial clustering to drainage facility selection, effectively improving the rationality and feasibility of drainage site selection decisions in mountainous areas.

[0089] In some embodiments, the processor can draw a K-Distance graph (e.g., Figure 11 As shown in the figure), the optimal distance parameter is determined to be the vertical coordinate corresponding to the inflection point in the figure, that is, eps=20.41. Clustering is performed using the DBSCAN algorithm. Figure 13 The clustering results of pouring points near the railway are shown. Each color represents a cluster, and a total of 16 clusters are classified. The clusters are identified one by one through principal component analysis (PCA). According to the dynamic threshold value method, with the first principal component variance contribution rate > 83% as the boundary, a total of 7 linear clusters and 5 nonlinear clusters are identified (such as Figure 12 , excluding single-point clusters), including 5 linear clusters that intersect the railway at large angles (see Figure 13(a) in the figure) and two linear clusters intersecting the railway at small angles (see Figure 13 (b) in the figure). For linear clusters that intersect at small angles, the least squares method is used to fit straight lines and accurately simulate the location of drainage ditches, ensuring smooth drainage and meeting the actual terrain requirements. However, linear clusters that intersect at large angles are classified as nonlinear clusters due to their spatial distribution characteristics and do not conform to the setting rules of roadbed drainage ditches. For clusters including multiple points ( Figure 13 (c)) and single point cluster ( Figure 13 For the nonlinear clusters in (b), the lowest elevation point within the cluster was uniformly selected as the culvert location. The drainage ditches and culvert locations along the railway were determined, resulting in a total of 14 culverts and 2 drainage ditches.

[0090] S360: Using the drainage facility recommendation layer, based on the cluster morphology classification results, the runoff data and the roadbed data, the standard parameters are reversed through the Manning equation to obtain the site selection results of the mountain railway drainage facilities, thereby completing the site selection of the mountain railway drainage facilities.

[0091] In some embodiments, the processor can use the SCS-CN model to calculate surface runoff based on the cluster morphology classification results, the runoff data and the roadbed data to obtain the surface runoff depth. Combined with the previously divided water unit area, the two can be multiplied to obtain the surface runoff of each water unit, thereby inversely calculating the size of the culvert and drainage ditch based on the surface runoff. According to the clustering and PCA analysis results, a total of 14 culvert points and 2 drainage ditches were identified in the study area. Figure 14 As shown. Combined with the surface runoff data, the allowable flow velocity of concrete masonry (4m / s) is used as the critical flow velocity. At the same time, considering the three conditions of no blockage, 1 / 3 blockage, and 2 / 3 blockage, the design flow of each drainage facility is calculated for flow design. (1) Culvert design: The lowest elevation point in the nonlinear cluster is selected as the culvert location, and the maximum runoff of the water catchment unit is used as the design flow. For example, under the rainfall condition of 100mm / h, the surface runoff of the water catchment unit (No. 4) is 2.02m³ / s. The corresponding culvert size can be designed according to the Manning equation. At the same time, economic benefits can be considered while ensuring safety. For example, a reinforced concrete circular culvert is used, with a culvert diameter of 1.2m and a flow capacity of 4.52m³ / s, which can meet the water flow under the condition of 2 / 3 congestion of the culvert. (2) Drainage ditch design: For the linear cluster (angle with the railway line <30°), the center line of the drainage ditch is fitted and the longitudinal slope is calculated. Taking the catchment unit (No. 16) as an example, under rainfall conditions of 100 mm / h, the surface runoff is 2.55 m³ / s. According to the Manning equation, a trapezoidal cross-section drainage ditch (bottom width 0.8 m, depth 0.6 m, slope ratio 1:1.5) can be designed with a flow capacity of 4.08 m³ / s, which also meets the water flow under 2 / 3 congestion conditions and meets the design requirements.

[0092] In some embodiments, a total of 15 culverts and 6 drainage ditches were actually laid out at the project site, and runoff in the remaining areas was drained through bridge drain holes and tunnel gutter systems. After comparative analysis, the culvert locations predicted by the model were highly consistent with the on-site implementation, with an accuracy rate of 92%. Figure 15 As shown, the prediction error primarily stems from the merging of two adjacent culverts. In practice, two independent slope-type debris flow gullies in this area require separate culverts to ensure smooth passage of debris flow and prevent siltation. In contrast, the predicted drainage ditch layout differs somewhat from the actual layout, with only two locations matching the predicted layout. The remaining four existing drainage ditches are not reflected in the prediction model. This is speculated to be related to the accuracy of local microtopographic feature extraction and the rainfall return period parameter settings. Compared to historical flood damage cases, the design flow of the original drainage facilities in the study area is generally 40% lower than the calculated value. The newly selected site covers 85% of the high-risk catchment area. GIS spatial overlay analysis shows that the optimized drainage system reduces the probability of waterlogging on the railway subgrade and significantly improves the railway's resilience to flooding. Furthermore, the design results agree with the field survey data at a rate of 92%, validating the engineering applicability of the site selection method based on DEM and cluster analysis.

[0093] The main error in culvert prediction is due to the merging of adjacent gullies on the slope, which fails to reflect the actual independent layout requirements. The deviation in drainage ditch prediction is related to the insufficient accuracy of local microtopography extraction. In addition, in actual situations, drainage ditches are usually distributed on both sides of the culvert to enable timely drainage of water from these ditches. Subsequent research can, based on the culvert prediction results, arrange drainage ditches in sequence according to this distribution pattern, and further combine it with roadbed slope data to improve the drainage system.

[0094] In some embodiments of the present specification, a method for site selection of drainage facilities for mountain railways based on a hydrological spatial site selection analysis model is provided. By preprocessing the collected data, mountain railway environmental data is obtained. The three-dimensional hydrological spatial site selection analysis model is used to perform spatial clustering dimensionality reduction analysis and surface runoff calculation on the mountain railway environmental data to obtain the site selection results of drainage facilities for mountain railways. This improves the accuracy of site selection of drainage facilities for mountain railways and their applicability under complex terrain conditions; reduces the probability of water accumulation in railway roadbeds, and significantly enhances the railway's ability to resist water damage.

Claims

1. A site selection method for railway drainage facilities in mountainous areas based on a hydrological spatial site selection analysis model is characterized by: include: S1: Preprocess the collected data to obtain mountain railway environmental data; S2: Construct a three-dimensional hydrological spatial site selection analysis model using the terrain factor analysis layer, runoff calculation layer, pour point screening layer, spatial cluster analysis layer, geometric identification classification layer, and drainage facility recommendation layer; S3: Using the three-dimensional hydrological spatial site selection analysis model, perform spatial clustering dimensionality reduction analysis and surface runoff calculation on the mountain railway environmental data to obtain the site selection results of the mountain railway drainage facilities, and complete the site selection of the mountain railway drainage facilities.

2. The method for site selection of drainage facilities for mountain railways based on the hydrological spatial site selection analysis model according to claim 1 is characterized in that: Said S1 comprises: The digital elevation model data, land use type data, soil hydrological condition information, and meteorological rainfall data are converted to the same coordinate system and spatial resolution, and boundary gaps and overlapping areas are repaired through topological consistency checking to obtain the processed digital elevation model data, land use type data, soil hydrological condition information, and meteorological rainfall data; The local window minimum filtering algorithm was used to filter the areas blocked by buildings under the forest in the digital elevation model data, and the exposed slope areas were smoothed and optimized. The elevation difference threshold method was used to eliminate abnormal elevation disturbances to obtain optimized digital elevation model data. The spatial distribution of soil use type data was checked and joint classification was performed using soil texture information to obtain joint classification data; Based on meteorological rainfall data, an image is constructed using the inverse distance weighted method to obtain a regional rainfall spatial distribution map; wherein the digital elevation model data, land use type data, soil hydrological condition information and meteorological rainfall data belong to the collected data, and the optimized digital elevation model data, the joint classification data and the regional rainfall spatial distribution map belong to the mountain railway environment data.

3. The method for site selection of drainage facilities for mountain railways based on the hydrological spatial site selection analysis model according to claim 2 is characterized in that: The S3 includes: S310: Using the terrain factor analysis layer, extracting terrain features from the optimized digital elevation model data to obtain a terrain basic factor map; S320: Using the runoff calculation layer, based on the terrain basic factor map, the joint classification data, the soil hydrological condition information, and the regional rainfall spatial distribution map, a slope correction is introduced to perform calculations to obtain the water catchment unit, runoff data, and initial pour point information; S330: Using the pour point screening layer, performing flood safety buffer analysis and feature elimination on the initial pour point information and railway line vector data to obtain a set of threatening pour points; S340: using a spatial clustering analysis layer, clustering and adaptively optimizing the threatening pouring point set to obtain cluster spatial characteristics; S350: using the geometric recognition classification layer, performing principal component analysis on the cluster spatial features, and obtaining cluster morphology classification results by setting a variance explanation rate threshold; S360: Using the drainage facility recommendation layer, based on the cluster morphology classification results, the runoff data and the roadbed data, the standard parameters are reversed through the Manning equation to obtain the site selection results of the mountain railway drainage facilities, and complete the site selection of the mountain railway drainage facilities; among them, the terrain factor analysis layer, the runoff calculation layer, the pour point screening layer, the spatial clustering analysis layer, the geometric identification classification layer and the drainage facility recommendation layer belong to the three-dimensional hydrological space site selection analysis model.

4. The method for site selection of drainage facilities for mountain railways based on the hydrological spatial site selection analysis model according to claim 3 is characterized in that: The expression of the runoff data is: ； ； ； ； in, represents runoff data, represents the total rainfall, Indicates the dynamic initial loss coefficient, that is, at position The initial rainfall loss ratio at Represents the horizontal coordinate of the grid cell, Represents the vertical coordinate of the grid cell, represents the potential water storage capacity, represents the modified general condition, Indicates wet conditions, Indicates drought conditions, represents a natural constant, represents the local slope, represents the normalized difference vegetation index, It is used to perform interval restriction processing on the dynamic initial loss ratio expression to ensure the rationality of its physical meaning and value range.

5. The method for site selection of drainage facilities for mountain railways based on a hydrological spatial site selection analysis model according to claim 3 is characterized in that: The expression of the actual buffer distance of the threatening pouring point set is: ； in, Indicates the actual buffer distance, represents the railway flood prevention safety distance threshold, represents the maximum slope in the study area, Indicates the local slope.

6. The method for site selection of drainage facilities for mountain railways based on a hydrological spatial site selection analysis model according to claim 3, characterized in that: The S340 includes: Using the spatial clustering analysis layer, the distance between each point and its minimum number of neighboring points is calculated, and the K-distance curve is formed after arranging them in ascending order. The mutation gradient at the inflection point is selected as the neighborhood radius threshold for adaptive optimization. Clustering the threatening pouring point set using the adaptively optimized neighborhood radius threshold to obtain a clustering result; The clustering results are screened twice to eliminate invalid clusters located in tunnel and bridge structure areas, and the spatial characteristics of clusters are obtained.

7. The method for site selection of drainage facilities for railways in mountainous areas based on a hydrological spatial site selection analysis model according to claim 3, characterized in that: The S350 includes: Using the geometric recognition classification layer, principal component analysis is performed on the cluster spatial characteristics to obtain the variance contribution rate of each principal component; Based on the variance explanation rate threshold, the variance contribution rate of the principal components is analyzed to obtain linear clusters and nonlinear clusters; The least squares method is used to fit the points in the linear cluster to extract the main axis direction angle, and the layout position and direction of the linear drainage ditch are determined by combining the railway direction and the drainage slope layout requirements. For nonlinear clusters, the lowest elevation point is selected as the recommended facility location by traversing the point set within the cluster, and the preferentially matched centralized drainage facilities are obtained; wherein, the layout position and direction of the linear drainage ditch and the centralized drainage facilities belong to the cluster morphology classification results.

Images