Highway layered slope dynamic beam radar monitoring and linkage early warning method and system

Abstract

The present application relates to the field of monitoring and early warning, and more particularly to a dynamic beam radar monitoring and linkage early warning method and system for highway layered slopes, which method comprises: layered scanning of the layered geological structure of the slope, synchronous acquisition of phase-coherent echo signals, multi-threshold scattering point extraction, elimination of interference signals caused by vehicle traffic through prior constraint model and geological vegetation identification network analysis to obtain a spatio-temporal coherent scattering point set, input into a deformation calculation pipeline, differential interferometry, adaptive atmospheric disturbance correction and slope structure parameter inversion, generation of displacement data flow, spatio-temporal correlation analysis with layered dip angle time series data and meteorological hydrological data, dynamic baseline update of slope stability evaluation indexes, and generation of early warning instructions. The present application forms a technical system from interference suppression, accurate calculation to decision-making closed loop, ensures high reliability of highway slope monitoring results, and provides intelligent support for driving safety and emergency control.

Description

Technical Field

[0001] This invention relates to the field of monitoring and early warning technology, and in particular to a dynamic beam radar monitoring and linkage early warning method and system for layered slopes of highways. Background Technology

[0002] Currently, highway slope safety monitoring mainly relies on manual inspections, which involve periodic on-site surveys combined with visual inspections and simple instrument measurements to assess slope condition. However, manual inspections depend on personnel's subjective judgment of slope stability based on experience, resulting in insufficient monitoring accuracy. Furthermore, the time intervals between inspections create blind spots in slope condition monitoring, making it difficult to achieve continuous, 24 / 7 coverage.

[0003] Furthermore, traditional manual methods have limited ability to detect sudden, minute deformations and cannot provide real-time feedback on slope dynamics. This is especially true when slopes enter a phase of accelerated deformation, where monitoring delays may cause missed warnings. On the other hand, manual inspections require personnel to work in complex terrain environments, and safety risks in slope areas increase significantly, particularly during severe weather conditions such as heavy rain and strong winds, further highlighting potential safety hazards for personnel.

[0004] Therefore, existing technologies are insufficient to meet the needs of real-time monitoring and accurate early warning of highway slope stability, and there is an urgent need for a technical solution that can overcome the shortcomings of manual inspection and improve monitoring coverage and timeliness. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a dynamic beam radar monitoring and linkage early warning method and system for layered slopes of highways, which solves the technical problems of existing manual inspections having subjective errors, monitoring blind spots and safety risks, and being unable to provide real-time and accurate early warning of highway slope stability.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, the main technical solutions adopted by the present invention include:

[0009] In a first aspect, embodiments of the present invention provide a dynamic beam radar monitoring and coordinated early warning method for layered slopes of highways, comprising:

[0010] A continuous wave radar array with adjustable beam angle is used to perform layered scanning of the layered geological structure of highway slopes, and simultaneously acquire phase coherent echo signals of the slope surface and internal structure.

[0011] Multi-threshold scattering points are extracted from the phase coherent echo signal. The data is then analyzed and processed by a prior constraint model constructed from the acquired slope geometric features and a geological vegetation recognition network optimized by transfer learning. Interference signals caused by vehicle traffic are excluded to obtain a set of spatiotemporal coherent scattering points.

[0012] The spatiotemporal coherent scattering point set is input into the preset deformation calculation pipeline, and differential interferometry, adaptive atmospheric disturbance correction and slope structure parameter inversion are executed simultaneously to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics.

[0013] The displacement data stream is spatiotemporally correlated with the acquired layered dip angle time series data and meteorological and hydrological data. The slope stability assessment indicators are updated in real time through dynamic baselines and linked with augmented reality display terminals to automatically generate multi-level early warning instructions.

[0014] Optionally, an adjustable beam angle continuous wave radar array is used to perform layered scanning of the layered geological structure of the highway slope, simultaneously acquiring phase coherent echo signals of the slope surface and internal structure, including:

[0015] Based on the dielectric constant distribution and layer thickness parameters of the layered geological structure of the highway slope, a layered scanning strategy matching each soil and rock layer is dynamically generated and a beamforming network is constructed. Combined with the preset scanning parameter constraints, the beam incident angle adjustment sequence of each scanning cycle of the continuous wave radar array is determined.

[0016] Based on the scanning parameter constraints, the spatial arrangement parameters of the transmitting / receiving units of the radar array and the combination of the operating frequency bands are initialized, and the hardware configuration templates corresponding to each scanning layer are constructed.

[0017] The control continuous wave radar array switches the transmitted beam angle according to the beam incident angle adjustment sequence. When switching to each scanning layer, the hardware configuration template corresponding to the current layer is loaded. The amplitude and phase distribution weights of the beamforming network are adjusted in real time according to the dielectric constant and target detection depth of the current scanning layer, so that the spatial resolution of the radar array and the electromagnetic wave penetration depth are adaptively matched with the scanning layer.

[0018] During the layered scanning process, the combination of parameters including transmission power, pulse repetition frequency and coherent accumulation time is dynamically adjusted based on the signal-to-noise ratio spectrum characteristics of the echo signal to maintain the signal phase stability of each scanning layer within a preset stable range.

[0019] The received multi-level echo signals are orthogonally separated, and the phase-consistent radar image of the slope surface and internal structure is reconstructed by a digital beamforming matrix. The output is a set of phase-coherent echo signals that meet the preset differential interferometry requirements.

[0020] Optionally, based on the dielectric constant distribution and layer thickness parameters of the layered geological structure of the highway slope, a layered scanning strategy matching each soil and rock layer is dynamically generated, and a beamforming network is constructed. Combined with preset scanning parameter constraints, the beam incidence angle adjustment sequence for each scanning cycle of the continuous wave radar array is determined, including:

[0021] Based on the imported geological survey data of the highway slope, the dielectric constant distribution and layer thickness parameters of each soil and rock layer were extracted.

[0022] Based on the dielectric constant distribution and layer thickness parameters of each soil and rock layer, a layered scanning strategy function is constructed with the joint optimization objective of minimizing interlayer phase error and total scanning time. The interlayer phase error is quantified by the sum of the squares of the phase delay of the electromagnetic wave propagation path of each layer and the corresponding threshold deviation. The total scanning time is calculated by multiplying and accumulating the beam switching frequency and the dwell time.

[0023] In the layered scanning strategy optimization function, the phase stability error threshold, the maximum interval between adjacent beam incident angles, and the upper limit of single-cycle scanning time are configured as hard constraints. The beam incident angle adjustment sequence that satisfies all constraints is obtained by solving the constraint optimization algorithm.

[0024] A beamforming network with a matching layered structure is constructed synchronously, and the amplitude and phase distribution weights of the beamforming network are dynamically initialized according to the dielectric constant distribution of each soil and rock layer.

[0025] Optionally, the received multi-level echo signals are orthogonally separated, and a phase-coherent radar image of the slope surface and internal structure is reconstructed using a digital beamforming matrix. The output set of phase-coherent echo signals that meets the preset differential interferometry requirements includes:

[0026] Based on the dip angle and layer thickness parameters of the slope layered geological structure extracted from the imported geological exploration data, an orthogonal temporal coding sequence and spatial beam pointing code set are assigned to each scanning layer.

[0027] A composite control matrix for spatiotemporal joint coding is constructed based on time-domain orthogonal coding sequences and spatial beam pointing code sets;

[0028] By performing cross-correlation decoupling operations on the multi-level echo signals received by the radar array through a composite control matrix, the spatiotemporal coupling components of the echo signals at different levels are separated, and the baseband signal channels of each scanning level are generated.

[0029] Spatial selective enhancement processing is performed on each baseband signal channel using a spatial beam pointing code set to suppress interference from adjacent layers;

[0030] Based on the enhanced signal, the three-dimensional complex scattering characteristic distribution data of each scanning layer are reconstructed, and the phase distortion caused by multipath effect is eliminated by phase gradient consistency analysis.

[0031] The distortion-free data is fitted with the full aperture phase plane and coherently accumulated to generate a spatially continuous phase-consistent radar image.

[0032] Extract the set of scattering points that meet the preset spatial continuity criterion from the radar image sequence, and output the set of phase coherent echo signals.

[0033] Optionally, multi-threshold scattering points are extracted from the phase coherent echo signal. This is then analyzed using a prior constraint model constructed from the acquired slope geometric features and a geological vegetation recognition network optimized through transfer learning. Interference signals caused by vehicle traffic are excluded, resulting in a spatiotemporal coherent scattering point set including:

[0034] Based on the slope angle, layer thickness parameters and structural surface spatial orientation parameters extracted from the imported geological exploration data, a three-dimensional slope prior model is constructed, which includes constraints on the spatial density of scattering points, the continuity of displacement vectors, and the orientation of structural surfaces and radar line of sight.

[0035] Based on the joint criterion of amplitude deviation index and phase stability threshold, candidate scattering points are screened from phase coherent echo signals. The spatial density constraint of the three-dimensional slope prior model is used to perform layer classification test on the candidate scattering points in order to eliminate outliers that deviate from the geological features of their respective layers.

[0036] The scattering points that have undergone stratification are input into a pre-constructed transfer learning-optimized geological vegetation recognition network, which outputs the geological attribute classification results and confidence scores for each scattering point.

[0037] Based on the vehicle Doppler parameters and acceleration observations obtained from the phase coherent echo signal, a dynamic target scattering model containing multiple scattering centers is constructed based on the radar transmitted signal wavelength, the vehicle's three-dimensional attitude, and the azimuth angle of the scattering point. Each scattering center in the model is characterized by the complex reflection coefficient, the time-varying radial distance, and the acceleration modulation frequency term. The time-varying radial distance includes the initial distance, the radial velocity, and the second motion component of the acceleration. A rectangular window function is introduced to limit the effective illumination time window of each scattering point.

[0038] The time-frequency distribution feature spectrum of dynamic interference signals is generated by using a moving target scattering model. The time-varying scattering intensity envelope curve and the rate of change of frequency are extracted from the time-frequency distribution feature spectrum to establish a vehicle interference feature library.

[0039] Cross-validation was performed based on the vehicle interference feature library and the geological attribute classification results. Transient interference signals caused by vehicle passage were identified and filtered out according to the preset vehicle interference judgment criteria to obtain a spatiotemporal coherent scattering point set. The vehicle interference judgment criteria included: the scattering intensity envelope was positively correlated with the vehicle speed and the envelope width matched the lane occupancy time; the frequency modulation rate change curve conformed to the nonlinear characteristics of the acceleration / braking mode; and the geological attribute classification result of the target area was a non-rock and soil structure.

[0040] Optionally, the spatiotemporal coherent scattering point set is input into a preset deformation calculation pipeline, and differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion are performed simultaneously to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics, including:

[0041] By receiving a set of spatiotemporal coherent scattering points through a deformation solution pipeline pre-deployed on edge computing nodes, performing spatiotemporal reference unification processing, and establishing standardized observation data registered with the slope geographic coordinate system;

[0042] Based on observational data, the following parallel operations are performed in the deformation calculation pipeline: scattering points of adjacent time phases are selected for phase difference calculation to generate an interferometric phase map covering the entire slope area; an atmospheric delay correction surface is constructed based on meteorological sensor data deployed on the slope to compensate for spatial variation errors in the interferometric phase map and obtain the residual phase field; the elastic modulus and Poisson's ratio from the preset geotechnical mechanics database are called, and the deformation gradient field of each soil and rock layer is calculated using the displacement-strain conversion relationship;

[0043] The interferometric phase map, residual phase field and deformation gradient field are fused from multiple sources, and the three-dimensional deformation vector is solved by the confidence-weighted adaptive least squares algorithm.

[0044] By analyzing the statistical characteristics of deformation vectors through a time-series sliding window, the time-series evolution characteristics including slope surface displacement rate, acceleration, and deep slip trend parameters are extracted, and sub-millimeter displacement data streams are generated in time sequence.

[0045] Optionally, the displacement data stream is spatiotemporally correlated with the acquired layered dip angle time-series data and meteorological and hydrological data. The slope stability assessment indicators are updated in real time through dynamic baselines, and linked with augmented reality display terminals to automatically generate multi-level early warning instructions, including:

[0046] The displacement data stream is decomposed into frequency bands to extract the first deformation component, the second deformation component, and the third deformation component. Simultaneously, the layered dip angle time series data collected by the dip angle sensors deployed on the slope are time-domain aligned to generate a joint observation sequence of deformation and dip angle.

[0047] The acquired rainfall data is converted into a slope surface runoff intensity distribution map through spatial interpolation, and the pore water pressure gradient field is inverted based on the acquired groundwater level monitoring data and the pore water pressure gradient change rate within a preset time period is calculated.

[0048] The time-delay coupling strength between the first deformation component and the rate of change of pore water pressure gradient is calculated to obtain the time delay parameter of displacement lagging behind water pressure change. The spatial correlation region between the second deformation component and the runoff intensity distribution map is analyzed to calibrate the deformation sensitive area and output the spatial coordinate set of the deformation sensitive area. The frequency domain coherence between the third deformation component and the time series data of stratified dip angle is detected to extract the activation characteristic frequency of the structural surface.

[0049] An assessment baseline was established based on joint observation sequences, including displacement acceleration threshold, tilt rate of change threshold, hysteresis time threshold, and structural surface activation hazard frequency band.

[0050] The density of sensitive points in each slope zone is calculated based on the spatial coordinate set of the deformation sensitive area. The evaluation baseline is dynamically adjusted according to the density of sensitive points. Among them, the displacement acceleration threshold increases proportionally with the increase of the density of sensitive points, the dip angle change rate threshold tightens linearly with the increase of the density of sensitive points, the hysteresis time threshold is corrected based on the density of sensitive points, and the structural surface activation danger frequency band is adaptively expanded with the increase of density.

[0051] When multiple conditions are met, such as displacement acceleration being greater than the adjusted displacement acceleration threshold, tilt angle change rate being greater than the adjusted tilt angle change rate threshold, delay time parameter being less than the adjusted hysteresis time threshold, and structural surface activation characteristic frequency being in the adjusted structural surface activation danger frequency band, slope stability deterioration judgment is triggered.

[0052] Based on the real-time data obtained at the time of triggering the judgment, including displacement exceeding limits, slope trend of tilt angle change, and activation characteristic frequency of structural surfaces, combined with the evaluation baseline, a comprehensive instability index is calculated through nonlinear weighting to generate a set of slope stability evaluation indicators.

[0053] The slope stability assessment index set is input into the augmented reality display terminal to visualize the slope deformation field and the three-dimensional geological model, and automatically generates multi-level early warning instructions that are linked with the highway operation management system.

[0054] Optionally, the slope stability assessment index set can be input into an augmented reality display terminal to visualize the slope deformation field and the three-dimensional geological model through virtual-real overlay, and to automatically generate multi-level early warning instructions that are linked with the highway operation management system, including:

[0055] Real-time slope stability assessment indicators are synchronized with augmented reality display terminals to achieve spatial registration and overlay display of slope deformation field and three-dimensional geological model;

[0056] Based on the comprehensive instability index level, generate a multi-level early warning signal that includes at least one of the following: speed limit control, lane closure, and emergency support instructions;

[0057] Multi-level early warning signals are encoded into traffic control protocol data packets, and linked with the gate control unit, information board display unit, and maintenance dispatch platform of the highway operation management system through a 5G-V2X communication module.

[0058] Secondly, embodiments of the present invention provide a dynamic beam radar monitoring and linkage early warning system for layered slopes of highways. The system is used to execute the method described above, including:

[0059] The radar hardware unit is used to perform layered scanning of the layered geological structure of the highway slope according to the layered scanning strategy issued by the edge computing unit, and simultaneously acquire the phase coherent echo signals of the slope surface and internal structure, and upload them to the edge computing unit.

[0060] The edge computing unit, connected to the radar hardware unit, includes a signal processing module, a deformation calculation module, and an early warning module. The signal processing module extracts multi-threshold scattering points from the phase coherent echo signal, combines the analysis and processing of a prior constraint model constructed from the acquired slope geometric features with a geological vegetation recognition network optimized by transfer learning, and eliminates interference signals caused by vehicle traffic to obtain a spatiotemporal coherent scattering point set. The deformation calculation module inputs the spatiotemporal coherent scattering point set into a preset deformation calculation pipeline, simultaneously performing differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics. The displacement data stream is then spatiotemporally correlated with the acquired layered dip angle time-series data and meteorological and hydrological data. The early warning module updates the slope stability assessment indicators in real time through a dynamic baseline and links with the augmented reality display terminal to automatically generate multi-level early warning instructions.

[0061] Optionally, the radar hardware unit includes: a continuous wave radar array, consisting of multiple transmitting units and multiple receiving units forming a MIMO architecture; and a beam control module, connected to the continuous wave radar array via a high-speed bus, used to control the transmitting units of the radar array to transmit frequency modulation signals according to the beam incidence angle adjustment sequence and scanning parameter constraints of each scanning cycle of the continuous wave radar array in the issued layered scanning strategy, and synchronously adjust the beam combining direction of the receiving units to match the spatial coverage requirements of each scanning layer according to the beam incidence angle adjustment sequence under the scanning parameter constraints.

[0062] (III) Beneficial Effects

[0063] The beneficial effects of this invention are:

[0064] First, by employing a continuous wave radar array with dynamically adjustable beam angles, the monitoring blind zone limitation of the traditional fixed scanning mode is overcome, enabling full-dimensional synchronous scanning of the surface and internal structure of layered slopes, significantly improving the ability to detect hidden geological defects.

[0065] Secondly, by combining multi-threshold scattering point extraction with a recognition network optimized through transfer learning, this invention intelligently distinguishes between vegetated areas and areas with abnormal geological structures while preserving effective scattering characteristics of the soil and rock mass, effectively overcoming the risk of subjective errors caused by experience differences in manual judgment. Crucially, the introduction of a dynamic interference suppression mechanism effectively identifies and eliminates transient interference signals generated by vehicle traffic. Compared to traditional fixed-threshold filtering methods, this invention significantly reduces the risk of misjudgment caused by interference such as vehicle vibration and metal reflection while preserving the true scattering points of the slope.

[0066] Furthermore, by using the parallel processing architecture of the deformation calculation pipeline, multi-task collaborative calculation of differential interferometry, atmospheric correction and structural inversion is realized, optimizing the traditional step-by-step processing flow into real-time continuous calculation, and solving the problem of early warning lag caused by the long cycle of manual inspection.

[0067] Based on this, a dynamic baseline update mechanism is constructed by integrating displacement data, dip angle time series and hydrological and meteorological parameters through multimodal spatiotemporal correlation analysis. This enables the stability assessment index to adapt to changes in slope geological conditions and environment, avoiding the shortcomings of fixed threshold systems that are slow to respond to local risks.

[0068] Ultimately, through visualization linkage with augmented reality systems and automatic generation of multi-level early warning commands, a complete technical system is formed, encompassing interference suppression, precise calculation, and decision-making closed loop. This ensures the high reliability of highway slope monitoring results and provides intelligent support for driving safety and emergency management. Attached Figure Description

[0069] Figure 1 A flowchart illustrating the method provided in an embodiment of the present invention;

[0070] Figure 2 This is a schematic diagram illustrating the specific process of step S1 of the method provided in this embodiment of the invention;

[0071] Figure 3 This is a schematic flowchart illustrating step S11 of the method provided in this embodiment of the invention.

[0072] Figure 4 This is a detailed flowchart illustrating step S15 of the method provided in this embodiment of the invention.

[0073] Figure 5 This is a detailed flowchart illustrating step S2 of the method provided in this embodiment of the invention;

[0074] Figure 6 This is a detailed flowchart illustrating step S3 of the method provided in this embodiment of the invention.

[0075] Figure 7 This is a detailed flowchart illustrating step S4 of the method provided in this embodiment of the invention;

[0076] Figure 8 A schematic diagram of the specific process of step S48 of the method provided in the embodiment of the present invention. Detailed Implementation

[0077] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0078] like Figure 1 As shown in the embodiment of the present invention, a dynamic beam radar monitoring and linkage early warning method for layered slopes of highways includes: using a continuous wave radar array with adjustable beam angle to perform layered scanning of the layered geological structure of the highway slope, and simultaneously acquiring phase coherent echo signals of the slope surface and internal structure; extracting multi-threshold scattering points from the phase coherent echo signals, and analyzing and processing them through a prior constraint model constructed from the acquired slope geometric features and a geological vegetation recognition network optimized by transfer learning, while excluding interference signals caused by vehicle traffic, to obtain a spatiotemporal coherent scattering point set; inputting the spatiotemporal coherent scattering point set into a preset deformation calculation pipeline, and simultaneously performing differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics; performing spatiotemporal correlation analysis on the displacement data stream with the acquired layered dip angle time-series data and meteorological and hydrological data, updating the slope stability assessment index in real time through dynamic baselines, and linking with an augmented reality display terminal to automatically generate multi-level early warning instructions.

[0079] First, by employing a continuous wave radar array with dynamically adjustable beam angles, this invention overcomes the monitoring blind zone limitations of traditional fixed scanning modes, enabling full-dimensional synchronous scanning of the surface and internal structure of layered slopes, and significantly improving the ability to detect hidden geological defects.

[0080] Secondly, by combining multi-threshold scattering point extraction with a recognition network optimized through transfer learning, this invention intelligently distinguishes between vegetated areas and areas with abnormal geological structures while preserving effective scattering characteristics of the soil and rock mass, effectively overcoming the risk of subjective errors caused by experience differences in manual judgment. Crucially, the introduction of a dynamic interference suppression mechanism effectively identifies and eliminates transient interference signals generated by vehicle traffic. Compared to traditional fixed-threshold filtering methods, this invention significantly reduces the risk of misjudgment caused by interference such as vehicle vibration and metal reflection while preserving the true scattering points of the slope.

[0081] Furthermore, by using the parallel processing architecture of the deformation calculation pipeline, multi-task collaborative calculation of differential interferometry, atmospheric correction and structural inversion is realized, optimizing the traditional step-by-step processing flow into real-time continuous calculation, and solving the problem of early warning lag caused by the long cycle of manual inspection.

[0082] Based on this, a dynamic baseline update mechanism is constructed by integrating displacement data, dip angle time series and hydrological and meteorological parameters through multimodal spatiotemporal correlation analysis. This enables the stability assessment index to adapt to changes in slope geological conditions and environment, avoiding the shortcomings of fixed threshold systems that are slow to respond to local risks.

[0083] Ultimately, through visualization linkage with augmented reality systems and automatic generation of multi-level early warning commands, a complete technical system is formed, encompassing interference suppression, precise calculation, and decision-making closed loop. This ensures the high reliability of highway slope monitoring results and provides intelligent support for driving safety and emergency management.

[0084] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and that the scope of the present invention can be fully conveyed to those skilled in the art.

[0085] Specifically, embodiments of the present invention provide a dynamic beam radar monitoring and coordinated early warning method for layered slopes of highways, comprising:

[0086] S1. An adjustable beam angle continuous wave radar array is used to perform layered scanning of the layered geological structure of the highway slope, and to simultaneously acquire phase coherent echo signals of the slope surface and internal structure.

[0087] Furthermore, such as Figure 2 As shown, step S1 includes:

[0088] S11. Based on the dielectric constant distribution and layer thickness parameters of the layered geological structure of the highway slope, dynamically generate a layered scanning strategy that matches each soil and rock layer and construct a beamforming network. Combined with the preset scanning parameter constraints, determine the beam incident angle adjustment sequence for each scanning cycle of the continuous wave radar array.

[0089] Furthermore, such as Figure 3 As shown, step S11 includes:

[0090] S111. Based on the imported geological survey data of the highway slope, extract the dielectric constant distribution and layer thickness parameters of each soil and rock layer. The dielectric constant distribution varies with depth and is determined by the electromagnetic properties of different soil and rock types, while the layer thickness parameters characterize the vertical extension scale of each soil and rock layer.

[0091] S112. Based on the dielectric constant distribution and layer thickness parameters of each soil and rock layer, a layered scanning strategy function is constructed with the joint optimization objective of minimizing interlayer phase error and total scanning time. The interlayer phase error is quantified by the sum of the squares of the phase delay of the electromagnetic wave propagation path of each layer and the corresponding threshold deviation. The total scanning time is calculated by multiplying and accumulating the beam switching frequency and the dwell time.

[0092] S113. In the layered scanning strategy optimization function, the phase stability error threshold, the maximum interval between adjacent beam incident angles, and the upper limit of single-cycle scanning time are configured as hard constraints. The beam incident angle adjustment sequence that satisfies all constraints is obtained by solving the constraint optimization algorithm.

[0093] The layered scanning strategy optimization function achieves adaptive optimization of scanning parameters through deep coupling of geological parameters and electromagnetic propagation model. Based on the dielectric constant distribution and layer thickness parameters of the layered geological structure of the slope, the layered scanning strategy optimization function is constructed as follows: ;

[0094] In the formula, θ n For the first n The incident angle of each scanning beam, f m For the first m Carrier frequency at each frequency point λ ( f m ) represents frequency f m The corresponding wavelength, d k These are the layer thickness parameters for the layered geological structure of the slope. The dielectric constant distribution of the layered geological structure of the slope. This is the phase stability threshold. T scan Total scan time α t For time weighting coefficients, z q Indicates the first q Depth coordinates of each geological stratum Q This represents the number of geological strata.

[0095] In the process of solving the constrained optimization algorithm, a candidate solution set is first constructed based on the initially randomly generated beam incidence angle sequence. Then, the phase stability error of each layer, the adjacent beam angle interval, and the scanning time constraint are transformed into penalty terms of the objective function using the penalty function method, forming an optimization model that comprehensively considers detection accuracy and timeliness. Subsequently, a sequential quadratic programming algorithm is used to iteratively optimize the objective function: in each iteration, the phase error gradient and constraint violation amount corresponding to the current solution are calculated, the incidence angle step size and frequency combination are dynamically adjusted, and the optimal step direction is determined through linear search to satisfy the constraint boundary conditions. Finally, a beam incidence angle adjustment sequence that satisfies the following scanning parameter constraints is generated: phase stability error of each layer ≤ 0.08 rad; beam incidence angle interval between adjacent layers ≤ 2°; single-cycle scanning time... Seconds. Therefore, by dynamically controlling the radar array beam direction based on the output scanning parameters, adaptive optimization of the electromagnetic characteristics of layered slope scanning is achieved.

[0096] S114. Simultaneously construct a beamforming network with a matched layered structure and initialize the amplitude and phase distribution weights of the beamforming network. The beamforming network is a signal processing architecture in a radar system that achieves spatial directivity synthesis of electromagnetic beams by adjusting the amplitude and phase distribution weights between transmitting / receiving units. When initializing the amplitude and phase distribution weights of the beamforming network, the ideal beam directivity pattern for each layer is pre-calculated using an electromagnetic field simulation model. The amplitude and phase parameters are then fitted using a least-squares approximation algorithm to ensure that the initial main lobe width and side lobe suppression ratio meet the inter-layer resolution requirements.

[0097] S12. Based on the scanning parameter constraints, initialize the spatial arrangement parameters of the radar array's transmitting / receiving units and the combination of operating frequency bands, and construct the hardware configuration template corresponding to each scanning layer.

[0098] In this step, based on the scanning parameter constraints, the spatial arrangement parameters of the radar array's transmitting / receiving units and the combination of operating frequency bands are defined;

[0099] Hardware configuration templates were constructed to address the differences in geological characteristics at different strata, as shown in Tables 1, 2, and 3 below:

[0100] Table 1 Surface Scan Template

[0101]

[0102] Table 2. Mid-layer scanning template

[0103]

[0104] Table 3 Deep Scan Template

[0105]

[0106] S13. Control the continuous wave radar array to switch the transmission beam angle according to the beam incident angle adjustment sequence, and when switching to each scanning layer, load the hardware configuration template corresponding to the current layer, and adjust the amplitude and phase distribution weight of the beamforming network in real time according to the dielectric constant and target detection depth of the current scanning layer, so that the spatial resolution of the radar array and the electromagnetic wave penetration depth adapt to the change of the scanning layer.

[0107] In this step, the corresponding hardware configuration template is invoked the instant the target scanning layer is switched.

[0108] First, configure the frequency: surface layer 14GHz → middle layer 12GHz → deep layer 10GHz, changing the frequency layer by layer;

[0109] Secondly, adjust the array element activation mode: all array elements in the surface layer are working → selectively dormant edge elements in the middle layer → the deep layer sparse array is enabled; then, dynamically adjust the amplitude and phase weights: according to the current layer dielectric constant and target depth, perform amplitude optimization, and adjust the array element excitation amplitude through the least squares approximation algorithm to make the beam main lobe width adapt to the layer thickness.

[0110] S14. During the layered scanning process, the combination of parameters including transmit power, pulse repetition frequency, and coherent accumulation time is dynamically adjusted based on the signal-to-noise ratio spectral characteristics of the echo signal to maintain the signal phase stability of each scanning layer within a preset stable range. Preferably, the signal phase stability of each scanning layer is maintained within a 0.1 rad threshold.

[0111] S15. Perform orthogonal separation on the received multi-level echo signals, reconstruct the phase-consistent radar image of the slope surface and internal structure through a digital beamforming matrix, and output a set of phase-coherent echo signals that meet the preset differential interferometry requirements.

[0112] Furthermore, such as Figure 4 As shown, step S15 includes:

[0113] S151. Based on the dip angle and layer thickness parameters of the slope layered geological structure extracted from the imported geological exploration data, assign orthogonal temporal coding sequences and spatial beam pointing codes to each scanning layer. The temporal coding sequences of adjacent layers satisfy a cross-correlation peak-to-average power ratio (PAPR) ≤ -30dB, and the spatial beam pointing code includes azimuth and elevation phase modulation parameters to ensure no main lobe overlap between layers (angular interval ≥ 1.5 times the beamwidth).

[0114] S152. Construct a spatiotemporal joint coding composite control matrix based on time-domain orthogonal coding sequences and spatial beam pointing code sets.

[0115] S153. Perform cross-correlation decoupling operation on the multi-level echo signals received by the radar array through the composite control matrix, separate the spatiotemporal coupling components of the echo signals at different levels, and generate the baseband signal channels for each scanning level.

[0116] S154. Spatial selective enhancement processing is performed on each baseband signal channel using a spatial beam pointing code set and adjacent layer interference is suppressed to achieve a sidelobe level ≤ -25dB.

[0117] S155. Based on the enhanced signal, reconstruct the three-dimensional complex scattering characteristic distribution data of each scanning layer, and eliminate the phase distortion caused by multipath effect through phase gradient consistency analysis. The phase gradient change rate is controlled within 0.2 radians per wavelength.

[0118] S156. Perform full-aperture phase plane fitting and coherent accumulation on the distortion-free data to generate a spatially continuous phase-consistent radar image.

[0119] S157. Extract the set of scattering points that meet the preset spatial continuity criteria (region density ≥ 5 points / m², phase standard deviation ≤ 0.05λ) from the radar image sequence, and output the set of phase coherent echo signals.

[0120] In one specific embodiment, a composite control matrix integrating time-domain coded sequences and spatial-domain beam pointing codes combines the orthogonal modulation signal from the radar transmitter with the adaptive beamforming from the receiver through spatiotemporal joint coding. This achieves spatial selectivity enhancement and temporal interference suppression of echo signals at different scanning layers, providing precise beam pointing control and signal separation capabilities for layered scanning. The specific formula is as follows: In the formula, For time-domain orthogonal coding matrix ( T The length of the encoded sequence is a power of 2; a longer encoding results in higher distance resolution but also increases computational complexity. For the spatial beam pointing code matrix, O The number of transmitting units, M The number of receiving units, The Kronecker product is used to expand the spatiotemporal coding dimension, ensuring that the matrix dimension is strictly matched to the radar architecture. This is the Hadamard product, used to represent element-wise multiplication. The array manifold matrix is ​​the phase response matrix describing the relationship between the array geometry and the beam pointing direction. Each element corresponds to the spatial phase delay of the array element. θ The beam pointing azimuth angle, ϕ The beam pointing elevation angle. Accurate modeling of the array manifold matrix can compensate for wavefront distortion caused by layered structures.

[0121] The three-dimensional complex scattering characteristic distribution data is a three-dimensional spatial distribution matrix characterizing the scattering characteristics of each layer of the slope, while for the th l The three-dimensional spatial distribution matrix of each scanning layer is as follows: , For the first o The first launch unit, the first m The echo data from the receiving unit includes K A multiple signal in a fast-moving sequence, For the corresponding layer l and( p,m ) The beamforming sub-matrix of the transmit-receive pair includes azimuth and elevation modulation parameters. H This represents the conjugate transpose operation. The cascaded reconstruction process of the three-dimensional complex scattering matrix effectively separates the scattering contributions from each layer.

[0122] Next, the reconstructed complex scattering matrix S l The phase gradient is calculated pixel by pixel, and the gradient change rate is constrained to not exceed a threshold (e.g., 0.2 radians per wavelength). Phase compensation is performed on regions that exceed the threshold. The phase difference between adjacent pixels is minimized through iterative optimization, and a distortion-free three-dimensional complex scattering characteristic matrix is ​​output. Its phase continuity meets the requirements of interferometry.

[0123] S2. Multi-threshold scattering point extraction is performed on the phase coherent echo signal. The analysis and processing are carried out by the prior constraint model constructed from the acquired slope geometric features and the geological vegetation recognition network optimized by transfer learning, and the interference signal caused by vehicle traffic is excluded to obtain the spatiotemporal coherent scattering point set.

[0124] Furthermore, such as Figure 5 As shown, step S2 includes:

[0125] S21. Based on the slope angle, layer thickness parameters, and spatial orientation parameters of structural surfaces extracted from the imported geological exploration data, construct a three-dimensional a priori model of the slope, including constraints on the spatial density of scattering points, the continuity of displacement vectors, and the orientation of structural surfaces relative to radar line of sight. Specifically, establish an a priori model containing the following constraints: the spatial density threshold of scattering points for each soil and rock layer, according to the formula... calculate, h i The tolerance for thickness and directional continuity of displacement vectors of adjacent layers is ≤15°, the rate of change of displacement modulus is ≤30%, and the allowable deviation range of the angle between the orientation of the structural surface and the radar line of sight is ±20° to avoid specular reflection blind spots.

[0126] S22. Based on the joint criterion of amplitude deviation index and phase stability threshold, candidate scattering points are screened from the phase coherent echo signal. Using the spatial density constraint of the three-dimensional slope prior model, a layer attribution test is performed on the candidate scattering points to eliminate outliers that deviate from the geological characteristics of their respective layers. A dual-condition screening is performed using an amplitude deviation index threshold (≥0.7) and a phase stability threshold (≥0.85); a layer attribution test is performed based on the spatial density threshold to eliminate scattering points with a density deviation exceeding ±30%.

[0127] S23. Input the scattering points that have undergone stratification verification into the pre-constructed transfer learning optimized geological vegetation recognition network, and output the geological attribute classification results and confidence scores of each scattering point.

[0128] Specifically, the geological vegetation recognition network uses a pre-trained ResNet-50 as its backbone. It maps the natural image feature space to the radar scattering feature space of the phase-coherent echo signal through a domain adaptation module. Furthermore, it introduces an attention mechanism by integrating a CBAM module after Stage 3 of the ResNet-50 to extract features from geological structure edges and vegetation textures. Finally, the network outputs geological type labels (rock / soil / vegetation) and confidence scores (range 0-1) for each scattering point.

[0129] S24. Based on the vehicle Doppler parameters and acceleration observations obtained from the phase coherent echo signal, a dynamic target scattering model containing multiple scattering centers is constructed based on the radar transmitted signal wavelength, the vehicle's three-dimensional attitude, and the azimuth angle of the scattering point. Each scattering center in the model is characterized by the complex reflection coefficient, time-varying radial distance, and acceleration modulation frequency term. The time-varying radial distance includes the initial distance, radial velocity, and the second-order motion component of acceleration. A rectangular window function is introduced to limit the effective illumination time window for each scattering point. The moving target scattering model is as follows: ;

[0130] In the formula, S m ( t ) is the radar array's... m The synthesized echo signal of the vehicle target from each receiving unit is composed of the superposition of echoes from multiple scattering centers. t The variable is time, representing the cumulative time from the start of radar signal observation to the current moment. N This refers to the number of independent scattering points on the vehicle; for example, the front of the vehicle, the body, and the wheels can all be considered as different scattering centers. N ≥1, For the first n Complex reflection coefficient at each scattering point It is a complex exponential function, here j The imaginary unit is used to represent the complex phase of a signal. Considering the Doppler frequency shift of the vehicle's three-dimensional attitude, v The radial velocity of the vehicle along the radar line of sight. The wavelength of the radar transmitted signal. θ n For the first n The azimuth angle of each scattering point relative to the radar line of sight. For the frequency modulation term caused by acceleration, a This refers to the radial acceleration of the vehicle along the radar's line of sight. For a time-varying distance model containing acceleration, Initial time t Radial distance = 0, v 0 is the initial velocity. a For acceleration, To characterize the effective illumination time window of the scattering point, For the first n The time delay of each scattering point entering the radar beam T n The dwell time of the scattering point in the radar beam is denoted as . This model, by introducing an acceleration term, three-dimensional attitude parameters, and a multi-scattering center architecture, effectively solves the model mismatch problem of traditional methods in complex traffic scenarios, providing theoretical support for the subsequent accurate removal of interference signals.

[0131] S25. Generate a time-frequency distribution feature map of dynamic interference signals through a moving target scattering model, extract the time-varying scattering intensity envelope curve and frequency modulation rate change curve from the time-frequency distribution feature map, and establish a vehicle interference feature library.

[0132] S26. Based on the cross-validation of the vehicle interference feature library and geological attribute classification results, identify and filter transient interference signals caused by vehicle passage according to the preset vehicle interference judgment criteria to obtain a spatiotemporal coherent scattering point set. Among them, the vehicle interference judgment criteria include: the scattering intensity envelope is positively correlated with the vehicle speed and the envelope width matches the lane occupancy time (the error between the envelope width and the lane occupancy time is ≤0.5 seconds); the frequency modulation rate change curve conforms to the nonlinear characteristics of the acceleration / braking mode (the correlation coefficient between the frequency modulation rate change curve and the theoretical acceleration / braking mode is ≥0.8, where the theoretical acceleration / braking mode refers to the non-uniform modulation characteristics of the radar echo signal caused by the vehicle's motion state in the time and frequency domain during acceleration or braking, which is obtained by fitting historical data); and the geological attribute classification result of the target area is a non-rock and soil structure (confidence level ≥0.9).

[0133] S3. Input the spatiotemporal coherent scattering point set into the preset deformation calculation pipeline, and simultaneously perform differential interferometry, adaptive atmospheric disturbance correction and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics.

[0134] Furthermore, such as Figure 6 As shown, step S3 includes:

[0135] S31. Receive spatiotemporal coherent scattering point sets through a deformation solution pipeline pre-deployed on edge computing nodes, perform spatiotemporal reference unification processing, and establish standardized observation data registered with the slope geographic coordinate system.

[0136] S32. Based on observation data, the following parallel operations are performed in the deformation calculation pipeline: scattering points of adjacent time phases are selected for phase difference calculation to generate an interferometric phase map covering the entire slope area; an atmospheric delay correction surface is constructed based on meteorological sensor data deployed on the slope to compensate for spatial variation errors in the interferometric phase map and obtain the residual phase field; the elastic modulus and Poisson's ratio of the preset geotechnical database are called, and the deformation gradient field of each geotechnical layer is calculated using the displacement-strain conversion relationship.

[0137] S33. The interferometric phase diagram, residual phase field and deformation gradient field are fused from multiple sources, and the three-dimensional deformation vector is solved by the confidence-weighted adaptive least squares algorithm.

[0138] S34. Analyze the statistical characteristics of deformation vectors through a time-series sliding window, extract time-series evolution characteristics including slope surface displacement rate, acceleration and deep slip trend parameters, and organize and generate sub-millimeter displacement data streams according to time sequence.

[0139] In a specific embodiment, the parallel execution channels in the deformation calculation pipeline include:

[0140] (1) Differential interference channel: Select adjacent time phase scattering points to perform millimeter wave interference and generate an interference phase map covering the entire slope.

[0141] (2) Atmospheric correction channel: An atmospheric delay correction surface is constructed based on meteorological sensors (temperature / humidity / pressure) deployed on the slope, and the formula is: In the formula, The phase delay of the radar signal is caused by changes in atmospheric refractive index. P q For air pressure, T w For temperature, H s For humidity, k 1, k 2, k 3 represents the delay correction coefficient for the corresponding parameter. k 1 = 0.12 rad / kPa, k 2 = 0.08 rad / ℃ k= 0.05rad / %RH, the corrected residual error of the interference phase is ≤0.3rad.

[0142] (3) Structural inversion channel: calling the elastic modulus E and Poisson's ratio from the geotechnical mechanics database. γ, Input displacement field obtained by radar interferometry inversion u =[ u x , u y , u z ] T , u x Let represent the displacement components of the slope in the east-west direction. u y Let represent the displacement components of the slope in the north-south direction. u z Let represent the vertical displacement component of the slope. Based on the small deformation assumption, calculate the strain tensor. In the formula, displacement vector components u i coordinates x j The partial derivative, displacement vector components u j coordinates x i The partial derivative, i , j = x , y , z , i and j It is a free index, representing the three directions in the coordinate system. x , y , z When the formula appears i , j = x , y , z When, it means that all possible [conditions] need to be considered. i , j The calculations are performed in combination.

[0143] Constructing a displacement-strain transformation model using the generalized Hooke's law: , σ ij For stress tensor, ε kk For volumetric strain, δ ij Let Kronecker function be used.

[0144] Considering the deformation gradient field The spatial derivative of the displacement field is actually derived from the strain tensor, stress tensor, and time-varying strain energy density gradient. This is composed of elements that output the deformation gradient field of each soil and rock layer. The time-varying strain energy density gradient is the derivative of the elastic strain energy stored per unit volume of soil and rock along the spatial direction, characterizing the spatiotemporal variation of strain energy accumulation or release. .

[0145] Next, the following weights are set for atmospheric correction: ( σ res The residual phase standard deviation is 0.27 rad (measured). σ total (The original phase standard deviation is 1.2 rad, as measured). For the structural inversion weights, the following weights are set: W str = corr ( ε ij , E (Correlation coefficient with elastic modulus, with a value range of...) Then, the weighted least squares algorithm is used to solve the three-dimensional deformation vector.

[0146] Then, using a 24-hour window length and sliding along the time axis with a 1-hour step, the following statistical characteristics of the deformation vector within the window were calculated:

[0147] (1) Displacement rate: The instantaneous rate is calculated based on the difference in the mean values ​​of adjacent windows.

[0148] (2) Acceleration: The acceleration is obtained by performing a second difference on the displacement rate sequence.

[0149] (3) Deep slip trend parameters: extract the rate of change of the slip surface dip angle. Δα With time-varying strain energy density gradient The rate of change of the sliding surface inclination angle refers to the rate of change of the inclination angle of the potential sliding surface (usually with the horizontal plane as a reference) over time.

[0150] Furthermore, displacement rate, acceleration, and trend parameters are bound to timestamps and organized in time sequence to generate sub-millimeter displacement data streams.

[0151] Finally, the sub-millimeter displacement data stream is organized and output according to the ISO 8601 time-coded format.

[0152] S4. Perform spatiotemporal correlation analysis on the displacement data stream with the acquired layered dip angle time series data and meteorological and hydrological data, update the slope stability assessment indicators in real time through dynamic baseline, and link with the augmented reality display terminal to automatically generate multi-level early warning instructions.

[0153] Furthermore, such as Figure 7 As shown, step S4 includes:

[0154] S41. The displacement data stream is decomposed into frequency bands to extract the first deformation component, the second deformation component, and the third deformation component, which include the 0.1-10Hz frequency band, the 10-50Hz frequency band, and the 50-100Hz frequency band. Simultaneously, the layered dip angle time series data collected by the dip angle sensors deployed on the slope are time-domain aligned to generate a joint observation sequence of deformation and dip angle.

[0155] S42. The acquired rainfall data is converted into a slope surface runoff intensity distribution map through spatial interpolation. The pore water pressure gradient field is inverted based on the acquired groundwater level monitoring data, and the pore water pressure gradient change rate within the preset time period is calculated, as well as its 24-hour change rate.

[0156] S43. Calculate the time-delay coupling strength between the first deformation component and the pore water pressure gradient change rate to obtain the delay time parameter of displacement lagging behind water pressure change. Analyze the spatial correlation region between the second deformation component and the runoff intensity distribution map to calibrate the deformation sensitive area and output the spatial coordinate set of the deformation sensitive area. Extract the structural surface activation characteristic frequency by detecting the frequency domain coherence between the third deformation component and the stratified dip angle time series data.

[0157] S44. Establish an assessment baseline based on joint observation sequences, including displacement acceleration threshold, tilt rate of change threshold, hysteresis time threshold, and structural surface activation hazard frequency band.

[0158] S45. Calculate the density of sensitive points in each slope zone based on the spatial coordinate set of the deformation sensitive area, and dynamically adjust the evaluation baseline according to the density of sensitive points. Among them, the displacement acceleration threshold increases proportionally with the increase of the density of sensitive points, the dip angle change rate threshold tightens linearly with the increase of the density of sensitive points, the hysteresis time threshold is corrected based on the density of sensitive points, and the structural surface activation danger frequency band expands adaptively with the increase of density.

[0159] S46. When multiple conditions are met, namely, the displacement acceleration is greater than the adjusted displacement acceleration threshold, the rate of change of inclination angle is greater than the adjusted rate of change of inclination angle threshold, the delay time parameter is less than the adjusted hysteresis time threshold, and the characteristic frequency of structural surface activation is in the adjusted dangerous frequency band of structural surface activation, the slope stability deterioration judgment is triggered.

[0160] S47. Based on the displacement exceeding the limit, the slope of the tilt angle change trend, and the activation characteristic frequency of the structural surface obtained in real time when the trigger judgment is made, combined with the evaluation baseline, the comprehensive instability index is calculated by a nonlinear weighted algorithm to generate a set of slope stability evaluation indicators.

[0161] In yet another specific embodiment, performing spatiotemporal correlation analysis includes:

[0162] First, the time-delay coupling strength between the first deformation component and the rate of change of pore water pressure gradient is calculated to obtain the time delay parameter of displacement lagging behind water pressure change. τ Analyze the spatial correlation region between the second deformation component and the runoff intensity distribution map to determine the spatial coordinate set of the deformation-sensitive area. ; Detect the frequency domain coherence of the third deformation component and the tilt sensor data, and extract the activation characteristic frequencies of the structural surface. .

[0163] Next, an initial assessment baseline was established based on the deformation-tilt joint observation sequence during historical stable periods, including: displacement acceleration threshold. Threshold for rate of change of tilt angle and the threshold of the lag time parameter .

[0164] Furthermore, based on the real-time acquired spatial coordinate set of the deformation-sensitive area... Calculate the density of sensitive points in each slope zone to reflect the degree of deformation sensitivity of each slope zone: In the formula, A k Slope zoning, P Count the sensitive points. A rea The area represents the zone.

[0165] Stability degradation is triggered when the following conditions are met simultaneously:

[0166] 1. Current displacement acceleration , (This serves as the baseline value for the distribution of sensitive points along the entire slope);

[0167] 2. Rate of change of tilt angle ;

[0168] 3. Lag Time Parameter ;

[0169] 4. Characteristic frequency Falling into dangerous frequency band ;

[0170] S45, Exceeding the displacement acceleration limit Inclination trend slope and structural surface activation characteristic frequencies Substitute into the following formula to calculate the comprehensive instability index: ;

[0171] In the formula, The density-dependent critical frequency threshold, as ρ k It increases linearly.

[0172] S48. Input the slope stability assessment index set into the augmented reality display terminal to visualize the slope deformation field and the three-dimensional geological model by superimposing them, and automatically generate multi-level early warning instructions that are linked with the highway operation management system.

[0173] Furthermore, such as Figure 8 As shown, step S48 includes:

[0174] S481. Synchronize real-time slope stability assessment indicators with augmented reality display terminals to achieve spatial registration and overlay display of slope deformation field and three-dimensional geological model.

[0175] S482. Generate a multi-level early warning signal based on the comprehensive instability index level, including at least one of the following: speed limit control, lane closure, and emergency support instructions.

[0176] S483. Encode multi-level early warning signals into traffic control protocol data packets and establish linkage with the gate control unit, information board display unit and maintenance dispatch platform of the highway operation management system through the 5G-V2X communication module.

[0177] In another specific embodiment, the three-dimensional deformation vector in the submillimeter-level displacement data stream is mapped to a color gradient-coded displacement field heatmap, and the comprehensive instability index is... I f Quantified into risk isosurfaces with varying degrees of transparency, the spatial coordinate set of deformation-sensitive areas is... Convert to spatial anchor point marker.

[0178] By using BeiDou / GNSS positioning data and inertial measurement unit data, the spatial reference coordinate system of the radar array and the actual slope scene is determined. The feature point matching algorithm is used to align the phase consistency radar image and the structural surface features of the BIM geological model, and the spatial coverage area of ​​the displacement field heat map in the augmented reality scene is updated in real time.

[0179] In augmented reality display terminals, displacement field heatmaps are overlaid onto real-world slope video streams at a spatial resolution of 0.5-2m. Based on the transparency parameters of risk isosurfaces, instability risk levels are rendered at corresponding geological layers. Pulse alarm effects are applied to spatial anchor point markers, with the effect frequency and characteristic frequency... f c Maintain a 1:3 harmonic ratio.

[0180] When displacement acceleration is detected a curr Exceeding the current dynamic grading threshold hour:

[0181] like It generates a level 3 early warning command, triggering the highway information board to issue a speed limit of 40km / h and a slope monitoring reminder.

[0182] like A level-two warning instruction is generated, and drone inspections are initiated to close the outer lane.

[0183] like A Level 1 early warning instruction is generated, and full-section closure is implemented and the emergency support system is activated.

[0184] Finally, the warning instructions are encoded into standard traffic control protocol data packets and linked with the gate control unit, information board display unit, and maintenance dispatch platform of the highway operation management system via a 5G-V2X communication module.

[0185] Specifically, taking the Chongqing-Kunming Expressway as an example, the multi-level early warning linkage effect shown in Table 4 is obtained:

[0186] Table 4 Multi-level Early Warning Linkage Data Table

[0187]

[0188] Additionally, embodiments of the present invention provide a dynamic beam radar monitoring and linkage early warning system for layered slopes of highways. The system is used to perform the method described above, including:

[0189] The radar hardware unit is used to perform layered scanning of the layered geological structure of the highway slope according to the layered scanning strategy issued by the edge computing unit, simultaneously acquiring phase coherent echo signals of the slope surface and internal structure, and uploading them to the edge computing unit. The radar hardware unit includes: a continuous wave radar array, consisting of multiple transmitting units and multiple receiving units forming a MIMO architecture; and a beam control module, connected to the continuous wave radar array via a high-speed bus, used to control the transmitting units of the radar array to transmit frequency-modulated signals according to the beam incidence angle adjustment sequence and scanning parameter constraints of each scanning cycle of the continuous wave radar array in the issued layered scanning strategy, and simultaneously adjusting the beam combining direction of the receiving units to match the spatial coverage requirements of each scanning layer.

[0190] The edge computing unit, connected to the radar hardware unit, includes a signal processing module, a deformation calculation module, and an early warning module. The signal processing module extracts multi-threshold scattering points from the phase coherent echo signal, combines the analysis and processing of a prior constraint model constructed from the acquired slope geometric features with a geological vegetation recognition network optimized by transfer learning, and eliminates interference signals caused by vehicle traffic to obtain a spatiotemporal coherent scattering point set. The deformation calculation module inputs the spatiotemporal coherent scattering point set into a preset deformation calculation pipeline, simultaneously performing differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics. The displacement data stream is then spatiotemporally correlated with the acquired layered dip angle time-series data and meteorological and hydrological data. The early warning module updates the slope stability assessment indicators in real time through a dynamic baseline and links with the augmented reality display terminal to automatically generate multi-level early warning instructions.

[0191] It also includes: a multi-source interface unit for accessing data from radar-visual fusion sensors and meteorological and hydrological equipment, and for linkage with the highway operation management system via the 5G-V2X communication protocol. The multi-source interface unit provides: a radar-visual fusion data channel supporting the access of calibration data from laser displacement gauges and fiber optic grating sensors; and a meteorological and hydrological adapter compatible with protocol data from rain gauges, groundwater level gauges, and pore water pressure gauges. The radar hardware unit and the edge computing unit interact via a data bus, and the edge computing subunit communicates with the early warning platform unit, with an end-to-end transmission latency of ≤10ms.

[0192] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0193] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the invention should be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0194] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from the spirit and scope of the invention. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, then this invention should also include these modifications and variations.

Claims

1. A dynamic beam radar monitoring and linkage early warning method for layered slopes of highways, characterized in that, include: A continuous wave radar array with adjustable beam angle is used to perform layered scanning of the layered geological structure of highway slopes, and simultaneously acquire phase coherent echo signals of the slope surface and internal structure. Multi-threshold scattering points are extracted from the phase coherent echo signal. The data is then analyzed and processed by a prior constraint model constructed from the acquired slope geometric features and a geological vegetation recognition network optimized by transfer learning. Interference signals caused by vehicle traffic are excluded to obtain a set of spatiotemporal coherent scattering points. The spatiotemporal coherent scattering point set is input into the preset deformation calculation pipeline, and differential interferometry, adaptive atmospheric disturbance correction and slope structure parameter inversion are executed simultaneously to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics. The displacement data stream is spatiotemporally correlated with the acquired layered dip angle time-series data and meteorological and hydrological data. Slope stability assessment indicators are updated in real time through dynamic baselines, and linked with augmented reality display terminals to automatically generate multi-level early warning commands. This includes: decomposing the displacement data stream into frequency bands to extract the first deformation component (0.1-10Hz), the second deformation component (10-50Hz), and the third deformation component (50-100Hz); simultaneously aligning the layered dip angle time-series data collected by the slope's deployed dip angle sensors in the time domain to generate a joint observation sequence of deformation and dip angle; and further processing the acquired dip angle data... Rainfall data is spatially interpolated to convert into a slope surface runoff intensity distribution map. Based on acquired groundwater level monitoring data, the pore water pressure gradient field is inverted, and the rate of change of pore water pressure gradient within a preset time period is calculated. The time-delay coupling strength between the first deformation component and the rate of change of pore water pressure gradient is calculated to obtain the time delay parameter indicating that displacement lags behind water pressure changes. The spatial correlation region between the second deformation component and the runoff intensity distribution map is analyzed to calibrate deformation-sensitive areas and output the spatial coordinate set of deformation-sensitive areas. The frequency domain coherence between the third deformation component and the stratified dip angle time series data is detected to extract the structural surface activation characteristic frequencies. Based on the joint observation sequence... An assessment baseline is established, including displacement acceleration threshold, dip angle change rate threshold, hysteresis time threshold, and structural surface activation hazard frequency band. Sensitive point density for each slope zone is calculated based on the spatial coordinate set of the deformation-sensitive area. The assessment baseline is dynamically adjusted according to the sensitive point density. Specifically, the displacement acceleration threshold increases proportionally with the increase in sensitive point density, the dip angle change rate threshold tightens linearly with the increase in sensitive point density, the hysteresis time threshold is corrected based on the sensitive point density, and the structural surface activation hazard frequency band adaptively expands with the increase in density. When the displacement acceleration is greater than the adjusted displacement acceleration threshold and the dip angle change rate is greater than the adjusted dip angle change rate threshold, ... When multiple conditions are met, such as the delay time parameter being less than the adjusted hysteresis time threshold and the structural surface activation characteristic frequency being in the adjusted structural surface activation danger frequency band, a slope stability deterioration judgment is triggered. Based on the displacement acceleration exceeding the limit, the slope trend of the tilt angle change, and the structural surface activation characteristic frequency obtained in real time at the time of triggering the judgment, combined with the evaluation baseline, a comprehensive instability index is calculated through nonlinear weighting to generate a slope stability assessment index set. The slope stability assessment index set is input into an augmented reality display terminal to visualize the slope deformation field and the three-dimensional geological model through virtual and real overlay, and multi-level early warning instructions are automatically generated in conjunction with the highway operation management system.

2. The dynamic beam radar monitoring and linkage early warning method for layered slopes of highways as described in claim 1, characterized in that, Multi-threshold scattering points were extracted from the phase coherent echo signal. The signal was then analyzed using a prior constraint model constructed from the acquired slope geometric features and a geological vegetation recognition network optimized through transfer learning. Interference signals caused by vehicle traffic were excluded, resulting in a spatiotemporal coherent scattering point set including: Based on the slope angle, layer thickness parameters and structural surface spatial orientation parameters extracted from the imported geological exploration data, a three-dimensional slope prior model is constructed, which includes constraints on the spatial density of scattering points, the continuity of displacement vectors, and the orientation of structural surfaces and radar line of sight. Based on the joint criterion of amplitude deviation index and phase stability threshold, candidate scattering points are screened from phase coherent echo signals. The spatial density constraint of the three-dimensional slope prior model is used to perform layer classification test on the candidate scattering points in order to eliminate outliers that deviate from the geological features of their respective layers. The scattering points that have undergone stratification are input into a pre-constructed transfer learning-optimized geological vegetation recognition network, which outputs the geological attribute classification results and confidence scores for each scattering point. Based on the vehicle Doppler parameters and acceleration observations obtained from the phase coherent echo signal, a dynamic target scattering model containing multiple scattering centers is constructed based on the radar transmitted signal wavelength, the vehicle's three-dimensional attitude, and the azimuth angle of the scattering point. Each scattering center in the model is characterized by the complex reflection coefficient, the time-varying radial distance, and the acceleration modulation frequency term. The time-varying radial distance includes the initial distance, the radial velocity, and the second motion component of the acceleration. A rectangular window function is introduced to limit the effective illumination time window of each scattering point. The time-frequency distribution feature spectrum of dynamic interference signals is generated by using a moving target scattering model. The time-varying scattering intensity envelope curve and the rate of change of frequency are extracted from the time-frequency distribution feature spectrum to establish a vehicle interference feature library. Cross-validation was performed based on the vehicle interference feature library and the geological attribute classification results. Transient interference signals caused by vehicle passage were identified and filtered out according to the preset vehicle interference judgment criteria to obtain a spatiotemporal coherent scattering point set. The vehicle interference judgment criteria included: the scattering intensity envelope was positively correlated with the vehicle speed and the envelope width matched the lane occupancy time; the frequency modulation rate change curve conformed to the nonlinear characteristics of the acceleration / braking mode; and the geological attribute classification result of the target area was a non-rock and soil structure.

3. The dynamic beam radar monitoring and linkage early warning method for layered slopes of highways as described in claim 1, characterized in that, The spatiotemporal coherent scattering point set is input into a preset deformation calculation pipeline, and differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion are performed simultaneously to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics, including: By receiving a set of spatiotemporal coherent scattering points through a deformation calculation pipeline pre-deployed on edge computing nodes, performing spatiotemporal reference unification processing, and establishing standardized observation data registered with the slope geographic coordinate system; Based on observational data, the following parallel operations are performed in the deformation calculation pipeline: scattering points of adjacent time phases are selected for phase difference calculation to generate an interferometric phase map covering the entire slope area; an atmospheric delay correction surface is constructed based on meteorological sensor data deployed on the slope to compensate for spatial variation errors in the interferometric phase map and obtain the residual phase field; the elastic modulus and Poisson's ratio from the preset geotechnical mechanics database are called, and the deformation gradient field of each soil and rock layer is calculated using the displacement-strain conversion relationship; The interferometric phase map, residual phase field and layer deformation gradient field are fused from multiple sources, and the three-dimensional deformation vector is solved by the confidence-weighted adaptive least squares algorithm. By analyzing the statistical characteristics of deformation vectors through a time-series sliding window, the time-series evolution characteristics including slope surface displacement rate, acceleration, and deep slip trend parameters are extracted, and sub-millimeter displacement data streams are generated in time sequence.

4. The dynamic beam radar monitoring and linkage early warning method for layered slopes of highways as described in claim 1, characterized in that, The slope stability assessment index set is input into the augmented reality display terminal to visualize the slope deformation field and the three-dimensional geological model through virtual and real overlay, and to automatically generate multi-level early warning instructions that are linked with the highway operation management system, including: Real-time slope stability assessment indicators are synchronized with augmented reality display terminals to achieve spatial registration and overlay display of slope deformation field and three-dimensional geological model; Based on the comprehensive instability index level, generate a multi-level early warning signal that includes at least one of the following: speed limit control, lane closure, and emergency support instructions; Multi-level early warning signals are encoded into traffic control protocol data packets, and linked with the gate control unit, information board display unit, and maintenance dispatch platform of the highway operation management system through a 5G-V2X communication module.

5. A dynamic beam radar monitoring and linkage early warning system for layered slopes of highways, characterized in that, The system is configured to perform the method as described in any one of claims 1-4, comprising: The radar hardware unit is used to perform layered scanning of the layered geological structure of the highway slope according to the layered scanning strategy issued by the edge computing unit, and simultaneously acquire the phase coherent echo signals of the slope surface and internal structure, and upload them to the edge computing unit. The edge computing unit, connected to the radar hardware unit, includes a signal processing module, a deformation calculation module, and an early warning module. The signal processing module extracts multi-threshold scattering points from the phase coherent echo signal, combines the analysis and processing of a prior constraint model constructed from the acquired slope geometric features with a geological vegetation recognition network optimized by transfer learning, and eliminates interference signals caused by vehicle traffic to obtain a spatiotemporal coherent scattering point set. The deformation calculation module inputs the spatiotemporal coherent scattering point set into a preset deformation calculation pipeline, simultaneously performing differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics. The displacement data stream is then spatiotemporally correlated with the acquired layered dip angle time-series data and meteorological and hydrological data. The early warning module updates the slope stability assessment indicators in real time through a dynamic baseline and links with the augmented reality display terminal to automatically generate multi-level early warning instructions.

6. The dynamic beam radar monitoring and linkage early warning system for layered slopes of highways as described in claim 5, characterized in that, The radar hardware unit includes: A continuous wave radar array, consisting of multiple transmitting units and multiple receiving units forming a MIMO architecture; The beam control module is connected to the continuous wave radar array via a high-speed bus. It is used to control the transmitting unit of the radar array to transmit frequency modulation signals according to the beam incident angle adjustment sequence and scanning parameter constraints of each scanning cycle of the continuous wave radar array in the layered scanning strategy. Under the scanning parameter constraints, it controls the transmitting unit to transmit frequency modulation signals according to the beam incident angle adjustment sequence, and synchronously adjusts the beam combining direction of the receiving unit to match the spatial coverage requirements of each scanning layer.

Images