Road optimization design method and measurement system suitable for complex terrain

Abstract

The invention discloses a road optimization design method and a measurement system suitable for a complex terrain, and relates to the technical field of road engineering design, and the method comprises the steps: collecting data through an unmanned aerial vehicle to generate a DEM model fusing geological information, calculating a slope stability coefficient under a rainfall condition in real time through a dynamic slope safety evaluation module, and marking a reinforcement region; correcting the safe sight distance by adopting a real-time sight distance compensation model; synchronously optimizing the longitudinal slope, the horizontal curve and the earthwork volume by using a multi-target genetic algorithm; seismic working condition checking and feedback optimization are carried out through a BIM-geological disaster coupling platform, a road scheme meeting slope stability, safety and economical efficiency is finally output, a matched measurement system has the functions of self-adaptive leveling, multi-source fusion collection, edge calculation and data credibility evaluation, accuracy and reliability of data collection in a complex environment are ensured, and the method is suitable for large-scale popularization and application. According to the method, the technical bottlenecks of a traditional method in the aspects of early warning lag, insufficient sight distance, poor economy, weak environmental adaptability and the like are effectively solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of road engineering design, more particularly, to a road optimization design method and measurement system suitable for complex terrain. BACKGROUND

[0002] The road design of complex terrain, such as mountainous areas, hilly areas, cliff areas and canyon sections, needs to coordinate the three goals of geological stability, driving safety and engineering economy. Such complex terrain is generally characterized by dramatic terrain undulations, broken geological structure and variable climate environment, which leads to the risk of slope instability, loss of matric suction in the unsaturated zone caused by rainfall infiltration, and significant reduction of rock-soil shear strength. The traditional design method relies on static geological parameters and does not consider the dynamic influence of rainfall infiltration on the pore water pressure of unsaturated soil. Sudden landslides frequently occur during the construction period due to the sudden drop in slope safety factor caused by heavy rainfall. The actual sight distance in the curve area may be lower than the standard value due to the influence of terrain obstruction and dynamic obstacles such as construction vehicles and temporary facilities. The existing technology ignores the need for correction of the safety sight distance due to changes in vehicle braking acceleration, which significantly increases the risk of driving accidents on cliff sections. The GNSS positioning drift of conventional equipment in the canyon area is > 15 cm, and the laser sensor fails > 60% in rain and fog. Manual leveling takes > 5 minutes per time when the ground is uneven, which seriously restricts the efficiency and accuracy of complex terrain survey.

[0003] The existing technology for slope stability analysis does not couple real-time weather data, and cannot establish a dynamic mapping relationship between rainfall intensity and safety factor. The sight distance calculation model lacks a dynamic obstacle tracking mechanism and does not introduce vehicle kinematic parameter correction. The positioning module of the measurement equipment is single, relying only on GNSS, and the leveling response is slow. The use of mechanical leveling has poor environmental robustness and no rain and fog alternative. SUMMARY

[0004] In order to overcome the problems of the existing technology, such as the slope stability analysis not coupling real-time weather data, the inability to establish a dynamic mapping relationship between rainfall intensity and safety factor, the lack of a dynamic obstacle tracking mechanism in the sight distance calculation model, and the single positioning module of the measurement equipment relying only on GNSS, slow leveling response, the use of mechanical leveling, poor environmental robustness, and no rain and fog alternative, the present application discloses a road optimization design method and measurement system suitable for complex terrain, which can effectively solve the above technical problems.

[0005] The present application aims to provide a road optimization design method and measurement system for complex terrain.

[0006] To solve the above technical problems, the technical solution of the present application is as follows:

[0007] The application provides a road optimization design method suitable for complex terrain, comprising the following steps:

[0008] S1, collecting terrain data by a multi-spectral sensor and a laser radar carried by a drone to generate a DEM three-dimensional digital elevation model fused with geological lithology classification;

[0009] S2, based on terrain elevation, rock-soil parameters, slope direction and terrain characteristics, and surface cover data output by DEM, wherein the rock-soil parameters include rock-soil cohesion, internal friction angle, and soil bulk density, a dynamic slope safety evaluation module is used to calculate a slope stability coefficient under different rainfall conditions in real time, and when the stability coefficient is lower than a threshold value, a reinforcement area is marked, and the threshold value of the stability coefficient is 1.25;

[0010] S3, a real-time sight distance compensation model is used to correct the minimum safety sight distance in combination with a design vehicle speed, a curve radius, and a dynamic obstacle profile identified by a laser point cloud;

[0011] S4, a multi-objective genetic algorithm is used to optimize a longitudinal slope gradient, a horizontal curve radius, and a filling and excavating amount as optimization targets of slope stability, driving safety, and engineering economy, and output a road alignment primary scheme;

[0012] S5, slope stability checking and reinforcement under a seismic condition, the road alignment primary scheme output in S4 is imported into a BIM-geological disaster coupling platform, site seismic parameters are input, a Newmark slider displacement method is used to calculate a permanent displacement of the slope under the action of the earthquake, when the displacement is greater than 30 cm, an anti-seismic reinforcement scheme is generated and feedback optimization is performed on the scheme;

[0013] S6, outputting a final road alignment scheme meeting the requirements of slope stability under rainfall and seismic conditions, driving safety, and economy.

[0014] Preferably, characterized in that the dynamic slope safety evaluation module performs the following steps:

[0015] S201, inputting rock-soil cohesion, internal friction angle, soil bulk density derived from DEM, and real-time monitored rainfall intensity data, constructing a finite element (FEM) or finite difference (FDM) numerical model of the slope based on DEM, and discretizing the slope rock-soil into unit grids;

[0016] S202, using real-time rainfall intensity as a boundary condition on the model, using an iterative calculation with a time step ≤10 minutes to drive a non-saturated soil seepage-stress coupling model to simulate, and based on a van Genuchten model, determining a matrix suction-saturation relationship of each unit grid in each time step:

[0017]

[0018] wherein, is the volumetric saturation, is the residual saturation, is the maximum saturation, Y is the matrix suction, and parameters a, n, m are calibrated by geotechnical tests; the pore water pressure distribution field inside the slope at each time step is dynamically calculated by the model;

[0019] S203, at the end of each time step simulation, based on the pore water pressure distribution field of the current step, the stability coefficient Fs of the slope is calculated by using the limit equilibrium method; the stability coefficient Fs is the ratio of the total anti-sliding force to the total sliding force on the potential sliding surface; wherein, the total sliding force is mainly determined by the self-weight of the sliding soil, and the total anti-sliding force is calculated according to the Mohr-Coulomb criterion and directly depends on the pore water pressure value output by step S202, and the pore water pressure value is the absolute value of the matrix suction Y; the numerical change of Fs is monitored in real time;

[0020] S204, when it is monitored that the calculated stability coefficient Fs is less than 1.25 for 3 consecutive hours, it is determined that the slope is in a continuous instability risk, and the module automatically generates a drainage ditch or anchor rod reinforcement scheme, and marks the corresponding area as a reinforcement area.

[0021] Preferably, the real-time visual range compensation model comprises:

[0022] The millimeter wave radar is used to replace the laser radar to scan the position, speed and heading angle of the construction vehicle, and the dynamic obstacle area is projected in the DEM;

[0023] The basic visual range formula is corrected based on the vehicle braking acceleration:

[0024] Corrected visual range = standard visual range × (1 + acceleration / braking coefficient)

[0025] Wherein, the braking coefficient is dynamically adjusted according to the road adhesion coefficient:

[0026] Braking coefficient = reference value × (1 - 0.2 × road wetness index)

[0027] The wetness index is determined by the surface water content determined by the multispectral sensor, and the value range is 0.8-1.2;

[0028] When the corrected visual range is lower than 120% of the standard value, the curve widening instruction is triggered.

[0029] Preferably, the multi-objective genetic algorithm is used for synchronous optimization in step S4, which specifically includes the following steps:

[0030] S401, automatically generating or receiving a road longitudinal slope gradient sequence and a horizontal curve radius sequence preset by a designer according to DEM terrain features as optimization variables, generating an initial population in a real number coding mode, and each chromosome of the individual represents a potential road alignment scheme;

[0031] S402, before calculating the fitness of each scheme, the following rigid constraints are applied to each individual in the population:

[0032] (a) the road longitudinal slope gradient value, for general mountainous road sections, the gradient value ≤ 8%; for cliff road sections, the gradient value ≤ 5%;

[0033] (b) the balance ratio of filling and excavation, the ratio of total filling to total excavation should satisfy ∈ [0.9, 1.1];

[0034] S403, the fitness value F of each individual scheme in the population is calculated to evaluate the pros and cons of the scheme, and the fitness function F is defined as:

[0035]

[0036] is the total earthwork cost of the scheme, which is calculated by the filling and excavation amount and the unit earthwork cost, is the slope penalty term, slope is the absolute value of the percentage of the longitudinal slope of the section, this term is used to exponentially punish the scheme with excessive slope value, sight is the sight compliance rate of the scheme, which is calculated by the real-time sight compensation model in step S3, 、 、 is the weight coefficient, which corresponds to the relative importance of the three optimization objectives of earthwork economy, longitudinal slope flatness and driving safety respectively, and its value is dynamically calculated and allocated according to the distribution range of each index in the current population by the entropy weight method;

[0037] S404, selection, crossover and mutation operations are performed on the population to generate a new generation of population, and steps S402-S403 are repeated for iterative optimization until the maximum number of iterations or the fitness function converges;

[0038] S405, the individual with the optimal fitness value F is selected from the final generation of population, and the decoded output is the primary road alignment scheme.

[0039] Preferably, the road optimization design method suitable for complex terrain step S5 specifically comprises:

[0040] S501, import the road alignment primary scheme output by S4, the DEM, and the geological lithology classification data into a BIM-geological disaster coupling analysis platform to construct an integrated coupling model including a road BIM model and a geological body three-dimensional model;

[0041] S502, input site ground motion parameters as seismic loads, the site ground motion parameters being based on the peak ground acceleration (PGA) and characteristic seismic wave spectrum recorded in the national authoritative standards and the seismic safety evaluation report of the region where the project is located;

[0042] S503, in the coupling analysis platform, the Newmark slider displacement method is used to calculate the permanent displacement of the reinforced area and the slope of the cliff road section marked in step S2 under the seismic load; the calculation includes:

[0043] Based on the cohesion, internal friction angle, and pore water pressure of the rock and soil body, the critical acceleration of the slope on different potential sliding surfaces is calculated by the limit equilibrium method; the critical acceleration is the minimum horizontal acceleration when the slope stability coefficient Fs is exactly equal to 1.0;

[0044] The input seismic wave acceleration time history is double-integrated; only when the seismic wave acceleration exceeds the calculated critical acceleration, the acceleration exceeding the critical acceleration is integrated to calculate the cumulative velocity and displacement of the slider;

[0045] S504, when the maximum permanent displacement calculated is greater than 30 cm, it is determined that the slope has a high risk of instability under the seismic working condition, the platform automatically matches and generates a targeted anti-seismic reinforcement scheme from a pre-set anchor reinforcement scheme library, and feeds back the reinforcement engineering quantity and constraint conditions to the multi-objective genetic algorithm of S4 for a new round of optimization until all working condition requirements are met.

[0046] Preferably, a road measurement system suitable for complex terrain comprises:

[0047] An adaptive leveling leg system: including three independent servo hydraulic legs, each leg having a built-in displacement sensor and adjusting the extension and retraction amount in real time through a PID controller, so that the inclination of the system platform is ≤0.5°;

[0048] A multi-source fusion acquisition module: integrating a dual-frequency RTK GNSS receiver, a millimeter wave radar, and an electronic compass, wherein the millimeter wave radar collects terrain data in rainy and foggy weather;

[0049] An edge computing unit: running a slope stability coefficient algorithm and a line-of-sight compensation calculation in real time;

[0050] A data credibility label generation module for evaluating the credibility of the measurement results according to real-time sensor data and triggering data reacquisition when the credibility is lower than a threshold.

[0051] Preferably, the adaptive leveling leg system comprises:

[0052] Each leg tip is equipped with a conical screw anchor, whose screwing depth is self-adaptively adjusted by soil resistance; the soil resistance is obtained by real-time monitoring of the current value of the servo motor driving the conical screw anchor, and when the current value reaches a preset threshold, it is determined that the screwing depth meets the requirements.

[0053] The PID controller completes platform leveling within 3 seconds according to three-axis accelerometer data.

[0054] Preferably, the multi-source fusion acquisition module further comprises:

[0055] The RTK FNSS-UWB dual-mode positioning switching unit realizes centimeter-level positioning through the pre-deployed UWB base station when the GNSS signal is lost.

[0056] The millimeter wave radar is equipped with an active temperature control cover containing a semiconductor refrigeration piece and a temperature control microcontroller. When the millimeter wave radar temperature is > 50℃, forced air cooling is started, and when the millimeter wave radar temperature is <-10℃, resistance heating is started.

[0057] Preferably, the edge computing unit is configured to perform the following operations:

[0058] The sliding window RANSAC algorithm is used to process the point cloud data collected by the millimeter wave radar to extract the terrain feature line.

[0059] After the precise algorithm is used to calculate the slope stability coefficient and times out, it is switched to an emergency calculation mode, and a pre-trained convolutional neural network model (CNN) is called to quickly estimate the stability coefficient.

[0060] The CNN takes the slope height, slope angle, and rock-soil properties as input and outputs an approximate stability coefficient value with an error of no more than 5% compared to the precise solution.

[0061] Preferably, a road measurement system suitable for complex terrain further comprises:

[0062] The data reliability label generation module calculates the reliability according to the following formula:

[0063]

[0064] The GNSS signal-to-noise ratio is directly obtained from the raw data output of the GNSS receiver, the point cloud density is obtained by real-time statistics of each frame of radar point cloud data by the edge computing unit, and the temperature offset coefficient is calculated according to the deviation of the millimeter wave radar temperature sensor reading from the preset ideal temperature.

[0065] When the reliability is < 0.6, the data reacquisition is automatically triggered.

[0066] The beneficial effects of the present application are:

[0067] The present application significantly improves the safety and economy of complex terrain road engineering through the collaborative innovation of a dynamic slope safety evaluation module, a real-time visibility compensation model, a multi-objective genetic algorithm, a BIM-geological disaster coupling platform, and an intelligent measurement system. The dynamic slope safety evaluation module uses a non-saturated soil seepage-stress coupling model, accurately quantifies the influence of rainfall infiltration on matrix suction based on the van Genuchten equation, and sets the stability coefficient being less than 1.25 for 3 consecutive hours as the reinforcement trigger threshold, solving the defect of large prediction error of the safety factor of traditional static models and improving the landslide warning response speed. The real-time visibility compensation model integrates millimeter wave radar dynamic obstacle tracking and brake acceleration correction formula, and through global evaluation of the visibility compliance rate, it ensures that the actual visibility of cliff sections always maintains more than 120% of the standard value.

[0068] The multi-objective genetic algorithm dynamically allocates weights using the entropy weight method, and under the constraints of the fill-dig ratio ∈ [0.9, 1.1] and longitudinal slope classification control (mountain ≤ 8% / cliff ≤ 5%), it realizes the reduction of earthwork cost. The BIM-geological disaster coupling platform simulates the seismic displacement field through the Newmark slider analysis method, and when the permanent displacement exceeds 30 cm, it automatically matches the anchor reinforcement scheme, which can compress the disaster treatment scheme generation time and improve the accuracy of anchor selection. The seamless integration of conventional design and disaster prevention design on one BIM platform realizes the transition from the manual iteration mode of "design-checking-modification" to the integrated mode of "design-automatic checking-intelligent reinforcement", improving efficiency and reliability.

[0069] The supporting measurement system, with a PID hydraulic leveling mechanism, RTK-UWB dual-mode positioning, and temperature-controlled millimeter wave radar, can work all-weather from -20°C to 65°C, solving the problems of low efficiency and low data integrity in rain and fog. The edge computing unit uses a 3-layer convolutional neural network to output an approximate solution of the slope stability coefficient, reducing the calculation time from 10 minutes to 5 seconds with an error of ≤5%, and dynamically triggering re-collection combined with data reliability labels, reducing the rate of invalid data.

[0070] In summary, the present application breaks through the technical bottlenecks of geological instability warning lag, insufficient visibility safety redundancy, rigid economic constraints, and weak adaptability to surveying environments in complex terrain road engineering, providing reliable technical support for mountain, cliff highway, railway, and emergency passage construction. BRIEF DESCRIPTION OF DRAWINGS

[0071] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only exemplary, and for those skilled in the field, other drawings can be obtained from the provided drawings without creative labor. Figure 1 The overall flowchart of the road optimization design method suitable for complex terrain of the present application;

[0072] Figure 2 The specific implementation flowchart of the dynamic slope safety evaluation module in the road optimization design method suitable for complex terrain provided by the embodiment 1 of the present application;

[0073] Figure 3 The specific implementation flowchart of the multi-objective genetic algorithm optimization in the road optimization design method suitable for complex terrain provided by the embodiment 1 of the present application;

[0074] Figure 4 The specific implementation flowchart of the slope stability checking and reinforcement in the earthquake working condition in the road optimization design method suitable for complex terrain provided by the embodiment 1 of the present application;

[0075] Figure 5 The overall flowchart of the road measurement system suitable for complex terrain provided by the embodiment 2 of the present application. DETAILED DESCRIPTION

[0076] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined application purposes, the exemplary embodiments will be described in detail, and the examples are shown in the drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all the embodiments consistent with the present application. Instead, they are only examples of methods and systems consistent with some aspects of the present application as detailed in the appended claims.

[0077] The terms used in the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "an" and "the" used in the present application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein means and includes any or all possible combinations of one or more associated listed items.

[0078] The specific embodiments, features and effects according to the present application will be described in detail below in combination with the drawings and preferred embodiments.

[0079] Embodiment 1

[0080] Referring to Figure 1 The embodiment provides a road optimization design method suitable for complex terrain, including the following steps:

[0081] S1, collecting terrain data by a multi-spectral sensor and a laser radar carried by a drone to generate a DEM three-dimensional digital elevation model fused with geological lithology classification;

[0082] S2, based on terrain elevation, rock-soil parameters, slope direction and terrain characteristics, and surface cover data output by DEM, wherein the rock-soil parameters include cohesion of rock-soil body, internal friction angle, and soil bulk density, a dynamic slope safety evaluation module is used to calculate a slope stability coefficient under different rainfall conditions in real time, and when the stability coefficient is lower than a threshold value, a reinforcement area is marked, and the threshold value of the stability coefficient is 1.25;

[0083] S3, a real-time sight distance compensation model is used to correct the minimum safety sight distance in combination with a design vehicle speed, a curve radius, and a dynamic obstacle profile identified by laser point cloud;

[0084] S4, a multi-objective genetic algorithm is used to take slope stability, driving safety, and engineering economy as optimization targets to simultaneously optimize longitudinal slope gradient, horizontal curve radius, and fill and cut volume to output a road alignment primary scheme;

[0085] S5, slope stability checking and reinforcement under seismic conditions, the road alignment primary scheme output in S4 is imported into a BIM-geological disaster coupling platform, site seismic parameters are input, and a Newmark slider displacement method is used to calculate permanent displacement of the slope under the action of the earthquake, when the displacement is greater than 30 cm, an anti-seismic reinforcement scheme is generated and feedback optimization is performed on the scheme;

[0086] S6, outputting a final road alignment scheme that meets the requirements of slope stability under rainfall conditions and seismic conditions, driving safety, and economy.

[0087] Further referring to Figure 2 The dynamic slope safety evaluation module performs the following steps:

[0088] S201, inputting cohesion of rock-soil body, internal friction angle, soil bulk density derived from DEM, and real-time monitored rainfall intensity data, constructing a finite element (FEM) or finite difference (FDM) numerical model of the slope based on DEM, and discretizing the rock-soil body of the slope into unit grids;

[0089] S202, using real-time rainfall intensity as the upper boundary condition of the model, using an iterative calculation with a time step ≤10 minutes to drive a non-saturated soil seepage-stress coupling model to simulate, and at each time step, determining the matrix suction-saturation relationship of each unit grid based on a van Genuchten model:

[0090]

[0091] in, is the volume saturation, is the residual saturation, is the maximum saturation, Y is the matrix suction, and the parameters a, n, and m are calibrated by geotechnical tests. The pore water pressure distribution field inside the slope is dynamically calculated at each time step through this model.

[0092] S203, after each time step simulation is completed, the slope stability factor Fs is calculated using the limit equilibrium method based on the pore water pressure distribution field of the current step; the stability factor Fs is the ratio of the total anti-sliding force to the total sliding force on the potential sliding surface; the total sliding force is mainly determined by the deadweight of the sliding soil, and the total anti-sliding force is calculated according to the Mohr-Coulomb criterion and directly depends on the pore water pressure value output in step S202, which is the absolute value of the matrix suction Y; the value of Fs is monitored in real time;

[0093] S204: When the calculated stability coefficient Fs is monitored to be less than 1.25 for three consecutive hours, the slope is determined to be at risk of continuous instability. The module automatically generates a drainage ditch or anchor reinforcement plan and marks the corresponding area as a reinforcement area.

[0094] Furthermore, the real-time sight distance compensation model includes: scanning the position, speed, and heading angle of the construction vehicle by millimeter wave radar, and projecting the dynamic obstacle area in the DEM;

[0095] Correction of basic sight distance formula based on vehicle braking acceleration:

[0096] Corrected sight distance = standard sight distance × (1 + acceleration / braking coefficient)

[0097] The braking coefficient is dynamically adjusted according to the road adhesion coefficient:

[0098] Braking coefficient = reference value × (1 - 0.2 × road slip index)

[0099] The anti-skid coefficient test of the anti-skid performance test released by the Guangzhou National Inspection Center shows that the corresponding saturated water-containing road surface ( =25%), the adhesion coefficient is reduced to 80% of the dry road surface. In the Australian standard AS / NZS4586 anti-skid performance test grade assessment requirements, dry road surfaces (w<5%) are allowed to exceed the benchmark value by 20% due to the enhanced adhesion of micro-texture. Therefore, the wet skid index value range is 0.8~1.2;

[0100] When the corrected sight distance is lower than 120% of the standard value, the curve widening instruction is triggered.

[0101] Preferably, please refer toFigure 3 In step S4, a multi-objective genetic algorithm is used for synchronous optimization, which specifically includes the following steps:

[0102] S401, automatically generate according to DEM terrain features, at the same time, receive the road longitudinal slope gradient sequence and horizontal curve radius sequence preset by the designer as optimization variables. The designer can give a very rough route direction, such as the starting point, the ending point and several approximate control points, or completely initialize randomly by the algorithm. Based on the above input, the algorithm will randomly generate multiple sets of "longitudinal slope gradient sequence" and "horizontal curve radius sequence". For example, the first set of sequences may be [slope: 2%, 4%, 1%] and [radius: 300m, 200m], and the second set of sequences may be [slope: 5%, 3%, 0.5%] and [radius: 250m, 350m]. These initial sequences are usually of poor quality, but they constitute the starting point of the search.

[0103] For each sequence, the algorithm will fit a continuous three-dimensional road curve on the DEM. Then, this road curve will be sent for earthwork calculation, for slope stability analysis, for line-of-sight analysis, and for comprehensive analysis of all these results to calculate the fitness value F of this sequence, i.e., the fitness value of this scheme. An initial population is generated using real number encoding, and each individual chromosome represents a potential road alignment scheme;

[0104]

[0105] S402, before calculating the fitness of each scheme, the following rigid constraints are applied to each individual in the population:

[0106] (a) Road longitudinal slope gradient value: for general mountainous road sections, the slope value ≤ 8%; for cliff road sections, the slope value ≤ 5%;

[0107] (b) Balance ratio of filling and excavation volume: the ratio of total filling volume to total excavation volume should satisfy ∈ [0.9, 1.1];

[0108] The constraint conditions of the multi-objective genetic algorithm are set according to the "Design Speed of 40 km / h Mountainous Area of Three-level Highway, Maximum Longitudinal Slope Not Exceeding 8%; Continuous Long and Steep Downhill Road Section (Longitudinal Slope > 5%) Should Set Emergency Lane" clause in "Highway Traffic Safety Facilities Design Specification" (JTG D81-2017). The mountain road longitudinal slope gradient is ≤ 8%, which meets the national standard bottom line, or the cliff road section is ≤ 5%, which actively improves the safety redundancy for cliff scenarios.

[0109] When the filling and excavation volume ratio is 1, it represents the ideal state of earthwork balance. The filling and excavation volume ratio error is set to ±10%. When the filling and excavation volume ratio When the filling and excavation volume ratio ​

[0110] If a scheme does not meet the steps a and b, the scheme is directly determined as invalid individual, and does not participate in subsequent fitness calculation and genetic operation.

[0111] S403, the fitness value F of each individual scheme in the population is calculated, and the pros and cons of the scheme are evaluated by the fitness value, and the fitness function F is defined as:

[0112]

[0113] The total earthwork cost of the scheme is calculated by the amount of filling and excavation and the unit earthwork cost, The slope penalty term is the absolute value of the slope percentage of the section, which is used to exponentially punish the scheme with excessive slope value, and the sight is the sight compliance rate of the scheme, which is calculated by the real-time sight compensation model in step S3, 、 、 The weight coefficient corresponds to the relative importance of the three optimization objectives of earthwork economy, longitudinal slope flatness and driving safety, respectively, and the value is dynamically calculated and allocated by the entropy weight method according to the distribution range of each index in the current population, and in all curves of the road section, After the calculation of the real-time sight compensation model in step S3, the number of curves that meet the modified sight≥design specification value×120% accounts for the proportion, which quantifies the sight safety level of the whole line. In the early stage of optimization, the sight of all schemes may be poor, and the entropy is small and the weight is high. The algorithm will prioritize optimizing the sight. After the sight is up to standard, the sight difference becomes smaller, the entropy becomes larger, and the weight decreases. The algorithm will focus on optimizing the earthwork cost or slope, realizing the automatic adjustment of the focus in different optimization stages;

[0114] S404, selection, crossover and mutation operations are performed on the population to generate a new generation of population, and steps S402-S403 are repeated for iterative optimization until the maximum iteration number or the fitness function converges;

[0115] S405, select the individual with the optimal fitness value F from the final generation of population, and output the decoded road alignment primary scheme.

[0116] Further, please refer to Figure 4 , step S5 specifically includes:

[0117] S501, the road alignment primary scheme output by S4, that is, the final road center line, longitudinal slope, cross section and other geometric data, and DEM and geological lithology classification data are imported into the BIM-geological disaster coupling analysis platform to construct an integrated coupling model including the road BIM model and the geological body three-dimensional model;

[0118] S502, according to the national authority standard, such as "China ground motion parameter zoning map" and the earthquake safety evaluation report of the engineering site, determine the peak acceleration PGA and the seismic wave spectrum, and input the peak ground acceleration PGA and the characteristic seismic wave spectrum required by the design as the seismic load;

[0119] S503, in the coupling analysis platform, the reinforced area marked in step S2 and the slope of the cliff road section are calculated by Newmark slider displacement method under the permanent displacement of the seismic load; the calculation includes:

[0120] In the coupling platform, based on the cohesion, internal friction angle and pore water pressure value of the rock and soil mass, the critical acceleration of the slope on different potential sliding surfaces is calculated by iterative calculation of limit equilibrium method. Through iterative calculation, a hypothetical horizontal acceleration is applied until the sliding surface that makes the stability coefficient Fs=1 is searched, and the horizontal acceleration applied at this time is the critical acceleration of the sliding surface.

[0121] The entire seismic wave time history is input into the model, and the system reads the seismic wave acceleration value at each time. Only when the seismic wave acceleration value is greater than the critical acceleration of the sliding surface, the slope is considered to start plastic deformation sliding. The excess acceleration is integrated once to obtain the sliding velocity of the slider. The velocity is integrated again to obtain the cumulative permanent displacement in the time period. After the end of the entire seismic process, all the displacements generated in the sliding time period are accumulated to obtain the total permanent displacement of the potential sliding surface.

[0122] S504, in the engineering sense, the displacement is between 15-30cm, which may cause cracks and local damage, and needs attention. When the displacement is more than 30cm, it usually means that the slope has occurred substantial, non-negligible sliding, which may cause serious damage or failure of structures such as roadbed and retaining wall, and must be reinforced. Therefore, when the maximum permanent displacement calculated is greater than 30cm, it is determined that the slope has a high risk of instability under the seismic working condition, the platform automatically matches and generates the targeted seismic reinforcement scheme from the pre-set anchor reinforcement scheme library, and feeds back the reinforcement engineering quantity and constraint conditions to the multi-objective genetic algorithm of S4 for a new round of optimization until all working condition requirements are met.

[0123] Example 2

[0124] Please refer to Figure 5 A road measurement system suitable for complex terrain comprises:

[0125] The adaptive leveling leg system comprises three independent servo hydraulic legs, each of which is provided with a displacement sensor and a PID controller for real-time adjustment of the extension and retraction amount, so that the inclination of the system platform is less than or equal to 0.5°, and the platform itself is ensured not to be tilted, and all the collected terrain data is ensured not to have an angle deviation as far as possible.

[0126] The multi-source fusion acquisition module integrates a dual-frequency RTK GNSS receiver, a millimeter wave radar and an electronic compass, and the millimeter wave radar is used to collect terrain data in rainy and foggy weather.

[0127] The edge computing unit is used for real-time running of a slope stability coefficient algorithm and a line-of-sight compensation calculation.

[0128] The data credibility label generation module is used for evaluating the credibility of a measurement result according to real-time sensor data, and triggering data reacquisition when the credibility is lower than a threshold value.

[0129] Further, the adaptive leveling leg system comprises: a conical spiral ground spike is installed at the end of each leg, and the rotation depth is adaptively adjusted through soil resistance.

[0130] The soil resistance is obtained by real-time monitoring of the current value of the servo motor driving the conical spiral ground spike, and when the current value reaches a preset threshold value, it is determined that the rotation depth meets the requirements; the adaptive adjustment process of the conical spiral ground spike is that the servo motor at the end of the leg drives the ground spike to rotate and descend, the PID controller real-time monitors the driving current of the servo motor, the driving current is proportional to the output torque of the motor, and then reflects the soil resistance when the ground spike is rotated, the PID controller compares the real-time current value with a preset current threshold value representing sufficient grip, in hard ground, the current value quickly reaches the threshold value, and the motor stops working; in soft ground, the initial current value is low, the controller controls the motor to continuously rotate, so that the ground spike is continuously rotated deeper, until the driving current rises to the preset threshold value, so that stable support force can be obtained in different soil.

[0131] The three-axis accelerometer is used for real-time sensing of the current inclination angle of the platform, the PID controller receives the signal of the accelerometer, calculates the error between the current inclination angle and the target level, and then issues an instruction to drive the leg to act according to the proportional, integral and differential algorithm, and the PID controller completes the platform leveling within 3 seconds according to the three-axis accelerometer data, and quickly eliminates the inclination error.

[0132] Further, the multi-source fusion acquisition module further comprises an RTK FNSS-UWB dual-mode positioning switching unit. In an open outdoor area, the system uses RTK GNSS to obtain a centimeter-level absolute position. In places such as deep valleys, dense forests, tunnel entrances, and under bridges where GNSS signals are completely lost or severely deteriorated, RTK GNSS will fail. At this time, the system automatically switches to the UWB mode. Through pre-deployed UWB base stations, the UWB tags on the system can calculate their positions relative to the base stations by measuring the time of flight of wireless signals between the tags and the base stations, thereby obtaining centimeter-level positioning and generating point clouds with accurate geographic coordinates for the S1 step.

[0133] When the millimeter wave radar transmits electromagnetic waves of a specific frequency, the transmission frequency will drift with temperature changes. Therefore, the millimeter wave radar is equipped with an active temperature control cover containing a semiconductor refrigeration sheet and a temperature control microcontroller. When the temperature of the millimeter wave radar is greater than 50°C, forced air cooling is started. When the temperature of the millimeter wave radar is less than -10°C, resistance heating is enabled.

[0134] Further, the edge computing unit is the intelligent processing core of the system, which carries an optimized algorithm software package for real-time processing of sensor data and execution of key calculations. Its workflow includes two main stages: terrain feature extraction and intelligent calculation of slope stability coefficient. The edge computing unit receives raw laser point cloud data from the millimeter wave radar, which consists of millions of three-dimensional coordinate points containing terrain, vegetation, artificial objects, and other information. The system uses a sliding window RANSAC algorithm to accurately extract feature lines representing terrain undulations such as ridge lines, valley lines, slope top lines, and slope foot lines. After obtaining the terrain feature lines and calculating the slope geometric parameters, the edge computing unit starts the slope stability coefficient calculation process.

[0135] After the precise algorithm calculation of the slope stability coefficient times out, it switches to an emergency calculation mode and calls a pre-trained lightweight convolutional neural network model (CNN) for rapid estimation. When the emergency mode is started, the edge computing unit extracts the prepared input parameters: slope height, slope angle, and rock-soil type from the current calculation task. The parameters are input into the loaded lightweight CNN model. After forward propagation calculation, the CNN model outputs an approximate stability coefficient value with an error of no more than 5% from the precise solution.

[0136] Further, the data credibility label generation module calculates the credibility according to the following formula:

[0137]

[0138] The GNSS signal-to-noise ratio is directly derived from the raw data output of the GNSS receiver, and the point cloud density is obtained by real-time statistics of each frame of radar point cloud data by the edge computing unit; and the temperature offset coefficient is calculated according to the deviation of the millimeter wave radar temperature sensor reading from the preset ideal temperature.

[0139] When the credibility is less than 0.6, the module automatically sends an instruction to the system to discard the current batch of data and controls the acquisition module to re-scan and collect the current area. This process is automated and does not require human intervention. Ultimately, all credible data and results generate reports and charts.

[0140] The above is only a preferred embodiment of the present application, and is not intended to limit the present application in any form. Although the present application has been disclosed as above with a preferred embodiment, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content without departing from the scope of the technical solution of the present application, and any equivalent embodiments with equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present application, and any equivalent changes and modifications made to the above embodiments without departing from the technical solution of the present

Claims

1. A road optimization design method suitable for complex terrain, characterized by: The steps include: S1, using drones equipped with multispectral sensors and lidar to collect terrain data and generate a three-dimensional digital elevation model (DEM) that integrates geological and lithological classification; S2, based on the terrain elevation, geotechnical parameters, slope and aspect, terrain characteristics, and surface cover data output by the DEM. Geotechnical parameters include geotechnical cohesion, internal friction angle, and soil bulk density. The dynamic slope safety assessment module is used to calculate the slope stability coefficient under different rainfall conditions in real time. When the stability coefficient falls below a threshold, the reinforcement area is marked. The threshold of the stability coefficient is specifically 1.

25. S3 uses a real-time sight distance compensation model to correct the minimum safe sight distance based on the design vehicle speed, curve radius, and dynamic obstacle contours identified by laser point cloud recognition; S4, using a multi-objective genetic algorithm, with slope stability, driving safety, and engineering economy as optimization goals, simultaneously optimizes the longitudinal slope, horizontal curve radius, and fill and cut volume, and outputs a preliminary road alignment plan; S5, verification and reinforcement of slope stability under earthquake conditions: import the preliminary road alignment plan output from S4 into the BIM-geological disaster coupling platform, input the site seismic parameters, and use the Newmark slider displacement method to calculate the permanent displacement of the slope under earthquake. When the displacement is greater than 30cm, a seismic reinforcement plan is generated and feedback optimization is performed on the plan; S6, output the final road alignment plan that meets the slope stability requirements, driving safety and economy under both rainfall and earthquake conditions.

2. The road optimization design method suitable for complex terrain according to claim 1, characterized in that: The dynamic slope safety assessment module performs the following steps: S201, inputting the rock and soil cohesion, internal friction angle, soil bulk density and real-time monitored rainfall intensity data derived from the DEM; constructing a finite element (FEM) or finite difference (FDM) numerical model of the slope based on the DEM, and discretizing the slope rock and soil into unit grids; S202, using the real-time rainfall intensity as the upper boundary condition of the model, an iterative calculation with a time step of ≤10 minutes is performed to drive the unsaturated soil seepage-stress coupling model for simulation. Within each time step, the relationship between the matrix suction and saturation of each unit grid is determined based on the van Genuchten model: ； in, is the volume saturation, is the residual saturation, is the maximum saturation, Y is the matrix suction, and the parameters a, n, and m are calibrated by geotechnical tests. The pore water pressure distribution field inside the slope is dynamically calculated at each time step through this model. S203: After each time step simulation is completed, the slope stability factor Fs is calculated using the limit equilibrium method based on the pore water pressure distribution field of the current step. The stability factor Fs is the ratio of the total anti-sliding force to the total sliding force on the potential sliding surface. The total sliding force is primarily determined by the deadweight of the sliding soil. The total anti-sliding force is calculated according to the Mohr-Coulomb criterion and directly depends on the pore water pressure value output in step S202. The pore water pressure value is the absolute value of the matrix suction Y. The numerical changes of Fs are monitored in real time. S204: When the calculated stability coefficient Fs is monitored to be less than 1.25 for three consecutive hours, the slope is determined to be at risk of continuous instability. The module automatically generates a drainage ditch or anchor reinforcement plan and marks the corresponding area as a reinforcement area.

3. The road optimization design method suitable for complex terrain according to claim 1, characterized in that: The real-time line-of-sight compensation model includes: Use millimeter-wave radar to scan the position, speed, and heading angle of construction vehicles and project dynamic obstacle areas into the DEM; Correction of basic sight distance formula based on vehicle braking acceleration: Corrected sight distance = standard sight distance × (1 + acceleration / braking coefficient) The braking coefficient is dynamically adjusted according to the road adhesion coefficient: Braking coefficient = reference value × (1 - 0.2 × road slip index) The slippery index is determined by inverting the surface moisture content using a multispectral sensor, with a value range of 0.8 to 1.2; When the corrected sight distance is lower than 120% of the standard value, the curve widening instruction is triggered.

4. The road optimization design method applicable to complex terrain according to claim 1, characterized in that: In step S4, a multi-objective genetic algorithm is used to perform synchronous optimization, which specifically includes the following steps: S401, automatically generating a road longitudinal slope sequence and a horizontal curve radius sequence preset by a designer based on the DEM terrain features or receiving them as optimization variables, generating an initial population using a real number encoding method, where each individual chromosome represents a potential road alignment scheme; S402, before calculating the fitness of each solution, impose the following rigid constraints on each individual in the population: (a) Road longitudinal slope: for general mountainous sections, the slope should be ≤ 8%; for sections near cliffs, the slope should be ≤ 5%; (b) Fill-cut balance ratio: the ratio of total fill volume to total cut volume must satisfy ∈ [0.9, 1.1]; S403, calculate the fitness value F of each individual solution in the population, and use the fitness value to evaluate the quality of the solution. The fitness function F is defined as: ； is the total earthwork cost of the scheme, which is calculated from the filling and excavation volume and the unit earthwork cost. is the slope penalty term, slope is the absolute value of the longitudinal slope percentage of the road section, which is used to exponentially penalize the scheme with too large a slope value, and sight is the sight distance compliance rate of the scheme, which is calculated by the real-time sight distance compensation model in step S3 for the entire scheme. 、 、 is the weight coefficient, which corresponds to the relative importance of the three optimization objectives of earthwork economy, longitudinal slope gentleness, and driving safety. Its value is dynamically calculated and allocated according to the distribution range of various indicators in the current population through the entropy weight method; S404, performing selection, crossover, and mutation operations on the population to generate a new generation of population, and repeating steps S402-S403 for iterative optimization until the maximum number of iterations is reached or the fitness function converges; S405: Select the individual with the best fitness value F from the final generation population, and output it as a primary road alignment plan after decoding.

5. The road optimization design method suitable for complex terrain according to claim 1, characterized in that: The step S5 specifically includes: S501, importing the road alignment primary plan outputted in S4 together with the DEM and geological lithology classification data into the BIM-geological hazard coupling analysis platform to construct an integrated coupling model including the road BIM model and the geological body three-dimensional model; S502: Input site seismic parameters as seismic loads. Site seismic parameters are peak ground acceleration and characteristic seismic wave spectrum recorded in national authoritative standards and in the seismic safety assessment report for the project area. S503: In the coupled analysis platform, the permanent displacement of the reinforced area and the cliff-facing road slope marked in step S2 is calculated using the Newmark slider displacement method under seismic loads. The calculation includes: Based on the cohesion, internal friction angle and pore water pressure of the rock mass, the critical acceleration of the slope on different potential sliding surfaces is iteratively calculated using the limit equilibrium method. The critical acceleration is the minimum horizontal acceleration when the slope stability factor Fs is exactly equal to 1.

0. Double integration is performed on the input seismic wave acceleration time history. When and only when the seismic wave acceleration exceeds the calculated critical acceleration, the excess acceleration is integrated to calculate the accumulated velocity and displacement of the slider. At step S504, when the calculated maximum permanent displacement is greater than 30 cm, the slope is determined to be at risk of instability under earthquake conditions. The platform automatically matches and generates a targeted seismic reinforcement solution from a preset library of anchor cable reinforcement solutions, and feeds the reinforcement workload and constraints back into the multi-objective genetic algorithm of step S4 for a new round of optimization until all working condition requirements are met.

6. A road surveying system suitable for complex terrain, used to implement the road optimization design method suitable for complex terrain according to any one of claims 1 to 5, characterized in that: include: The adaptive leveling outrigger system includes three independent servo hydraulic outriggers. Each outrigger has a built-in displacement sensor and uses a PID controller to adjust the extension and retraction in real time to ensure that the system platform inclination is ≤0.5°. A multi-source fusion acquisition module integrates a dual-frequency RTK GNSS receiver, millimeter-wave radar, and an electronic compass. The millimeter-wave radar collects terrain data in rainy and foggy weather. Edge computing unit, running slope stability coefficient algorithm and sight distance compensation calculation in real time; The data credibility label generation module is used to evaluate the credibility of the measurement results based on the real-time sensor data and trigger data re-collection when the credibility is lower than the threshold.

7. The road measurement system suitable for complex terrain according to claim 6, characterized in that: The adaptive leveling leg system includes: A conical spiral grounding spike is installed at the end of each leg. The screw-in depth is adaptively adjusted based on the soil resistance. The soil resistance is obtained by real-time monitoring of the current value of the servo motor driving the conical spiral grounding spike. When the current value reaches a preset threshold, it is determined that the screw-in depth meets the requirement. The PID controller completes platform leveling within 3 seconds based on the three-axis accelerometer data.

8. The road measurement system suitable for complex terrain according to claim 6, characterized in that: The multi-source fusion acquisition module also includes: RTK FNSS-UWB dual-mode positioning switching unit, which enables centimeter-level positioning through pre-deployed UWB base stations when GNSS signals are lost; The millimeter-wave radar is equipped with an active temperature control cover, which includes a semiconductor refrigeration chip and a temperature control microcontroller. Forced air cooling is activated when the temperature of the millimeter-wave radar is greater than 50°C, and resistance heating is activated when the temperature of the millimeter-wave radar is less than -10°C.

9. The road measurement system suitable for complex terrain according to claim 6, characterized in that: The edge computing unit is configured to perform the following operations: The sliding window RANSAC algorithm is used to process the point cloud data collected by the millimeter wave radar to extract terrain feature lines; After the precise algorithm for calculating the slope stability coefficient times out, it switches to emergency calculation mode and calls a pre-trained convolutional neural network (CNN) model to quickly estimate the stability coefficient; The CNN takes slope height, slope angle, and geotechnical properties as input and outputs an approximate stability coefficient with an error of no more than 5% from the exact solution.

10. The road measurement system suitable for complex terrain according to claim 6, characterized in that: The data credibility label generation module calculates the credibility according to the following formula: ； The GNSS signal-to-noise ratio is directly derived from the raw data output of the GNSS receiver. The point cloud density is obtained by real-time statistics of each frame of radar point cloud data by the edge computing unit. The temperature offset coefficient is calculated based on the deviation between the millimeter-wave radar temperature sensor reading and the preset ideal temperature. When the confidence level is less than 0.6, data re-collection is automatically triggered.

Images