Railway intelligent route selection method based on multi-source data fusion

Abstract

This invention discloses an intelligent route selection method for seamless railways based on multi-source data fusion, belonging to the field of railway route selection technology. Addressing the issue that existing route selection methods do not consider the special requirements of seamless railways regarding ride comfort and temperature stress, this invention acquires multi-source geospatial data along the route and registers it to a gridded model. It calculates the basic cost of each grid cell, as well as the ride comfort cost (a weighted sum of elevation and direction change costs between adjacent cells). Then, it calculates the temperature stress risk coefficient based on historical air temperature and the locked track temperature range. Using the starting and ending points as constraints, it searches for the path that minimizes the weighted sum of the basic cost, ride comfort cost, and temperature stress risk coefficient. This invention can be used for automatic route selection of seamless railways in areas with large temperature differences, mitigating temperature stress and ride comfort risks from the outset.

Description

Technical Field

[0001] This invention relates to the field of railway alignment technology. More specifically, this invention relates to a method for intelligent railway alignment based on multi-source data fusion. Background Technology

[0002] Seamless railways, by eliminating rail joints in traditional jointed tracks, offer significant advantages such as smoother operation, less maintenance, and longer service life, making them the mainstream track structure for high-speed and heavy-haul railways. However, seamless railways place far higher demands on track geometry smoothness and temperature adaptability than jointed tracks. During the route selection phase, changes in the longitudinal profile gradient and the magnitude of the horizontal turning angle directly affect the additional stress and stability of the rails in seamless railways. Simultaneously, the rail temperature stress under large temperature differences (closely related to the locking rail temperature design value), if not adequately considered during the route selection phase, may lead to safety accidents such as rail bulging, runaway tracks, or rail breaks during operation.

[0003] Traditional railway route selection methods primarily rely on manual experience and two-dimensional topographic map analysis, determining the recommended route through comparison of multiple options. In recent years, with the development of Geographic Information Systems (GIS) and intelligent optimization algorithms, some intelligent route selection technologies based on gridded modeling and path search have emerged. For example, existing literature discloses methods that divide the route selection area into regular grids, store topographic, geological, and engineering investment attribute data for each grid cell, and use A* or genetic algorithms to search for the route with the lowest construction cost. However, these existing methods, when constructing cost functions, mainly focus on conventional factors such as terrain complexity, earthwork volume, and ecological impact, failing to establish corresponding constraints or cost models for the special requirements of seamless railways. These existing technologies all treat geometric constraints as mandatory hard conditions (hard constraints), rather than continuous costs (soft constraints) that can be balanced between construction cost and smoothness. They cannot handle situations where the ideal route requires slight deviations from constraints, nor do they incorporate operational risks such as temperature stress into the optimization objective. Specifically, existing technologies generally lack the following two aspects: First, they lack a quantitative model for balancing track smoothness (elevation change rate and horizontal turning angle) with construction costs. This results in searched routes having lower construction costs, but potentially failing to meet the geometric continuity requirements of seamless railways due to abrupt changes in gradient or excessively small curve radii. Second, they lack a quantitative assessment of temperature stress risks, especially in regions with large temperature differences. The extent to which historical temperature fluctuations exceed the track temperature range of seamless railways directly determines the amplitude of temperature stress generated within the rails. Existing route selection methods completely ignore this crucial factor, potentially leading to selected routes located in areas with drastic temperature fluctuations, posing hidden dangers for subsequent track temperature design and operational safety.

[0004] Furthermore, in special geographical environments such as North Africa, railway lines face multiple risks, including sandstorms, saline soil swelling, and long-term climate change. Current intelligent route selection technologies lack a comprehensive solution that can uniformly model and collaboratively optimize the geometric continuity constraints, temperature stress risks, sandstorm protection costs, salinization risks, and long-term climate adaptability costs of seamless railways. Therefore, there is an urgent need for an intelligent route selection method that comprehensively considers the special technical requirements of seamless railways, mitigating these risks from the outset during the design phase and achieving the automatic generation of safe, economical, and environmentally friendly routes. Summary of the Invention

[0005] One object of the present invention is to solve at least the above-mentioned problems and to provide at least the advantages that will be described later.

[0006] Another objective of this invention is to provide a railway intelligent route selection method based on multi-source data fusion, which solves the problem that existing route selection methods do not take into account the special requirements of seamless railways for smoothness and temperature stress.

[0007] To achieve these objectives and other advantages according to the present invention, a railway intelligent route selection method based on multi-source data fusion is provided, comprising the following steps: S1. Obtain multi-source geospatial data of the railway line area, wherein the multi-source geospatial data includes at least topographic data, geological data, ecologically sensitive area data and climate data; S2. Register and overlay the multi-source geospatial data into the same gridded geographic model, and calculate the basic cost for each grid cell. The basic cost is used to characterize the construction difficulty and ecological impact of building a railway in that grid cell. S3. Based on the geometric continuity requirements of seamless railway lines, define the smoothness cost between adjacent grid cells. The smoothness cost is the weighted sum of the elevation change cost and the planar direction change cost between adjacent grid cells. S4. Based on the historical temperature fluctuation range in the climate data and the preset seamless railway track temperature range, calculate the temperature stress risk coefficient of the area where each grid cell is located. The temperature stress risk coefficient is positively correlated with the degree to which the historical temperature fluctuation range exceeds the track temperature range. S5. Using the starting point and the ending point as constraints, a path search algorithm is used to find a path that minimizes the total cost. The total cost is the sum of the basic costs of all grid cells on the path multiplied by the first weight, plus the sum of the smoothness costs between all adjacent grid cells on the path multiplied by the second weight, plus the sum of the temperature stress risk coefficients of all grid cells on the path multiplied by the third weight.

[0008] Preferably, the elevation change cost is the ratio of the actual slope between two adjacent grid cells to the preset maximum allowable slope, wherein the actual slope is equal to the absolute value of the elevation difference divided by the horizontal distance between adjacent grid cells; the planar direction change cost is the ratio of the directional angle between two adjacent grid cells to the preset maximum allowable angle.

[0009] Preferably, the registration and overlay into the same gridded geographic model includes: unifying the multi-source geospatial data to the same coordinate system and the same spatial resolution, and generating the attribute vector of each grid cell through raster overlay analysis.

[0010] Preferably, the path with the minimum total cost found in S5 is output as the horizontal and vertical alignment design parameters of the railway line after curve smoothing.

[0011] Preferably, in step S2, when calculating the basic cost for each grid cell, the method further includes: Acquire wind and sand data for the area where the grid cell is located, wherein the wind and sand data includes at least the sand transport volume and the dune movement speed; A wind and sand protection cost model is constructed, which is used to calculate the cost of wind and sand protection engineering and maintenance required during the construction and operation of the railway in the grid unit. The wind and sand protection cost is positively correlated with the sand transport volume and the dune movement speed. The cost of wind and sand protection is weighted and included in the basic cost as an additional item.

[0012] Preferably, in step S2, when calculating the basic cost for each grid cell, the method further includes: Based on the cumulative function of the frequency and amplitude of the historical temperature fluctuation range exceeding the preset seamless railway track temperature range, the long-term climate adaptation cost of the area where the grid cell is located is calculated. The long-term climate adaptation cost increases nonlinearly with the increase of the cumulative frequency and amplitude of the exceeding events. The long-term climate adaptation costs are weighted and included in the base costs as an additional item.

[0013] Preferably, in step S2, when calculating the basic cost for each grid cell, the method further includes: Obtain data on the distribution of saline soil and the depth of groundwater in the area where the grid cell is located; Based on the saline soil distribution data and groundwater depth data, the salt swelling risk coefficient of the area where the grid unit is located is calculated. The salt swelling risk coefficient is proportional to the soluble salt content. When the capillary rise height is greater than the foundation depth, the risk coefficient is additionally multiplied by the ratio of the capillary rise height to the foundation depth. The salt swelling risk coefficient is weighted and included in the basic cost as an additional item.

[0014] Preferably, the temperature stress risk factor is calculated using the following formula: R = 1 + μ·ΔT; Where R is the temperature stress risk coefficient, ΔT=max(0, T1-T4) + max(0, T3-T2), T1 is the historical highest temperature, T2 is the historical lowest temperature, [T3, T4] is the preset seamless railway locking rail temperature range, and μ is the preset sensitivity coefficient.

[0015] The present invention has at least the following beneficial effects: First, by defining the weighted sum of the elevation change cost and the direction change cost between adjacent grid cells as the smoothness cost, and having it participate in the optimization of the total cost together with the basic cost, the selected line can meet the geometric continuity requirements of seamless railways for gentle gradients and smooth curves while satisfying the construction economy, thus avoiding repeated adjustments due to insufficient smoothness in the later stages. Second, by introducing the degree to which historical air temperature exceeds the locked rail temperature range, the temperature stress risk coefficient is calculated and added as an independent item to the total cost, so that the route selection results automatically avoid extreme temperature difference areas, thereby reducing the temperature stress risk of seamless railway rails from the source and ensuring the safety margin of the locked rail temperature design. Third, by weighting the costs of wind and sand protection, the costs of nonlinear long-term climate adaptation based on the accumulation of frequency and amplitude, and the risk coefficient of salt swelling based on the content of soluble salts and the height of capillary rise of groundwater into the basic cost, the system achieves synergistic avoidance of wind and sand, salt swelling and long-term climate change in special environments such as North Africa, and significantly reduces the maintenance costs and disaster risks during the operation period. Fourth, by smoothing the optimal polyline path and outputting the horizontal and vertical alignment parameters that conform to the design specifications, a seamless connection from discrete grid paths to continuous constructable lines is achieved, making this method practically applicable for directly guiding railway construction drawing design.

[0016] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Detailed Implementation

[0017] The present invention will be further described in detail below with reference to embodiments, so that those skilled in the art can implement it based on the description.

[0018] This invention discloses a railway intelligent route selection method based on multi-source data fusion, comprising the following steps: S1. Obtain multi-source geospatial data of the railway line area, wherein the multi-source geospatial data includes at least topographic data, geological data, ecologically sensitive area data and climate data; S2. Register and overlay the multi-source geospatial data into the same gridded geographic model, and calculate the basic cost for each grid cell. The basic cost is used to characterize the construction difficulty and ecological impact of building a railway in that grid cell. S3. Based on the geometric continuity requirements of seamless railway lines, define the smoothness cost between adjacent grid cells. The smoothness cost is the weighted sum of the elevation change cost and the planar direction change cost between adjacent grid cells. S4. Based on the historical temperature fluctuation range in the climate data and the preset seamless railway track temperature range, calculate the temperature stress risk coefficient of the area where each grid cell is located. The temperature stress risk coefficient is positively correlated with the degree to which the historical temperature fluctuation range exceeds the track temperature range. S5. Using the starting point and the ending point as constraints, a path search algorithm is used to find a path that minimizes the total cost. The total cost is the sum of the basic costs of all grid cells on the path multiplied by the first weight, plus the sum of the smoothness costs between all adjacent grid cells on the path multiplied by the second weight, plus the sum of the temperature stress risk coefficients of all grid cells on the path multiplied by the third weight. The temperature stress risk factor is calculated using the following formula: R = 1 + μ·ΔT; Where R is the temperature stress risk coefficient, ΔT=max(0, T1-T4) + max(0, T3-T2), T1 is the historical highest temperature, T2 is the historical lowest temperature, [T3, T4] is the preset seamless railway locking rail temperature range, and μ is the preset sensitivity coefficient.

[0019] In the above technical solution, a railway intelligent route selection method based on multi-source data fusion first executes steps S1 and S2. In steps S1 and S2, multi-source geospatial data of the railway line area is first acquired. This multi-source geospatial data includes at least topographic data, geological data, ecologically sensitive area data, and climate data. Specifically, topographic data can be collected using a drone equipped with an optical camera and lidar, or a satellite remote sensing platform. The resolution of the topographic data can be set to 5m to 30m. Geological data can be obtained from regional geological maps or field exploration reports. The gridded sampling interval for geological data can be set to 10m to 50m. Ecologically sensitive area data can be extracted from nature reserve boundary vector maps or land use status maps. The buffer distance for ecologically sensitive areas can be set to 200m to 500m. Climate data can be obtained from historical observation records of meteorological stations along the railway line. The statistical period for temperature data can be set to the past 30 years.

[0020] When registering and superimposing the above topographic data, geological data, ecological sensitive area data and climate data onto the same gridded geographic model, all data may first be converted to the same projected coordinate system, such as the UTM projection, and then the raster data of different resolutions are resampled to a unified spatial resolution through a bilinear interpolation method, and the unified resolution may be set to 10m×10m or 20m×20m.

[0021] Then the base cost is calculated for each grid cell. The base cost is obtained by weighted summation of the construction difficulty coefficient and the ecological impact coefficient. The construction difficulty coefficient is estimated according to the average slope, foundation bearing capacity and earthwork volume of the cell. The specific calculation method of the construction difficulty coefficient is as follows: the slope difficulty coefficient is determined according to the average slope. Specifically, when the average slope ≤10°, the slope difficulty coefficient is 0.2; when the average slope is between 10° and 20°, the slope difficulty coefficient is 0.5; when the average slope ≥20°, the slope difficulty coefficient is 1.0. The foundation difficulty coefficient is determined according to the foundation bearing capacity in the geological survey report. Specifically, when the bearing capacity ≥200kPa, the foundation difficulty coefficient is 0.2; when the bearing capacity is between 100kPa and 200kPa, the foundation difficulty coefficient is 0.5; when the bearing capacity <100kPa, the foundation difficulty coefficient is 1.0. Then the slope difficulty coefficient and the foundation difficulty coefficient are weighted and summed, wherein the weight of the slope difficulty coefficient is set to 0.6, and the weight of the foundation difficulty coefficient is set to 0.4, and the obtained result is the construction difficulty coefficient. The ecological impact coefficient is determined according to the distance d between the cell and the boundary of the ecological sensitive area. Specifically, when d ≤50m, the ecological impact coefficient is 1.0; when 50m <d ≤200m, the ecological impact coefficient is 0.5; when d>200m, the ecological impact coefficient is 0. Finally, the calculation formula of the base cost is: base cost = 0.6 × construction difficulty coefficient + 0.4 × ecological impact coefficient.

[0022] In step S3, the ride comfort cost between adjacent grid cells is defined. First, the elevation change cost is calculated, and the elevation change cost is the ratio of the actual slope of two adjacent grid cells to the preset maximum allowable slope, wherein the actual slope is equal to the absolute value of the elevation difference divided by the horizontal distance between adjacent grid cells (i.e., the grid resolution). The preset maximum allowable slope is determined according to the railway grade, which may be 15‰ for high-speed seamless railways and 25‰ for ordinary seamless railways. Then the plane direction change cost is calculated, and the plane direction change cost is the ratio of the direction rotation angle of two adjacent grid cells to the preset maximum allowable rotation angle. The preset maximum allowable rotation angle may be 30°, which corresponds to a minimum curve radius of about 300m. Finally, the ride comfort cost is the weighted sum of the elevation change cost and the plane direction change cost, wherein the weight of the elevation change cost is set to 0.5, and the weight of the plane direction change cost is set to 0.5.

[0023] In step S4, the temperature stress risk coefficient is calculated. Based on the historical temperature fluctuation range in the climate data, the historical maximum and minimum temperatures of the region where each grid cell is located are obtained. The historical maximum temperature can be the extreme maximum value of the past 30 years, denoted as T1, in °C; the historical minimum temperature can be the extreme minimum value of the past 30 years, denoted as T2, in °C. The preset track temperature range for seamless railways is set to 25 °C ± 5 °C for North Africa, i.e., the lower limit T3 = 20 °C and the upper limit T4 = 30 °C. The temperature stress risk coefficient is positively correlated with the degree to which the historical temperature fluctuation range exceeds the locked track temperature range, and can be calculated according to the formula R = 1 + μ·ΔT. Wherein, R is the temperature stress risk coefficient, ΔT = max(0, T1-T4) + max(0, T3-T2), in °C; μ is the preset sensitivity coefficient, with a value ranging from 0.02 / °C to 0.10 / °C, and its value is determined comprehensively based on the rail grade, the construction accuracy of the locked track temperature, and the operational safety factor. For U75V hot-rolled rails, μ can be taken as 0.05 / ℃; for higher strength rails, μ can be appropriately reduced. In the above formula, the product of μ and ΔT is dimensionless, therefore R is a dimensionless coefficient. When the historical temperature is completely within the locked rail temperature range, T1≤T4 and T2≥T3, then ΔT=0, R=1. For example, if the historical highest temperature in a certain area is 45℃ and the lowest temperature is 5℃, then ΔT = (45-30)+(20-5)=15+15=30℃, R=1+0.05×30=2.5.

[0024] In step S5, using the start and end points as constraints, a path search algorithm is employed to find the path with the minimum total cost. The A* algorithm can be used as the path search algorithm, and its heuristic function can be the Euclidean distance from the current grid cell to the end point. The total cost is calculated as follows: Total Cost = w1 × Σ(Basic Cost) + w2 × Σ(Smoothness Cost) + w3 × Σ(Temperature Stress Risk Coefficient). Where Σ(Basic Cost) represents the sum of the basic costs of all grid cells along the path, Σ(Smoothness Cost) represents the sum of the smoothness costs of all adjacent grid cell pairs along the path, and Σ(Temperature Stress Risk Coefficient) represents the sum of the temperature stress risk coefficients of all grid cells along the path; w1, w2, and w3 are the first, second, and third weights, respectively, and can be adjusted according to engineering requirements. When smoothness requirements are strict, w2 can be increased; when temperature stress risk is prominent, w3 can be increased. For example, w1, w2, and w3 can be set to 0.4, 0.4, and 0.2, respectively.

[0025] During the search process, starting from the initial grid, the search expands to eight adjacent grids in each direction. The cumulative total cost from the initial grid to the current grid is calculated, and a heuristic cost estimate is added. The grid with the lowest total cost is expanded first, until the endpoint grid is reached. After the search is complete, a polyline path connecting the center points of the grid cells is output. The base cost of each grid cell on this path, the smoothness cost between adjacent grid cells, and the temperature stress risk coefficient of each grid cell are all calculated and accumulated using the aforementioned method, ultimately yielding the path with the lowest total cost.

[0026] The beneficial effects of this technical solution are as follows: First, by including the smoothness cost (a weighted sum of elevation change cost and direction change cost) and the basic cost in the path optimization, the automatically generated route achieves an adjustable balance between construction economy and seamless railway geometric continuity, avoiding the infeasibility of later schemes due to neglecting gradient or turning angle constraints in manual route selection. Second, by introducing a temperature stress risk coefficient and independently including it in the total cost, the route selection results can prioritize areas with a high degree of overlap between historical temperature fluctuation ranges and locked rail temperature ranges, reducing the additional stress on rails caused by excessive temperature differences during operation and lowering the probability of rail buckling, runaway, or rail breakage. Third, the entire calculation process is based on a gridded geographic model and a standard path search algorithm, which can process large-scale geographic data in batches, and the output results are repeatable with good consistency among different designers. Fourth, the output polyline path can be further used for construction drawing design, shortening the route selection cycle and reducing the workload of site surveys and scheme comparisons.

[0027] In another technical solution, the elevation change cost is the ratio of the actual slope between two adjacent grid cells to the preset maximum allowable slope, wherein the actual slope is equal to the absolute value of the elevation difference divided by the horizontal distance between adjacent grid cells; the planar direction change cost is the ratio of the directional angle between two adjacent grid cells to the preset maximum allowable angle.

[0028] In the above technical solution, the elevation change cost is calculated as the ratio of the actual slope between two adjacent grid cells to the preset maximum allowable slope. Specifically, the elevation values ​​of two adjacent grid cells are obtained and denoted as z1 and z2, respectively, in meters. The absolute value of the elevation difference Δh = |z2 - z1|, in meters. The horizontal distance between adjacent grid cells is a fixed value L (e.g., equal to a grid resolution of 10 meters), in meters. The actual slope i = Δh / L, which is a dimensionless ratio. The preset maximum allowable slope is determined according to the railway grade and is also a dimensionless ratio. For example, for high-speed seamless railways, it can be taken as 15‰, i.e., 0.015; for ordinary seamless railways, it can be taken as 25‰, i.e., 0.025. The preset maximum allowable slope is denoted as P. Then the elevation change cost = i / P, which is a dimensionless ratio.

[0029] The cost of a change in planar orientation is calculated as the ratio of the directional angle between two adjacent grid cells to the preset maximum allowable angle. Specifically, the center points A, B, and C of three adjacent grid cells are selected, and the orientation angles of vectors AB and BC are calculated, with true north as 0° and measured clockwise. The orientation angle of vector AB is denoted as α1, and the orientation angle of vector BC is denoted as α2. Then, the directional angle θ = min(|α2 - α1|, 360° - |α2 - α1|), in degrees. The preset maximum allowable angle is determined according to the minimum curve radius in railway design specifications; for example, it can be 15° for high-speed railways and 30° for ordinary railways. The preset maximum allowable angle is denoted as Φ, in degrees. Therefore, the cost of a change in planar orientation = θ / Φ, which is a dimensionless ratio.

[0030] After calculating the elevation change cost and the horizontal direction change cost separately, they are substituted into the smoothness cost formula for weighted summation. For example, if the absolute value of the elevation difference between adjacent grid cells is 3 meters and the horizontal distance between grid cells is L=10 meters, then the actual slope i=3 / 10=0.3 (i.e., 30‰). The preset maximum allowable slope P is 15‰ (i.e., 0.015), then the elevation change cost = i / P=0.3 / 0.015=20 (dimensionless). If the direction angle θ=10° and the preset maximum allowable angle Φ=30°, then the horizontal direction change cost = θ / Φ ≈ 0.333 (dimensionless). Depending on the project focus, the elevation change cost and the horizontal direction change cost can be assigned a weight of 0.5 respectively, then the smoothness cost = 0.5×20+0.5×0.333=10.1665. The above calculation example is only for illustrating the calculation method; the actual weights can be adjusted according to project requirements.

[0031] The beneficial effects of this technical solution are as follows: By adopting dimensionless ratios for both elevation change cost (the ratio of actual slope to allowable slope) and horizontal direction change cost (the ratio of turning angle to allowable turning angle), the two engineering parameters can be weighted and summed under the same dimension, resulting in clear physical meaning. The elevation change cost directly reflects the excess multiple of the actual slope relative to the allowable slope between adjacent grid cells, while the horizontal direction change cost directly reflects the proportional relationship between the turning angle and the allowable limit. The weighted sum of the two can quantitatively evaluate the smoothness of local sections of the line. As an important component of the total path search cost, this smoothness cost can guide the algorithm to prioritize lines with gentle slopes and smooth turning angles, thereby meeting the stringent requirements of geometric continuity for seamless railways.

[0032] In another technical solution, the registration and overlay into the same gridded geographic model includes: unifying the multi-source geospatial data to the same coordinate system and the same spatial resolution, and generating the attribute vector of each grid cell through raster overlay analysis.

[0033] In the above technical solution, when registering and overlaying multi-source geospatial data onto the same gridded geographic model, the first step is to unify the coordinate system. Topographic data, geological data, ecologically sensitive area data, and climate data may originate from different projected coordinate systems or geographic coordinate systems. For example, topographic data may use the UTM projection, geological data may use the Gauss-Krüger projection, and ecologically sensitive area data may use geographic latitude and longitude coordinates. To perform overlay analysis within the same model, all data can be transformed to the same projected coordinate system. The UTM projection can be selected as the unified coordinate system, and a suitable central meridian can be chosen based on the longitude zone of the railway line area. The coordinate transformation can use a seven-parameter Bursa model or a three-parameter similarity transformation. The transformation parameters can be obtained from the local surveying department or calculated through common points. After the transformation, all data use the same unit of length, such as meters.

[0034] After establishing the coordinate system, spatial resolution is unified. Data from different sources have different original resolutions. For example, topographic data may have a resolution of 10m, geological data may have a sampling interval of 50m, and vector boundary maps of ecologically sensitive areas do not have a resolution concept but need to be rasterized. To enable pixel-by-pixel calculations within the same gridded model, all raster data needs to be resampled to a unified spatial resolution. This unified resolution can be set according to the accuracy requirements of railway alignment, for example, 10m×10m or 20m×20m. Resampling methods can include bilinear interpolation or the nearest neighbor method. Bilinear interpolation is suitable for continuous data such as elevation, while the nearest neighbor method is suitable for discrete data such as geological lithology coding. For vector data (such as ecologically sensitive area boundaries), it can first be rasterized into a raster layer consistent with the target resolution, where the value of each pixel indicates whether the location is within the ecologically sensitive area or its distance from the boundary.

[0035] After unifying the coordinates and resolution, raster overlay analysis is performed to generate attribute vectors for each grid cell. Specifically, for each grid cell within the study area, elevation values ​​are read from its corresponding topographic raster, foundation bearing capacity or lithology codes are read from the geological raster, distance to ecologically sensitive areas or protection levels are read from the ecological raster, and annual mean temperature, annual precipitation, or extreme temperature values ​​are read from the climate raster. These data are combined into a multidimensional attribute vector and stored in the data structure of that grid cell. For example, a grid cell located at coordinates (x=12345m, y=67890m) can have its attribute vector represented as [Elevation=156.3m, Slope=8.5°, Foundation Bearing Capacity=220kPa, Ecological Distance=120m, Annual Extreme Maximum Temperature=42℃, Annual Extreme Minimum Temperature=-2℃]. This attribute vector serves as the input data for subsequent calculations of basic cost and smoothness cost. This technical solution ensures that all grid cells have a consistent data spatial reference and resolution, providing a unified data foundation for subsequent path search algorithms.

[0036] The beneficial effects of this technical solution are as follows: By unifying the coordinate system and spatial resolution, spatial reference and scale differences between multi-source data are eliminated, enabling data from different sources to be calculated and overlaid within the same gridded model on a cell-by-cell basis. The attribute vectors generated by raster overlay analysis centrally store the multidimensional information of each grid cell, facilitating rapid retrieval and calculation of various costs during path search. This method effectively integrates multiple route selection elements such as terrain, geology, ecology, and climate, providing standardized data input for intelligent route selection and improving the scientific rigor and reliability of the route selection results.

[0037] In another technical solution, the path with the minimum total cost found in S5 is output as the design parameters for the horizontal and vertical alignment of the railway line after curve smoothing.

[0038] In the above technical solution, the path with the minimum total cost found in step S5 is a polygonal line formed by connecting the center points of a series of grid cells sequentially. Since the polygonal line exhibits abrupt changes in direction at the grid boundaries, it does not meet the railway design specifications for line smoothness; therefore, curve smoothing processing is required for the polygonal path. First, the planar coordinates (x, y, z) of all vertices on the polygonal line are extracted. i , y i ) and elevation value z iFor smooth horizontal alignments, a fitting method combining circular curves and transition curves can be used. Specifically, the inflection points of the broken lines are identified, and circular curves and transition curves are inserted between adjacent straight segments. The length of the transition curve is determined based on the design speed and superelevation rate. For example, for a high-speed railway with a design speed of 250 km / h, the transition curve length can be 180m to 220m; for a conventional railway with a design speed of 160 km / h, it can be 100m to 120m. The radius of the circular curve is determined based on the turning angle of the broken lines and the design speed, and is generally not less than the minimum curve radius requirement in the design specifications. After fitting, a horizontal alignment composed of straight segments, transition curves, and circular curves is obtained. The design parameters of the horizontal alignment can be output, including the coordinates of the intersection points, the curve radius, the length of the transition curve, the length of the circular curve, and the turning angle value.

[0039] For the smoothing of the longitudinal profile, ground elevation data is collected at equal intervals (e.g., 10m or 20m) along the smoothed horizontal alignment, and then the longitudinal profile slope design is carried out. The longitudinal profile consists of slope segments and vertical curves. First, the mileage and elevation of the slope change points are determined based on the trend of ground elevation changes. The line connecting adjacent slope change points is the slope segment, and its slope value should meet the maximum allowable slope requirements of seamless railways, for example, no more than 15‰ for high-speed railways. Then, vertical curves are set at the slope change points. The radius of the vertical curve is determined based on the design speed and the algebraic difference of the slope, for example, 20,000m to 25,000m for high-speed railways and 10,000m to 15,000m for ordinary railways. The vertical curves adopt the form of circular curves, and their tangent length and curve length are calculated based on the radius and the algebraic difference of the slope. After the longitudinal profile alignment design is completed, the longitudinal profile design parameters are output, including the mileage of the slope change point, the elevation of the slope change point, the slope and slope length of each slope segment, and the radius and length of each vertical curve.

[0040] After the curve smoothing process described above, the final complete railway line design parameters are output. The horizontal alignment design parameters can be directly used for the design of the line's horizontal alignment construction drawings, while the vertical alignment design parameters can be directly used for elevation control of the roadbed, bridges, and tunnels. The output parameters can be listed in tabular form, or generated as CAD script files or GIS layers, facilitating import into design software for subsequent detailed design. For example, the output results for a certain line are as follows: the horizontal alignment contains 5 intersection points with coordinates from (X1,Y1) to (X5,Y5), and the curve radii are 3200m, 2800m, 3500m, 3000m, and 4000m respectively; the vertical alignment contains 6 slope change points with mileages of DK0+000, DK1+250, DK3+780, DK7+200, DK10+500, and DK12+800, and the corresponding elevations are 125.3m, 118.7m, 121.5m, 135.2m, 128.9m, and 130.0m respectively.

[0041] The beneficial effects of this technical solution are as follows: By smoothing the curves, discrete polygonal paths are transformed into continuous horizontal and vertical alignments that conform to railway design specifications, achieving a seamless connection from gridded paths to constructable design parameters. The output horizontal and vertical alignment design parameters can be directly used in construction drawing design, avoiding the tedious work of manual adjustments and improving the efficiency of route selection and design. Simultaneously, the smoothing process considers refined design elements such as transition curves and vertical curves, ensuring the stringent requirements for track smoothness in seamless railways.

[0042] In another technical solution, when calculating the basic cost for each grid cell in S2, the following is also included: Acquire wind and sand data for the area where the grid cell is located, wherein the wind and sand data includes at least the sand transport volume and the dune movement speed; A wind and sand protection cost model is constructed, which is used to calculate the cost of wind and sand protection engineering and maintenance required during the construction and operation of the railway in the grid unit. The wind and sand protection cost is positively correlated with the sand transport volume and the dune movement speed. The cost of wind and sand protection is weighted and included in the basic cost as an additional item.

[0043] In the above technical solution, for each grid cell, wind and sand data of the area where the cell is located is additionally acquired when calculating the basic cost. This wind and sand data can be obtained from multi-year observation records of meteorological stations along the route, or extracted from regional wind and sand disaster distribution maps or remote sensing interpretation data. Specifically, the acquired wind and sand data includes at least the annual sediment transport volume and dune movement velocity. The annual sediment transport volume Q is expressed in cubic meters per meter per year (m³ / year). 3 The unit of sand transport velocity (v) is meters per year (m / a), which represents the amount of sand passing through each meter of width per year. The unit of sand dune movement velocity (v) is meters per year (m / a), which can be obtained through comparison of multiple remote sensing images or on-site monitoring. For areas lacking measured data, empirical formulas for wind and sand dynamics can be used to estimate the sand transport, such as calculating the sand transport based on average wind speed and surface vegetation cover.

[0044] When constructing a cost model for wind and sand protection, the main cost components of wind and sand protection projects are first analyzed. The wind and sand protection system for desert railways typically includes sand-blocking fences, sand-fixing checkerboard grids, sand transport channels, and sand removal and maintenance. Based on engineering practice of existing railways, the cost of protection projects is closely related to the annual sand transport volume and dune movement speed along the railway line. The annual sand transport volume Q determines the number of rows of sand-blocking fences, the area of ​​checkerboard grids, and the amount of sand-fixing materials required per unit width of railway; the dune movement speed v determines the expected effective lifespan of the protection system and the frequency and cost of sand removal and maintenance. To quantify Q and v into a dimensionless wind and sand hazard index, observational data from a typical severely wind-blown sand area along a railway line are selected as a benchmark: the benchmark value for annual sand transport volume Q0 = 30m. 3 / (m·a), taking the baseline value of dune movement speed v0=2m / a. For the measured or interpolated data Q and v of each grid cell, the dimensionless normalized annual sediment transport Q' and normalized dune movement speed v' are calculated using the following formulas: Q'= Q / Q0, v'=v / v0. Since Q and Q0 have the same unit, and v and v0 have the same unit, Q' and v' are both dimensionless ratios. When Q'≈1 and v'≈1, it indicates that the degree of wind and sand hazard in this area is comparable to that of a typical severe wind and sand area.

[0045] The cost of wind and sand protection, R1, is calculated in a linear form: R1 = w q ·Q'+w v ·v',w q and w v Let w be the weighting coefficient. q +w v = 1, w q and w v The coefficients are all set to 0.5 by default, indicating that the impact of sediment transport and dune movement speed on protection costs is equal. In engineering practice, this can be adjusted according to the dominant risk type in the local area. For example, in areas with extremely high sediment transport but slow dune movement, the weight of sediment transport can be increased. R1 is a dimensionless coefficient. It should be noted that those skilled in the art can also use other positive correlation functions, such as the product form (R1 = Q'·v'), as needed, without departing from the scope of this invention.

[0046] The calculated wind and sand protection cost R1 is added as an additional item to the base cost of the grid cell according to a set weight. The update formula for the base cost is: C' = C + k1 × R1, where C is the original value of the base cost, C' is the updated base cost, and k1 is the weight of the wind and sand protection cost (dimensionless), which can be set according to the severity of wind and sand in the engineering area. In areas with severe wind and sand hazards, such as the edge of the Sahara Desert, k1 can be set to 0.3 to 0.5; in areas with slight wind and sand impact, it can be set to 0.1 or ignored. For example, the annual sand transport of a certain grid cell is 45 m³. 3 Given a sand dune movement speed of 2.5 m / a, the normalized annual sand transport Q' = 45 / 30 = 1.5, the normalized sand dune movement speed v' = 2.5 / 2 = 1.25, and R1 = 0.5 × 1.5 + 0.5 × 1.25 = 1.375. If k1 is taken as 0.4, the increase in basic cost = 0.4 × 1.375 = 0.55, i.e., C' = C + 0.55. By incorporating wind and sand protection costs into the basic cost, the path search algorithm, when searching for the path with the minimum total cost, will tend to avoid areas with high sand transport and fast sand dune movement, or prioritize corridors with less wind and sand hazards.

[0047] The beneficial effects of this technical solution are: quantifying the cost of wind and sand protection and integrating it into the route selection cost function, enabling the route selection results to avoid areas severely affected by wind and sand disasters from the outset, reducing investment in subsequent wind and sand protection projects and maintenance costs during the operation period. By introducing benchmark values ​​and dimensionless processing, the degree of wind and sand hazard in different regions is made comparable. Linear and product forms provide two optional modeling approaches to adapt to different wind and sand environmental characteristics. This model has significant practical value for railway construction in desert regions such as North Africa, and can significantly reduce the impact of wind and sand on the operational safety of seamless railways.

[0048] In another technical solution, when calculating the basic cost for each grid cell in S2, the following is also included: Based on the cumulative function of the frequency and amplitude of the historical temperature fluctuations exceeding the preset seamless railway track temperature range, the long-term climate adaptation cost of the area where the grid cell is located is calculated. The long-term climate adaptation cost increases non-linearly with the increase of the cumulative frequency and amplitude of the exceeding events; the non-linear increase adopts the form of an exponential function or a power function. The long-term climate adaptation costs are weighted and included in the base costs as an additional item.

[0049] In the above technical solution, for each grid cell, a long-term climate adaptation cost is additionally calculated when calculating the basic cost. First, historical temperature time series data for the region where the grid cell is located is obtained. The statistical period can be set to the last 50 years, with a time resolution of year or month. The highest and lowest temperatures for each year (or month) are extracted from this series. A preset track temperature range for seamless railways is defined, denoted as [T3, T4], where T3 is the lower limit and T4 is the upper limit. For example, for North Africa, T3 = 20℃ and T4 = 30℃. For each year, its annual maximum temperature T is determined. max Does it exceed T4? If it does, record the excess magnitude ΔT. h =T max -T4; Determine the lowest annual temperature T min If it is lower than T3, record the excess amplitude ΔT. c =T3-T min In other cases, the excess is 0. The excess amounts for all years within the statistical period are summed to obtain the cumulative excess amount S = Σ(ΔT). h +ΔT c The unit is ℃. Also, count the total number of years N in which the excess events occurred, in units of times.

[0050] The long-term climate adaptation cost R2 is calculated using an exponential function, and its value increases non-linearly with the increase of the cumulative excess amount S and the excess frequency N. The specific formula is: R2 = exp(α·S + β·N), where α and β are preset coefficients, α is in units of 1 / ℃, β is in units of 1 / time, and R2 is a dimensionless coefficient. α can range from 0.01 / ℃ to 0.05 / ℃, and β can range from 0.02 / time to 0.1 / time. These coefficients can be determined based on the fitting results of existing railway operating costs and temperature exceedance data; for example, α can be 0.02 / ℃ and β can be 0.05 / time. When S=0 and N=0, R2=1, indicating no long-term climate risk. As S and N increase, R2 increases sharply according to an exponential law, reflecting the accelerating effect of cumulative damage. It should be noted that those skilled in the art can also use other non-linear increasing functions (e.g., power functions) as needed, without departing from the scope of this invention.

[0051] The calculated long-term climate adaptation cost R2 is added as an additional item to the base cost of the grid cell according to the set weights. The update formula for the base cost is: C''=C'+k2×R2, where C' is the base cost after including the wind and sand protection cost (if the wind and sand protection cost is not included, then C' is the original base cost), and k2 is the weight of the long-term climate adaptation cost, which is dimensionless and can be set according to the railway's design service life and the prediction reliability of climate change, generally between 0.2 and 0.4. For example, in a grid cell with 5 years of over-temperature events, the cumulative over-temperature total S=25℃, taking α=0.02 / ℃, β=0.05 / event, then R2=exp(0.02×25+0.05×5)= 2.117. If k2=0.3, then the increase in base cost=0.3×2.117=0.635, that is, C''=C'+0.635. The larger this increase value, the more severe the long-term climate fluctuations in the region, and the greater the cumulative impact on seamless railway operations. The path search algorithm will actively avoid such regions.

[0052] The beneficial effects of this technical solution are: it quantifies long-term climate fluctuations (the cumulative effect of frequency and amplitude) in the form of an exponential function and incorporates them into the alignment costs, enabling the alignment results to proactively mitigate the long-term risks of future climate change. The exponential function reflects the accelerating characteristics of cumulative damage, which is more consistent with the actual laws of material fatigue and operating costs than linear models. This feature complements the static temperature stress risk coefficient, together constructing a climate risk assessment system from the design baseline period to the entire life cycle.

[0053] In another technical solution, when calculating the basic cost for each grid cell in S2, the following is also included: Obtain data on the distribution of saline soil and the depth of groundwater in the area where the grid cell is located; Based on the saline soil distribution data and groundwater depth data, the salt swelling risk coefficient of the area where the grid unit is located is calculated. The salt swelling risk coefficient is proportional to the soluble salt content. When the capillary rise height is greater than the foundation depth, the risk coefficient is additionally multiplied by the ratio of the capillary rise height to the foundation depth. The salt swelling risk coefficient is weighted and included in the basic cost as an additional item.

[0054] In the above technical solution, for each grid cell, saline soil distribution data of the area where the cell is located is additionally obtained when calculating the base cost. Saline soil distribution data can be extracted from regional geological survey reports or soil salinization thematic maps, where the soluble salt content A is expressed as a percentage by mass (%), for example, A can range from 0.5% to 10%. The capillary rise height H is determined based on the soil type of the grid cell. c The capillary rise height for silt can be taken as 1.5m to 2.0m, for sandy soil as 0.5m to 1.0m, and for clay as 2.0m to 3.0m. H c The unit is meters. The foundation depth D is initially determined according to railway design standards. For roadbed engineering, the foundation depth can be taken as the difference between the shoulder elevation and the natural ground elevation, which is generally 1m to 3m. The unit of D is meters.

[0055] The risk factor R3 of salt swelling is directly proportional to the content of easily soluble salt A, and the capillary rise height H is also proportional to the risk factor R3. c When the depth is greater than the foundation depth D, the risk factor is multiplied by H. c / D (i.e., the ratio of capillary rise height to foundation depth). A preferred calculation method is to use a piecewise function: R3=A×max(1, H c / D). Where A is the content of easily soluble salts (dimensionless percentage, e.g., 5% is 0.05), H c Both A and D are in meters, and R3 is dimensionless. When A=0, R3=0, indicating no risk of salt swelling. When H... c When ≤ D, max(1, H) c / D)=1, therefore R3= A. At this point, the capillary rise height does not exceed the foundation depth, and salt is unlikely to enter the foundation area. The risk is only related to the content of easily soluble salts. When H c When >D, max(1, H) c / D)= H c / D, that is, R3=A× (H c / D), at this point, the capillary rise height exceeds the foundation depth, and salt easily accumulates in the foundation, increasing the risk with H. c / D increases linearly. For example, a certain grid cell has a soluble salt content A = 3% (i.e., 0.03%), and the soil is silt (H c Taking a depth of 1.8m, and a foundation depth of D = 1.2m, then Hc / D = 1.5 > 1, R3 = 0.03 × 1.5 = 0.045. If the region A = 8% (0.08), H c =1.8m, D=1.5m, then H c / D=1.2, R3=0.08×1.2=0.096.

[0056] The calculated salt heave risk coefficient R3 is added as an additional item to the base cost of the grid cell according to a set weight. The update formula for the base cost is: C''' = C'' + k3×R3, where C'' is the base cost after including wind and sand protection costs and long-term climate adaptability costs (if these costs are not included, then C'' is the original base cost), and k3 is the weight of the salt heave risk coefficient, which is dimensionless and can be set according to the severity of saline soil hazards in the area, generally between 0.2 and 0.5. For example, if a grid cell calculates R3=0.045, and k3=0.3, then the increase in base cost = 0.3×0.045=0.0135. By incorporating the salt heave risk coefficient into the base cost, the path search algorithm, when searching for the path with the minimum total cost, will tend to avoid areas with high soluble salt content and capillary rise height exceeding the foundation burial depth, thereby reducing the risk of uneven roadbed uplift and track irregularities caused by salt heave during operation.

[0057] It should be noted that the aforementioned wind and sand protection cost R1, long-term climate adaptability cost R2, and salt swelling risk coefficient R3 can be used independently or in combination. When used in combination, the original basic cost (i.e., the basic cost without any additional items) is added to each additional item according to its weight, i.e., the final basic cost = original basic cost + k1×R1 + k2×R2 + k3×R3, where k1, k2, and k3 are the weights of each additional item, which can be set independently according to the environmental characteristics of the engineering area. For example, in the Sahara Desert fringe of North Africa, where wind and sand hazards and saline soil problems are both prominent, the wind and sand protection cost and the salt swelling risk coefficient can be activated simultaneously, with k1=0.4 and k3=0.3 respectively, while the long-term climate adaptability cost can be selectively activated based on the reliability of climate predictions. Through the synergistic weighting of multi-dimensional risk coefficients, this method can achieve adaptive adjustment of the route selection strategy under different geographical environments. Those skilled in the art can choose to activate one, two, or all three additional items according to actual engineering needs, without departing from the scope of this invention.

[0058] Example 1 This example uses a section of a planned railway in Sudan as an illustration. The region has a tropical desert climate with annual rainfall of less than 100 mm and intense evaporation; during the hottest season, daytime high temperatures can reach over 47°C, while nighttime temperatures can drop to around 20°C, resulting in significant diurnal temperature variations; saline soils are distributed along the upstream riverbanks, and in some areas, sand dune movement rates are approximately 1-3 m / a, with annual sediment transport ranging from 10-50 m³ / year. 3 Between / (m·a).

[0059] The line begins near a town on the east bank of the Nile River (approximately 32°E, 15°N) and ends at a planned mining area about 80 km away, running along the eastern bank of the river valley. The design speed is 160 km / h, and it is classified as a conventional seamless railway.

[0060] The first step was data acquisition and gridded modeling. Topographic, geological, ecologically sensitive area, and climate data were acquired along the route. Topographic data was obtained using ASTER GDEM version 3 (30m resolution), and after resampling, the grid resolution was 20 m × 20 m. Geological data was sampled from a 1:200,000 regional geological map along the route, with soluble salt content (A) ranging from 0% to 6%, highest in the middle reaches of the alluvial-diluvial fan. The ecologically sensitive area includes an established dry valley nature reserve, covering approximately 12 km² within the study area. 2 A 200-meter buffer zone was established outward from the boundary of the protected area. Climate data were collected from meteorological observations over the past 30 years.

[0061] The foundation cost is calculated for each grid cell. The construction difficulty coefficient is calculated based on the average slope and foundation bearing capacity of the cell. The terrain of this section of the line is generally flat, with an average slope between 0° and 5°, and the slope difficulty coefficient is taken as 0.2; the foundation bearing capacity is between 80 kPa and 180 kPa, with the lowest foundation bearing capacity in the central river valley section (approximately 80-100 kPa), and the foundation difficulty coefficient is taken as 1.0. The calculated construction difficulty coefficient for the central section is 0.2 × 0.6 + 1.0 × 0.4 = 0.52. The ecological impact coefficient is determined according to the distance d from the boundary of the nature reserve, where the portion with d ≤ 50 m accounts for 2.2% of the total length, with an impact coefficient of 1.0; the portion with d between 50-200 m accounts for approximately 5%, with an impact coefficient of 0.5; and the remaining areas are assigned a coefficient of 0. The foundation cost is obtained by weighting the construction difficulty coefficient and the ecological impact coefficient in a 6:4 ratio.

[0062] Define the smoothness cost. The maximum allowable slope is preset to 20‰, and the maximum allowable turning angle to be 30° (corresponding to a minimum curve radius of approximately 300 m). The weights for elevation change cost and orientation change cost are each set to 0.5. Following the calculation methods for elevation change cost and horizontal orientation change cost, calculate the smoothness cost for each pair of adjacent grid cells.

[0063] Calculate the temperature stress risk factor. The historical highest temperature in this area is T1 = 47℃, and the historical lowest temperature is T2 = 2℃. The preset locking rail temperature range is taken near the local intermediate rail temperature, which is approximately (47+2) / 2 = 24.5℃. According to the design specifications, the locking rail temperature is taken as 25℃, and the allowable construction locking range is 20-30℃. The total excess is ΔT = (47-30) + (20-2) = 17 + 18 = 35℃. The preset sensitivity coefficient μ is taken as 0.05 / ℃, and R = 1 + 0.05 × 35 = 2.75. This coefficient is significantly greater than 1, indicating that the temperature stress risk in this area is relatively high.

[0064] Path search was performed. The weights for basic cost, smoothness cost, and temperature stress risk coefficient were set to 0.4, 0.4, and 0.2, respectively. The A* algorithm was used to search for the path with the minimum total cost. After testing and comparison, the path selected by this method was basically consistent with the manually planned reference scheme, but differences were observed in several details: In the middle reaches of the river with developed saline soil, this method automatically fine-tuned the alignment by shifting it westward by about 200-500 m to avoid areas with soluble salt content exceeding 4%; in the northern section where wind and sand are more active, the alignment tended to select areas of fixed sand dunes with slower dune movement, reducing the amount of wind and sand protection engineering required later.

[0065] All three additional risk factors (wind and sand protection, long-term climate adaptation, and salt expansion risk) were included in the calculation for comparative testing. The results show that, considering all three risk factors simultaneously, the optimal path is essentially consistent with the route generated when only temperature stress risk is considered, with only a slight increase in local alignment adjustments (an increase in offset of approximately 50-100 m). This verifies that multiple risk factors exhibit synergistic consistency in guiding the algorithm to avoid sensitive areas, and no route selection conflicts arise due to conflicting risk indicators. The total calculation time is approximately 15 minutes, which can be completed on a standard workstation, meeting the time requirements of the engineering design phase.

[0066] This embodiment demonstrates the feasibility and effectiveness of the intelligent route selection method defined in this invention in arid desert climates and saline soil environments. It can avoid the risks of temperature stress, wind and sand disasters and salt expansion from the source while ensuring the smoothness requirements of the route.

[0067] The beneficial effects of this implementation method are: quantifying the risk of salt swelling in saline soil and incorporating it into the route selection cost function, enabling the route selection results to avoid areas with severe salt swelling disasters from the source, thereby reducing subsequent foundation treatment costs and operation and maintenance expenses. By introducing the content of easily soluble salts and the ratio of capillary rise height to foundation burial depth, this model can reflect the risk differences in different saline soil regions, and has important engineering practical value for areas where saline soil is widely distributed, such as North Africa.

[0068] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and embodiments shown and described herein.

Claims

1. A railway intelligent route selection method based on multi-source data fusion, characterized in that, Includes the following steps: S1. Obtain multi-source geospatial data of the railway line area, wherein the multi-source geospatial data includes at least topographic data, geological data, ecologically sensitive area data and climate data; S2. Register and overlay the multi-source geospatial data into the same gridded geographic model, and calculate the basic cost for each grid cell. The basic cost is used to characterize the construction difficulty and ecological impact of building a railway in that grid cell. S3. Based on the geometric continuity requirements of seamless railway lines, define the smoothness cost between adjacent grid cells. The smoothness cost is the weighted sum of the elevation change cost and the planar direction change cost between adjacent grid cells. S4. Based on the historical temperature fluctuation range in the climate data and the preset seamless railway track temperature range, calculate the temperature stress risk coefficient of the area where each grid cell is located. The temperature stress risk coefficient is positively correlated with the degree to which the historical temperature fluctuation range exceeds the track temperature range. S5. Using the starting point and the ending point as constraints, a path search algorithm is used to find a path that minimizes the total cost. The total cost is the sum of the basic costs of all grid cells on the path multiplied by the first weight, plus the sum of the smoothness costs between all adjacent grid cells on the path multiplied by the second weight, plus the sum of the temperature stress risk coefficients of all grid cells on the path multiplied by the third weight.

2. The intelligent railway route selection method based on multi-source data fusion as described in claim 1, characterized in that, The elevation change cost is the ratio of the actual slope between two adjacent grid cells to the preset maximum allowable slope, where the actual slope is equal to the absolute value of the elevation difference divided by the horizontal distance between adjacent grid cells; the plane direction change cost is the ratio of the directional angle between two adjacent grid cells to the preset maximum allowable angle.

3. The intelligent railway route selection method based on multi-source data fusion as described in claim 1, characterized in that, The registration and overlay into the same gridded geographic model includes: unifying the multi-source geospatial data to the same coordinate system and the same spatial resolution, and generating attribute vectors for each grid cell through raster overlay analysis.

4. The intelligent railway route selection method based on multi-source data fusion as described in claim 1, characterized in that, The path with the minimum total cost found in S5 is output as the design parameters for the horizontal and vertical alignment of the railway line after curve smoothing.

5. The intelligent railway route selection method based on multi-source data fusion as described in claim 1, characterized in that, The calculation of the basic cost for each grid cell in S2 also includes: Acquire wind and sand data for the area where the grid cell is located, wherein the wind and sand data includes at least the sand transport volume and the dune movement speed; A wind and sand protection cost model is constructed, which is used to calculate the cost of wind and sand protection engineering and maintenance required during the construction and operation of the railway in the grid unit. The wind and sand protection cost is positively correlated with the sand transport volume and the dune movement speed. The cost of wind and sand protection is weighted and included in the basic cost as an additional item.

6. The intelligent railway route selection method based on multi-source data fusion as described in claim 1, characterized in that, The calculation of the basic cost for each grid cell in S2 also includes: Based on the cumulative function of the frequency and amplitude of the historical temperature fluctuation range exceeding the preset seamless railway track temperature range, the long-term climate adaptation cost of the area where the grid cell is located is calculated. The long-term climate adaptation cost increases nonlinearly with the increase of the cumulative frequency and amplitude of the exceeding events. The long-term climate adaptation costs are weighted and included in the base costs as an additional item.

7. The intelligent railway alignment method based on multi-source data fusion as described in claim 1, characterized in that, The calculation of the basic cost for each grid cell in S2 also includes: Obtain data on the distribution of saline soil and the depth of groundwater in the area where the grid cell is located; Based on the saline soil distribution data and groundwater depth data, the salt swelling risk coefficient of the area where the grid unit is located is calculated. The salt swelling risk coefficient is proportional to the soluble salt content. When the capillary rise height is greater than the foundation depth, the risk coefficient is additionally multiplied by the ratio of the capillary rise height to the foundation depth. The salt swelling risk coefficient is weighted and included in the basic cost as an additional item.

8. The intelligent railway route selection method based on multi-source data fusion as described in claim 1, characterized in that, The temperature stress risk factor is calculated using the following formula: R = 1 + μ·ΔT; Where R is the temperature stress risk coefficient, ΔT=max(0, T1-T4) + max(0, T3-T2), T1 is the historical highest temperature, T2 is the historical lowest temperature, [T3, T4] is the preset seamless railway locking rail temperature range, and μ is the preset sensitivity coefficient.