Dynamic beam radar monitoring and linkage early warning method and system for layered slope of expressway

Abstract

The invention relates to the field of monitoring and early warning, in particular to a dynamic beam radar monitoring and linkage early warning method and system for a layered side slope of an expressway, and the method comprises the steps: carrying out the layered scanning of a layered geologic structure of the side slope, synchronously obtaining a phase coherent echo signal, carrying out the multi-threshold scattering point extraction, and carrying out the multi-threshold scattering point extraction; through analysis of a prior constraint model and a geological vegetation recognition network, interference signals caused by vehicle passing are eliminated to obtain a space-time coherent scattering point set, the space-time coherent scattering point set is input to a deformation calculation assembly line, differential interference measurement, adaptive atmospheric disturbance correction and slope structure parameter inversion are executed, and a displacement data stream is generated. And carrying out space-time correlation analysis on the layered inclination angle time sequence data and the meteorological and hydrological data, updating a slope stability evaluation index through a dynamic baseline, and generating an early warning instruction. According to the invention, a technical system from interference suppression and precise calculation to decision closed loop is formed, high reliability of road slope monitoring results is ensured, and intelligent support is provided for traffic safety and emergency management and control.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring and early warning, and in particular to a dynamic beam radar monitoring and linkage early warning method and system for highway layered slopes. Background Art

[0002] Currently, the safety monitoring of highway slopes mainly relies on manual inspection methods, that is, the slope state assessment is realized through regular on-site surveys, combined with visual inspections and simple instrument measurements. However, since manual inspections rely on personnel experience to subjectively judge the slope stability, there are problems with insufficient monitoring accuracy; at the same time, due to the time interval in the inspection cycle, there are blind spots in the slope state monitoring, and it is difficult to achieve all-weather continuous coverage.

[0003] In addition, the traditional manual method has limited ability to capture sudden minor deformations and cannot provide real-time feedback on the dynamic changes of slopes. Especially when the slope enters the stage of accelerated deformation, the early warning opportunity may be delayed due to monitoring lag. On the other hand, manual inspections require personnel to operate in complex terrain environments. Especially during bad weather such as heavy rain and strong winds, the safety risks in the slope area increase significantly, further highlighting the potential safety hazards to personnel.

[0004] Therefore, the existing technologies are difficult to meet the requirements 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 defects of manual inspections and improve the monitoring coverage and timeliness. Summary of the Invention

[0005] (I) Technical Problems to be Solved In view of the above-mentioned shortcomings and deficiencies of the existing technology, the present invention provides a dynamic beam radar monitoring and linkage early warning method and system for highway layered slopes, which solves the technical problems that the existing manual inspections have subjective errors, monitoring blind spots and safety risks, and cannot provide real-time and accurate early warning of highway slope stability.

[0006] (II) Technical Solutions To achieve the above object, the main technical solutions adopted by the present invention include: In the first aspect, an embodiment of the present invention provides a dynamic beam radar monitoring and linkage early warning method for highway layered slopes, including: Using a continuous wave radar array with an adjustable beam angle to perform layered scanning on the layered geological structure of the highway slope, and synchronously obtaining the phase coherent echo signals of the slope surface and internal structure; Performing multi-threshold scatter point extraction on the phase coherent echo signals, analyzing and processing them through a priori constraint models constructed from the obtained slope geometric features and a geological vegetation recognition network optimized by transfer learning, and excluding interference signals caused by vehicle passage to obtain a spatio-temporal coherent scatter point set; Input the spatio-temporal coherent scattering point set into a preset deformation solution pipeline, and synchronously perform differential interferometric measurement, adaptive atmospheric disturbance correction, and inversion of slope structure parameters to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics; Perform spatio-temporal correlation analysis on the displacement data stream, the obtained hierarchical dip angle time series data, and meteorological and hydrological data, and dynamically update the slope stability evaluation index through a dynamic baseline, and link with an augmented reality display terminal to automatically generate multi-level early warning instructions.

[0007] Optionally, use a continuous wave radar array with an adjustable beam angle to perform hierarchical scanning on the layered geological structure of the highway slope, and synchronously obtain the phase coherent echo signals of the slope surface and internal structure, including: According to the dielectric constant distribution and layer thickness parameters of the layered geological structure of the obtained highway slope, dynamically generate a hierarchical scanning strategy matching each rock and soil layer, construct a beamforming network, and determine the beam incident angle adjustment sequence for each scanning period of the continuous wave radar array in combination with preset scanning parameter constraints; According to the scanning parameter constraints, initialize the spatial arrangement parameters and working frequency band combinations of the transmitting / receiving units of the radar array, and construct a hardware configuration template corresponding to each scanning layer; Control the continuous wave radar array to switch the transmitting 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-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 and electromagnetic wave penetration depth of the radar array adaptively match with the change of the scanning layer; During the hierarchical scanning process, dynamically adjust the parameter combination including transmitting power, pulse repetition frequency, and coherent integration time based on the signal-to-noise ratio spectral characteristics of the echo signal, so that the signal phase stability of each scanning layer is maintained within a preset stable range; 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 interferometric measurement requirements.

[0008] Optionally, according to the dielectric constant distribution and layer thickness parameters of the layered geological structure of the obtained highway slope, dynamically generate a hierarchical scanning strategy matching each rock and soil layer, construct a beamforming network, and determine the beam incident angle adjustment sequence for each scanning period of the continuous wave radar array, including: Extract the dielectric constant distribution and layer thickness parameters of each rock and soil layer according to the imported geological exploration data of the highway slope; According to the dielectric constant distribution and layer thickness parameters of each rock and soil layer, a hierarchical scanning strategy function is constructed with the joint optimization goal of minimizing the inter-layer phase error and the total scanning time. The inter-layer phase error is quantified by the sum of squares of the phase delays of the electromagnetic wave propagation paths in each layer and the deviation from the corresponding threshold values. The total scanning time is calculated by the cumulative product of the beam switching frequency and the dwell time; In the hierarchical scanning strategy optimization function, configure the phase stability error threshold, the maximum interval of adjacent beam incident angles, and the upper limit of the single-cycle scanning time as hard constraint conditions, and solve through a constrained optimization algorithm to obtain the beam incident angle adjustment sequence that satisfies all constraints; Synchronously construct a beamforming network that matches the layered structure, and dynamically initialize the amplitude-phase distribution weights of the beamforming network according to the dielectric constant distribution of each rock and soil layer.

[0009] Optionally, perform orthogonal separation on the received multi-level echo signals, reconstruct the phase-consistent radar images 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, including: According to the inclination angle and layer thickness parameters of the slope layered geological structure extracted from the imported geological exploration data, allocate an orthogonal time-domain coding sequence and a spatial-domain beam pointing code set for each scanning layer; Construct a composite control matrix for spatio-temporal joint coding based on the time-domain orthogonal coding sequence and the spatial-domain beam pointing code set; Perform cross-correlation decoupling operations on the multi-level echo signals received by the radar array through the composite control matrix, separate the spatio-temporal coupling components of the echo signals of different layers, and generate the baseband signal channels of each scanning layer; Perform spatial-domain selective enhancement processing on each baseband signal channel using the spatial-domain beam pointing code set and suppress adjacent layer interference; Reconstruct the three-dimensional complex scattering characteristic distribution data of each scanning layer according to the enhanced signals, and eliminate the phase distortion caused by the multi-path effect through phase gradient consistency analysis; Perform full-aperture phase plane fitting and coherent accumulation on the data after eliminating the distortion to generate a spatially continuous phase-consistent radar image; Extract the set of scatter points that meet the preset spatial continuity criterion from the radar image sequence, and output a set of phase-coherent echo signals.

[0010] Optionally, perform multi-threshold scatter point extraction on the phase-coherent echo signals, analyze and process them through a prior constraint model constructed from the obtained slope geometric features and a geological vegetation recognition network optimized by transfer learning, and exclude the interference signals caused by vehicle passage to obtain a spatio-temporal coherent scatter point set, including: Construct a 3D slope prior model that includes constraints on the spatial density of scatter points, the continuity of displacement vectors, and the orientation of structural planes and radar line-of-sight directions, based on the slope gradient angle, layer thickness parameters, and spatial orientation parameters of structural planes extracted from the imported geological exploration data. Based on the combined criterion of the amplitude deviation index and the phase stability threshold, select candidate scatter points from the phase-coherent echo signals, and use the spatial density constraint of the 3D slope prior model to perform layer membership tests on the candidate scatter points to eliminate outliers that deviate from the layered geological characteristics of their respective layers. Input the scatter points that have passed the layer membership test into a pre-constructed geological vegetation recognition network optimized by transfer learning, and output the geological attribute classification results and confidence scores of each scatter point. Based on the vehicle Doppler parameters and acceleration observations obtained from the phase-coherent echo signals, construct a dynamic target scattering model that includes multiple scattering centers based on the radar transmission signal wavelength, vehicle 3D attitude, and scatter point azimuth angle. Each scattering center in the model is jointly characterized by a complex reflection coefficient, a time-varying radial distance, and an acceleration frequency modulation term. The time-varying radial distance includes the initial distance, the radial velocity, and the quadratic motion component of the acceleration, and a rectangular window function is introduced to limit the effective illumination time window of each scatter point. Generate the time-frequency distribution feature map of the dynamic interference signal through the motion target scattering model, extract the time-varying scattering intensity envelope curve and the frequency modulation rate change curve from the time-frequency distribution feature map, and establish a vehicle interference feature library. Based on cross-validation between the vehicle interference feature library and the geological attribute classification results, identify and filter out transient interference signals caused by vehicle passage according to the preset vehicle interference judgment criteria to obtain a spatio-temporally coherent scatter point set. The vehicle interference judgment criteria include: the scattering intensity envelope is positively correlated with the vehicle motion speed and the envelope width matches the lane occupancy time, the frequency modulation rate change curve conforms to the non-linear characteristics in the acceleration / braking mode, and the geological attribute classification result of the target area is a non-rock-soil structure body.

[0011] Optionally, input the spatio-temporally coherent scatter point set into a preset deformation solution pipeline to synchronously perform differential interferometric measurement, adaptive atmospheric disturbance correction, and slope structure parameter inversion, and generate a displacement data stream that includes 3D deformation vectors and temporal evolution characteristics, including: Receive the spatio-temporally coherent scatter point set through the deformation solution pipeline pre-deployed on the edge computing node, perform spatio-temporal reference unification processing, and establish standardized observation data registered with the slope geographic coordinate system. Based on the observed data, the following operations are performed in parallel in the deformation calculation pipeline: Select scattering points in adjacent time phases for phase difference calculation to generate an interferometric phase map covering the entire slope area; construct an atmospheric delay correction surface based on the meteorological sensor data deployed on the slope to compensate for the spatial variation error of the interferometric phase map and obtain the residual phase field; call the elastic modulus and Poisson's ratio in the preset geomechanical database and calculate the deformation gradient field of each rock and soil layer using the displacement and strain conversion relationship. Perform multi-source data fusion on the interferometric phase map, the residual phase field, and the deformation gradient field, and use the confidence-weighted adaptive least squares algorithm to calculate the three-dimensional deformation vector. Analyze the statistical characteristics of the deformation vector through a time-series sliding window, extract the time-series evolution characteristics including the surface displacement rate, acceleration, and deep slip trend parameters of the slope, and organize them in time series to generate a sub-millimeter displacement data stream.

[0012] Optionally, perform spatio-temporal correlation analysis on the displacement data stream, the obtained hierarchical dip angle time-series data, and the meteorological and hydrological data, update the slope stability evaluation index in real time through a dynamic baseline, and link with the augmented reality display terminal to automatically generate multi-level warning instructions, including: Decompose the displacement data stream into frequency bands to extract the first deformation component, the second deformation component, and the third deformation component. Synchronously perform time-domain alignment processing on the hierarchical dip angle time-series data collected by the dip angle sensors deployed on the slope to generate a joint observation sequence of deformation and dip angle. Convert the obtained rainfall data into a surface runoff intensity distribution map of the slope through spatial interpolation, and invert the pore water pressure gradient field based on the obtained groundwater level monitoring data and calculate the change rate of the pore water pressure gradient within a preset time period. Calculate the time-delay coupling intensity between the first deformation component and the change rate of the pore water pressure gradient to obtain the delay time parameter of the displacement lagging behind the 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 activation characteristic frequency of the structural plane by detecting the frequency-domain coherence between the third deformation component and the hierarchical dip angle time-series data. Establish an evaluation baseline based on the joint observation sequence, including the displacement acceleration threshold, the dip angle change rate threshold, the lag time threshold, and the activation dangerous frequency band of the structural plane. Calculate the sensitive point density of each slope partition based on the spatial coordinate set of the deformation-sensitive area, and dynamically adjust the evaluation baseline according to the sensitive point density. Among them, the displacement acceleration threshold increases proportionally with the increase of the sensitive point density, the dip angle change rate threshold tightens linearly with the increase of the sensitive point density, the lag time threshold is corrected based on the sensitive point density, and the activation dangerous frequency band of the structural plane expands adaptively with the increase of the density. When multiple conditions are met, including that the displacement acceleration is greater than the adjusted displacement acceleration threshold, the inclination change rate is greater than the adjusted inclination change rate threshold, the delay time parameter is less than the adjusted lag time threshold, and the structural plane activation characteristic frequency is within the adjusted dangerous frequency band of structural plane activation, the determination of slope stability deterioration is triggered; Based on the displacement overlimit, the inclination change trend slope, and the structural plane activation characteristic frequency obtained in real time when the trigger determination is made, combined with the evaluation baseline, the comprehensive instability index is calculated through non-linear weighting to generate a set of slope stability evaluation indicators: The set of slope stability evaluation indicators is input into the augmented reality display terminal for visualizing the virtual-real superposition of the slope deformation field and the three-dimensional geological model, and multi-level warning instructions linked to the highway operation management system are automatically generated.

[0013] Optionally, inputting the set of slope stability evaluation indicators into the augmented reality display terminal for visualizing the virtual-real superposition of the slope deformation field and the three-dimensional geological model, and automatically generating multi-level warning instructions linked to the highway operation management system includes: Synchronize the real-time slope stability evaluation indicators with the augmented reality display terminal to achieve the spatial registration and superposition display of the slope deformation field and the three-dimensional geological model; Generate multi-level warning signals including at least one of speed limit control, lane closure, and emergency support instructions according to the comprehensive instability index level; Encode the multi-level warning signals into traffic control protocol data packets, and establish a linkage with the gate control unit, the information board display unit, and the maintenance dispatching platform of the highway operation management system through the 5G-V2X communication module.

[0014] In a second aspect, an embodiment of the present invention provides a dynamic beam radar monitoring and linkage warning system for highway layered slopes. The system is used to execute the method described above and includes: A radar hardware unit for performing layered scanning on the layered geological structure of the highway slope according to the layered scanning strategy issued by the edge computing unit, synchronously acquiring the phase coherent echo signals of the slope surface and the internal structure, and uploading them to the edge computing unit; The edge computing unit is connected to the radar hardware unit and includes a signal processing module, a deformation solution module and an early warning module. The signal processing module is used to extract multi-threshold scattering points from the phase-coherent echo signal, combine the analysis and processing of the prior constraint model constructed by the acquired slope geometric features and the geological vegetation recognition network optimized by transfer learning, and eliminate the interference signals caused by vehicle passage to obtain a set of spatiotemporal coherent scattering points; the deformation solution module is used to input the set of spatiotemporal coherent scattering points into the preset deformation solution 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 time series evolution characteristics, and perform spatiotemporal correlation analysis on the displacement data stream with the acquired layered inclination time series data and meteorological and hydrological data. The early warning module is used to update the slope stability assessment index in real time through a dynamic baseline, and link with the augmented reality display terminal to automatically generate multi-level early warning instructions.

[0015] Optionally, the radar hardware unit includes: a continuous wave radar array, which is composed of a MIMO architecture consisting of multiple transmitting units and multiple receiving units; a beam control module, which is connected to the continuous wave radar array through a high-speed bus, and is used to control the transmitting units 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 issued layered scanning strategy, and synchronously adjust the beam synthesis direction of the receiving unit to match the spatial coverage requirements of each scanning layer.

[0016] (3) Beneficial effects The beneficial effects of the present invention are: First, by adopting a continuous wave radar array that can dynamically adjust the beam angle, we can break through the monitoring blind spot limitations of the traditional fixed scanning mode, achieve full-dimensional synchronous scanning of the layered slope surface and internal structure, and significantly improve the detection capability of hidden geological defects.

[0017] Secondly, a recognition network that combines multi-threshold scattering point extraction with transfer learning optimization intelligently distinguishes between areas of vegetation coverage and areas of geological structural anomalies while preserving the effective scattering characteristics of the rock and soil mass, effectively overcoming the risk of subjective error caused by differences in experience 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 method significantly reduces the risk of misjudgment due to interference such as vehicle vibration and metal reflections while preserving the true slope scattering points.

[0018] Furthermore, through the parallel processing architecture of the deformation solution pipeline, multi-task collaborative calculation of differential interferometry, atmospheric correction and structural inversion is realized, and the traditional step-by-step processing flow is optimized into real-time continuous solution, solving the problem of early warning lag caused by long manual inspection cycles.

[0019] On this basis, a multi-modal spatio-temporal correlation analysis integrating displacement data, dip angle time series, and hydro-meteorological parameters is carried out to construct a dynamic baseline update mechanism, enabling the stability evaluation index to adapt to the slope geological conditions and environmental changes, and avoiding the defect that the fixed threshold system is insensitive to local risks.

[0020] Finally, through the visual linkage with the augmented reality system and the automatic generation of multi-level warning instructions, a full-chain technical system from interference suppression, precise calculation to decision-making closed-loop is formed to ensure the high reliability of the highway slope monitoring results and provide intelligent support for driving safety and emergency management. Brief Description of the Drawings

[0021] Figure 1 It is a schematic flowchart of the method provided by the embodiment of the present invention; Figure 2 It is a specific schematic flowchart of step S1 of the method provided by the embodiment of the present invention; Figure 3 It is a specific schematic flowchart of step S11 of the method provided by the embodiment of the present invention; Figure 4 It is a specific schematic flowchart of step S15 of the method provided by the embodiment of the present invention; Figure 5 It is a specific schematic flowchart of step S2 of the method provided by the embodiment of the present invention; Figure 6 It is a specific schematic flowchart of step S3 of the method provided by the embodiment of the present invention; Figure 7 It is a specific schematic flowchart of step S4 of the method provided by the embodiment of the present invention; Figure 8 It is a specific schematic flowchart of step S48 of the method provided by the embodiment of the present invention. Detailed Embodiments

[0022] To better explain the present invention for easy understanding, the present invention will be described in detail below with reference to the accompanying drawings through specific embodiments.

[0023] Such as Figure 1As shown in the figure, a dynamic beam radar monitoring and linkage warning method for highway layered slopes proposed by an embodiment of the present invention includes: using a continuous wave radar array with an adjustable beam angle to perform layered scanning on the layered geological structure of the highway slope, and synchronously obtaining the phase coherent echo signals of the slope surface and internal structure; extracting multi-threshold scattering points from the phase coherent echo signals, analyzing and processing them through a priori constraint models constructed from the obtained slope geometric features and a geological vegetation recognition network optimized by transfer learning, and excluding interference signals caused by vehicle passage to obtain a spatio-temporal coherent scattering point set; inputting the spatio-temporal coherent scattering point set into a preset deformation calculation pipeline, synchronously 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 features; performing spatio-temporal correlation analysis on the displacement data stream, the obtained layered dip angle time series data, and meteorological and hydrological data, updating the slope stability evaluation index in real time through a dynamic baseline, and linking with an augmented reality display terminal to automatically generate multi-level warning instructions.

[0024] First of all, by using a continuous wave radar array with a dynamically adjustable beam angle, the present invention breaks through the monitoring blind area limitation of the traditional fixed scanning mode, realizes the full-dimensional synchronous scanning of the surface and internal structure of the layered slope, and significantly improves the detection ability of hidden geological defects.

[0025] Secondly, combined with the multi-threshold scattering point extraction and the recognition network optimized by transfer learning, while retaining the effective scattering characteristics of the rock and soil mass, it can intelligently distinguish the vegetation-covered area and the geological structure abnormal area, effectively overcoming the subjective error risk caused by experience differences in manual discrimination. Particularly crucial is the introduction of a dynamic interference suppression mechanism, which can effectively identify and eliminate the transient interference signals generated by vehicle passage. Compared with the traditional fixed threshold filtering method, the present invention significantly reduces the misjudgment risk caused by interference such as vehicle vibration and metal reflection while retaining the real slope scattering points.

[0026] Furthermore, through the parallel processing architecture of the deformation calculation pipeline, the multi-task collaborative calculation of differential interferometry, atmospheric correction, and structure inversion is realized, and the traditional step-by-step processing flow is optimized into real-time continuous calculation, solving the problem of early warning lag caused by the long cycle of manual inspection.

[0027] On this basis, the multi-modal spatio-temporal correlation analysis of displacement data, dip angle time series, and hydrometeorological parameters is integrated to construct a dynamic baseline update mechanism, enabling the stability evaluation index to adapt to the slope geological conditions and environmental changes, and avoiding the defect of slow response of the fixed threshold system to local risks.

[0028] Finally, through the visual linkage with the augmented reality system and the automatic generation of multi-level warning instructions, a full-chain technical system from interference suppression, precise calculation to decision-making closed-loop is formed to ensure the high reliability of the highway slope monitoring results and provide intelligent support for driving safety and emergency management and control.

[0029] In order to better understand the above technical solution, the exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the 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 by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be understood more clearly and thoroughly, and the scope of the present invention can be fully conveyed to those skilled in the art.

[0030] Specifically, the embodiment of the present invention provides a dynamic beam radar monitoring and linkage warning method for highway layered slopes, which includes: S1. Use a continuous wave radar array with adjustable beam angles to perform layered scanning on the layered geological structure of the highway slope, and synchronously obtain the phase coherent echo signals of the slope surface and internal structure.

[0031] Further, as Figure 2 shown, step S1 includes: S11. According to the dielectric constant distribution and layer thickness parameters of the layered geological structure of the highway slope obtained, dynamically generate a layered scanning strategy matching each rock and soil layer position and construct a beamforming network. Combining the preset scanning parameter constraints, determine the beam incident angle adjustment sequence of each scanning cycle of the continuous wave radar array.

[0032] Even further, as Figure 3 shown, step S11 includes: S111. According to the imported geological exploration data of the highway slope, extract the dielectric constant distribution and layer thickness parameters of each rock and soil layer position. Among them, the dielectric constant distribution changes with depth and is determined by the electromagnetic characteristics of different rock and soil types, and the layer thickness parameter characterizes the vertical extension scale of each rock and soil layer.

[0033] S112. According to the dielectric constant distribution and layer thickness parameters of each rock and soil layer position, construct a layered scanning strategy function with the joint optimization goal of minimizing the interlayer phase error and the total scanning time, where the interlayer phase error is quantified by the sum of squares of the phase delay of the electromagnetic wave propagation path of each layer and the deviation from the corresponding threshold, and the total scanning time is calculated by the cumulative product of the beam switching frequency and the dwell time.

[0034] S113. Configure the phase stability error threshold, the maximum interval of incident angles of adjacent beams, and the upper limit of the single-cycle scanning time as hard constraint conditions in the hierarchical scanning strategy optimization function, and solve through a constraint optimization algorithm to obtain a beam incident angle adjustment sequence that satisfies all constraints; Among them, the hierarchical scanning strategy optimization function realizes the adaptive optimization of scanning parameters through the depth coupling of geological parameters and the electromagnetic propagation model. Based on the dielectric constant distribution and layer thickness parameters of the slope layered geological structure, the hierarchical scanning strategy optimization function is constructed as: ; In the formula, θ n is the incident angle of the n th scanning beam, f m is the carrier frequency of the m th frequency point, λ ( f m ) is the wavelength corresponding to the frequency f m , d k is the layer thickness parameter of the slope layered geological structure, is the dielectric constant distribution of the slope layered geological structure, is the phase stability threshold, T scan is the total scanning time, α t is the time weight coefficient, z q represents the depth coordinate of the q th geological layer, Q is the number of geological layers.

[0035] In the process of solving by the constraint optimization algorithm, first construct a candidate solution set based on the initially randomly generated beam incident angle sequence, and transform the phase stability error, adjacent beam angle interval, and scanning time constraints of each layer into penalty terms of the objective function through the penalty function method to form an optimization model that comprehensively considers detection accuracy and timeliness. Subsequently, the sequential quadratic programming algorithm can be used to iteratively optimize the objective function: in each iteration, calculate the phase error gradient and constraint violation amount corresponding to the current solution, dynamically adjust the incident angle step size and frequency combination, and determine the optimal step direction through linear search to meet the constraint boundary conditions. Finally, generate a beam incident angle adjustment sequence that satisfies the following conditions for the scanning parameter constraints: the phase stability error of each layer ≤ 0.08 rad; the interval of incident angles of scanning beams between adjacent layers ≤ 2°; the single-cycle scanning time seconds. Thus, dynamically control the beam pointing of the radar array according to the output scanning parameters to realize the adaptive optimization of the electromagnetic characteristics of slope hierarchical scanning.

[0036] S114. Synchronously construct a beamforming network that matches the layered structure, and initialize the amplitude-phase distribution weights of the beamforming network. The beamforming network is a signal processing architecture in a radar system that realizes the spatial directivity synthesis of electromagnetic beams by regulating the amplitude-phase distribution weights between the transmitting / receiving units. When initializing the amplitude-phase distribution weights of the beamforming network, pre-calculate the ideal beam directivity diagrams of each layer with the help of an electromagnetic field simulation model, and fit the amplitude-phase parameters through the least squares approximation algorithm to make the main lobe width and sidelobe suppression ratio of the initial beam meet the inter-layer resolution requirements.

[0037] S12. According to the scanning parameter constraints, initialize the spatial arrangement parameters and working frequency band combinations of the transmitting / receiving units of the radar array, and construct the hardware configuration templates corresponding to each scanning layer.

[0038] In this step, define the spatial arrangement parameters and working frequency band combinations of the transmitting / receiving units of the radar array according to the scanning parameter constraints; For the geological characteristic differences of different layers, construct the hardware configuration templates respectively, as shown in Table 1, Table 2 and Table 3 below: Table 1 Surface Scanning Template

[0039] Table 2 Middle Layer Scanning Template

[0040] Table 3 Deep Layer Scanning Template

[0041] S13. Control the continuous wave radar array to switch the transmitting beam angle according to the beam incident angle adjustment sequence. When switching to each scanning layer, load the hardware configuration template corresponding to the current layer, and adjust the amplitude-phase distribution weights 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 are adaptively matched as the scanning layer changes.

[0042] In this step, at the moment of switching to the target scanning layer, call the corresponding hardware configuration template, First, perform frequency configuration: gradually change the frequency layer by layer from 14 GHz on the surface layer to 12 GHz on the middle layer and then to 10 GHz on the deep layer; Secondly, adjust the array element activation mode: all array elements work on the surface layer → selectively sleep the edge units on the middle layer → enable the sparse array on the deep layer; then, dynamically adjust the amplitude-phase weights: according to the dielectric constant and target depth of the current layer, perform amplitude optimization, and adjust the excitation amplitude of the array elements through the least squares approximation algorithm to make the main lobe width of the beam adapt to the layer thickness.

[0043] S14. During the hierarchical scanning process, dynamically adjust the parameter combination including the transmit power, pulse repetition frequency, and coherent integration time based on the SNR spectrum characteristics of the echo signal, so that the signal phase stability of each scanning layer is maintained within a preset stable range. Preferably, the signal phase stability of each scanning layer is maintained within a threshold of 0.1 rad.

[0044] 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.

[0045] Furthermore, as Figure 4 shown, step S15 includes: S151. According to the dip angle and layer thickness parameters of the slope layered geological structure extracted from the imported geological exploration data, assign an orthogonal time-domain coding sequence and a spatial-domain beam pointing code set to each scanning layer. The time-domain coding sequences of adjacent layers satisfy the cross-correlation peak-to-average ratio ≤ -30 dB, and the spatial-domain beam pointing code includes phase modulation parameters in the azimuth and elevation directions to ensure that the main lobes of each layer do not overlap (angle interval ≥ 1.5 times the beam width).

[0046] S152. Construct a composite control matrix for space-time joint coding based on the time-domain orthogonal coding sequence and the spatial-domain beam pointing code set.

[0047] S153. Perform cross-correlation decoupling operations on the multi-level echo signals received by the radar array through the composite control matrix, separate the space-time coupling components of the echo signals of different layers, and generate the baseband signal channels of each scanning layer.

[0048] S154. Perform spatial-domain selective enhancement processing on each baseband signal channel using the spatial-domain beam pointing code set and suppress adjacent layer interference to achieve a sidelobe level ≤ -25 dB.

[0049] S155. Reconstruct the three-dimensional complex scattering characteristic distribution data of each scanning layer based on the enhanced signals, and eliminate the phase distortion caused by the multipath effect through phase gradient consistency analysis. The phase gradient change rate is controlled within 0.2 radians per wavelength.

[0050] S156. Perform full-aperture phase plane fitting and coherent integration on the data with eliminated distortion to generate a spatially continuous phase-consistent radar image.

[0051] S157. Extract the set of scatter points that meet the preset spatial continuity criteria (area density ≥ 5 points / m², phase standard deviation ≤ 0.05λ) from the radar image sequence, and output a set of phase-coherent echo signals.

[0052] In a specific embodiment, a composite control matrix that fuses the time-domain coding sequence and the spatial-domain beam pointing code combines the orthogonal modulation signal at the radar transmitting end with the adaptive beamforming at the receiving end through spatio-temporal joint coding, achieving spatial-domain selective enhancement and time-domain interference suppression of echo signals at different scanning levels, and providing precise beam pointing control and signal separation capabilities for layer scanning. The specific formula is as follows: , where is the time-domain orthogonal coding matrix ( T is the coding sequence length, taking values as powers of 2. The longer the coding, the higher the range resolution, but the computational complexity increases), is the spatial-domain beam pointing code matrix, O is the number of transmitting units, M is the number of receiving units, is the Kronecker product, which realizes the dimension expansion of time-space coding through the Kronecker product, making the matrix dimension strictly match the radar architecture. is the Hadamard product, used to represent element-wise multiplication, is the array manifold matrix, a phase response matrix that describes the relationship between the array geometry and the beam pointing. Each element corresponds to the spatial phase delay of the array element, θ is the azimuth angle of the beam pointing, ϕ is the elevation angle of the beam pointing. The accurate modeling of the array manifold matrix can compensate for the wavefront distortion caused by the layered structure.

[0053] The three-dimensional complex scattering characteristic distribution data is a three-dimensional spatial distribution matrix that characterizes the scattering characteristics of each layer of the slope. For the l th scanning level, its three-dimensional spatial distribution matrix is: , is the echo data of the o th transmitting unit and the m th receiving unit, containing K complex signals of snapshots, is the beamforming sub-matrix corresponding to the l and ( p,m ) transmitting and receiving pairs, containing azimuth and elevation modulation parameters, H represents the conjugate transpose operation. The cascaded reconstruction process of the three-dimensional complex scattering matrix effectively separates the scattering contributions of each layer.

[0054] Next, for the reconstructed complex scattering matrix S lCalculate the phase gradient pixel by pixel, constrain the rate of change of the gradient not to exceed a threshold (such as 0.2 radians per wavelength), perform phase compensation on the areas exceeding the threshold, and minimize the phase difference between adjacent pixels through iterative optimization to output a three-dimensional complex scattering characteristic matrix with distortion eliminated, and its phase continuity meets the requirements of interferometric measurement.

[0055] S2. Extract multi-threshold scattering points from the phase-coherent echo signals, analyze and process them through a prior constraint model constructed from the obtained slope geometric features and a geological vegetation recognition network optimized by transfer learning, and exclude the interference signals caused by vehicle passage to obtain a spatio-temporally coherent scattering point set.

[0056] Furthermore, as Figure 5 shown, step S2 includes: S21. Construct a three-dimensional slope prior model including constraints on the spatial density of scattering points, the continuity of displacement vectors, and the orientation of the structural plane and the radar viewing direction based on the slope angle, layer thickness parameter, and spatial orientation parameter of the structural plane extracted from the imported geological exploration data. Specifically, a prior model including the following constraint conditions is established: the spatial density threshold of scattering points in each rock and soil layer is calculated according to the formula calculate, h i is the thickness, the direction continuity tolerance of the displacement vector between adjacent layers ≤ 15°, the rate of change of the displacement modulus length ≤ 30%, and the allowable deviation range of the angle between the structural plane strike and the radar viewing direction (±20°, to avoid the blind area of specular reflection).

[0057] S22. Based on the joint criterion of the amplitude deviation index and the phase stability threshold, screen candidate scattering points from the phase-coherent echo signals, and use the spatial density constraint of the three-dimensional slope prior model to perform layer membership tests on the candidate scattering points to eliminate outliers deviating from the layered geological characteristics of the layer to which they belong. Perform double-condition screening using the amplitude deviation index threshold (≥0.7) and the phase stability threshold (≥0.85); perform layer membership tests based on the spatial density threshold to eliminate scattering points with density deviation values exceeding ±30%.

[0058] S23. Input the scattering points that have passed the layer membership test into a pre-constructed geological vegetation recognition network optimized by transfer learning, and output the geological attribute classification results and confidence scores of each scattering point.

[0059] Specifically, the geological vegetation recognition network uses the pre-trained ResNet-50 as the backbone network, maps the natural image feature space to the radar scattering feature space of the phase-coherent echo signals through a domain adaptation module, and introduces an attention mechanism to extract features of the geological structure edges and vegetation textures by integrating the CBAM module after Stage3 of ResNet-50. Finally, the geological vegetation recognition network outputs the geological type labels (rock / soil / vegetation) and confidence scores (in the range of 0-1) of each scattering point.

[0060] S24. Based on the vehicle Doppler parameters and acceleration observation values obtained from the phase-coherent echo signals, construct a dynamic target scattering model containing multiple scattering centers based on the radar transmission signal wavelength, vehicle three-dimensional attitude, and scattering point azimuth angle. Each scattering center of the model is jointly characterized by a complex reflection coefficient, time-varying radial distance, and acceleration frequency modulation term. The time-varying radial distance includes the initial distance, radial velocity, and quadratic motion component of acceleration, and a rectangular window function is introduced to limit the effective illumination time window of each scattering point. Among them, the moving target scattering model is: ; In the formula, S m ( t ) is the synthetic echo signal of the vehicle target for the m th receiving unit of the radar array, which is composed of the superposition of echoes from multiple scattering centers. t is the time variable, representing the cumulative time of the radar signal from the start of observation to the current moment. N is the number of independent scattering points on the vehicle. For example, the front of the vehicle, the body, the wheels, etc. can all be regarded as different scattering centers. N ≥1, is the complex reflection coefficient of the n th scattering point. is the complex exponential function. Here, j is the imaginary unit, which is used to represent the complex phase of the signal. Considering the Doppler frequency shift of the vehicle three-dimensional attitude. v is the radial velocity of the vehicle along the radar line of sight. is the radar transmission signal wavelength. θ n is the azimuth angle of the n th scattering point relative to the radar line of sight. is the frequency modulation term caused by acceleration. a is the radial acceleration of the vehicle along the radar line of sight. is the time-varying distance model containing acceleration. is the radial distance at the initial moment t =0, v 0 is the initial velocity. a is the acceleration. is used to characterize the effective illumination time window of the scattering point. is the time delay when the n th scattering point enters the radar beam. T n is the dwell time of the scattering point in the radar beam. By introducing the acceleration term, three-dimensional attitude parameters, and multi-scattering center architecture, this model effectively solves the model mismatch problem of traditional methods in complex traffic scenarios and provides theoretical support for the subsequent accurate elimination of interference signals.

[0061] S25. Generate the time-frequency distribution feature map of the dynamic interference signal through the moving target scattering model, extract the time-varying scattering intensity envelope curve and the chirp rate change rate curve from the time-frequency distribution feature map, and establish a vehicle interference feature library.

[0062] S26. Perform cross-validation based on the vehicle interference feature library and the geological attribute classification results, identify and filter out the transient interference signals caused by vehicle passage according to the preset vehicle interference judgment criteria, and obtain the spatio-temporal coherent scatter point set. Among them, the vehicle interference judgment criteria include: the scattering intensity envelope is positively correlated with the vehicle moving speed and the envelope width matches the lane occupancy time (the error between the envelope width and the lane occupancy time ≤ 0.5 seconds), the chirp rate change rate curve conforms to the non-linear characteristics in the acceleration / braking mode (the correlation coefficient between the chirp rate change rate curve and the theoretical acceleration / braking mode ≥ 0.8, where the theoretical acceleration / braking mode refers to the non-uniform modulation characteristics presented by the radar echo signal in the time-frequency domain caused by the vehicle's motion state during acceleration or braking, and is obtained by fitting historical data collection) and the geological attribute classification result of the target area is non-rock and soil structure body (confidence ≥ 0.9).

[0063] S3. Input the spatio-temporal coherent scatter point set into the preset deformation solution pipeline, synchronously perform differential interferometry, adaptive atmospheric disturbance correction and slope structure parameter inversion, and generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics.

[0064] Further, as Figure 6 shown, step S3 includes: S31. Receive the spatio-temporal coherent scatter point set through the deformation solution pipeline pre-deployed on the edge computing node, perform spatio-temporal reference unification processing, and establish standardized observation data registered with the slope geographic coordinate system.

[0065] S32. Based on the observation data, perform in parallel in the deformation solution pipeline: select adjacent-phase scatter points for phase difference calculation to generate an interference phase map covering the entire slope area; construct an atmospheric delay correction surface based on the meteorological sensor data deployed on the slope to compensate for the spatial variation error of the interference phase map and obtain the residual phase field; call the elastic modulus and Poisson ratio of the preset geotechnical mechanics database, and calculate the deformation gradient field of each rock and soil layer using the displacement-strain conversion relationship.

[0066] S33. Perform multi-source data fusion on the interference phase map, the residual phase field and the deformation gradient field, and use the confidence-weighted adaptive least squares algorithm to solve the three-dimensional deformation vector.

[0067] S34. Analyze the statistical characteristics of the deformation vector through a time-series sliding window, extract the time-series evolution characteristics including the surface displacement rate, acceleration, and deep slip trend parameters of the slope, and generate a sub-millimeter displacement data stream organized by time series.

[0068] In a specific embodiment, the parallel execution channels in the deformation solution pipeline include: (1) Differential interferometry channel: Select adjacent phase scattering points for millimeter-wave interferometry to generate an interferometric phase map covering the entire slope.

[0069] (2) Atmospheric correction channel: Construct an atmospheric delay correction surface based on the meteorological sensors (temperature / humidity / pressure) deployed on the slope. The formula is: ; where is the phase delay of the radar signal caused by the change in atmospheric refractive index, P q is the atmospheric pressure, T w is the temperature, H s is the humidity, k 1, k 2, k 3 are the delay correction coefficients for the corresponding parameters respectively, k 1 = 0.12 rad / kPa, k 2 = 0.08 rad / °C, k 3 = 0.05 rad / %RH, and the residual error of the corrected interferometric phase ≤ 0.3 rad.

[0070] (3) Structural inversion channel: Call the elastic modulus E and Poisson's ratio in the geotechnical mechanics database γ, Input the displacement field obtained by radar interferometry inversion u = u x , u y , u z T , u x is the displacement component of the slope in the east-west direction, u y is the displacement component of the slope in the north-south direction, u z is the displacement component of the slope in the vertical direction. Based on the small deformation assumption, calculate the strain tensor where is the displacement vector component u i partial derivative with respect to the coordinate x j , is the displacement vector component​u j Partial derivative with respect to coordinates x i , i , j = x , y , z , i and j are free indices representing three directions in the coordinate system x , y , z , when i , j = x , y , z appears in the formula, it means that all possible i , j combinations need to be calculated.

[0071] Use the generalized Hooke's law to construct a displacement-strain conversion model: , σ ij is the stress tensor, ε kk is the volumetric strain, δ ij is the Kronecker function.

[0072] Considering the deformation gradient field is the spatial derivative of the displacement field, actually composed of the strain tensor, the stress tensor and the time-varying strain energy density gradient , thus outputting the deformation gradient field of each rock and soil layer. Among them, the time-varying strain energy density gradient is the derivative of the elastic strain energy stored in the rock and soil body per unit volume along the spatial direction, characterizing the spatio-temporal variation characteristics of the accumulation or release of strain energy, .

[0073] Next, for the atmospheric correction weight, set the following weights: ( σ res is the standard deviation of the residual phase, measured as 0.27 rad, σ total is the standard deviation of the original phase, measured as 1.2 rad); for the structural inversion weight, set the following weights: W str = corr ( ε ij , E ) (the correlation coefficient with the elastic modulus, the value range is , and then use the weighted least squares algorithm to solve the three-dimensional deformation vector.

[0074] Then, using a 24-hour window length and sliding along the time axis with a step length of 1 hour, the following statistical features of the deformation vector within the window are calculated: (1) Displacement rate: The instantaneous rate is calculated based on the mean difference between adjacent windows.

[0075] (2) Acceleration: Acceleration is obtained by performing a quadratic difference on the displacement rate sequence.

[0076] (3) Deep slip trend parameters: Extract the sliding surface inclination change rate Δα Time-varying strain energy density gradient The sliding surface inclination change rate refers to the rate at which the inclination angle of the potential sliding surface (usually with reference to the horizontal plane) changes over time.

[0077] Furthermore, the displacement rate, acceleration and trend parameters are bound to the timestamp and organized in time sequence to generate a sub-millimeter displacement data stream.

[0078] Finally, the submillimeter displacement data stream is organized and output according to the ISO 8601 time coding format.

[0079] S4. Perform spatiotemporal correlation analysis on the displacement data stream, the acquired layered inclination time series data, and the meteorological and hydrological data. Update the slope stability assessment index in real time through the dynamic baseline, and link it with the augmented reality display terminal to automatically generate multi-level warning instructions.

[0080] Further, if Figure 7 As shown, step S4 includes: S41. Decompose the displacement data stream into frequency bands to extract the first deformation component including the 0.1-10 Hz frequency band, the second deformation component including the 10-50 Hz frequency band, and the third deformation component including the 50-100 Hz frequency band. Simultaneously, perform time domain alignment processing on the layered inclination time series data collected by the inclination sensors deployed on the slope to generate a joint observation sequence of deformation and inclination.

[0081] S42. Convert the acquired rainfall data into a slope surface runoff intensity distribution map through spatial interpolation, and invert the pore water pressure gradient field based on the acquired groundwater level monitoring data and calculate the pore water pressure gradient change rate within a preset time period, and calculate its 24-hour change rate.

[0082] S43. Calculate the time-delay coupling strength between the first deformation component and the rate of change of the pore water pressure gradient to obtain the delay time parameter of the displacement lagging behind the water pressure change; analyze the spatial correlation area between the second deformation component and the runoff intensity distribution diagram to calibrate the deformation sensitive area and output the spatial coordinate set of the deformation sensitive area; and extract the characteristic frequency of structural surface activation by detecting the frequency domain coherence between the third deformation component and the layered inclination time series data.

[0083] S44. Establish an evaluation baseline including displacement acceleration threshold, inclination change rate threshold, lag time threshold, and structural plane activation dangerous frequency band based on the joint observation sequence.

[0084] S45. Calculate the sensitive point density of each slope section based on the spatial coordinate set of the deformation-sensitive area, and dynamically adjust the evaluation baseline according to the sensitive point density. Among them, the displacement acceleration threshold increases proportionally with the increase of the sensitive point density, the inclination change rate threshold tightens linearly with the increase of the sensitive point density, the lag time threshold is corrected based on the sensitive point density, and the structural plane activation dangerous frequency band expands adaptively with the increase of the density.

[0085] S46. When multiple conditions are met, including the displacement acceleration being greater than the adjusted displacement acceleration threshold, the inclination change rate being greater than the adjusted inclination change rate threshold, the delay time parameter being less than the adjusted lag time threshold, and the structural plane activation characteristic frequency being within the adjusted structural plane activation dangerous frequency band, trigger the determination of slope stability deterioration.

[0086] S47. Based on the displacement overlimit, inclination change trend slope, and structural plane activation characteristic frequency obtained in real time when the trigger determination is made, combined with the evaluation baseline, calculate the comprehensive instability index through a non-linear weighting algorithm to generate a slope stability evaluation index set. <![CDATA[ ]]><![CDATA[

[0087] ]]>In another specific embodiment, performing spatio-temporal correlation analysis includes: First, calculate the time-delay coupling strength between the first deformation component and the change rate of pore water pressure gradient to obtain the delay time parameter of the displacement lagging behind the 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 between the third deformation component and the inclination sensor data to extract the structural plane activation characteristic frequency .

[0088] Next, establish an initial evaluation baseline based on the deformation-inclination joint observation sequence during the historical stable period, including: displacement acceleration threshold , inclination change rate threshold , and lag time parameter threshold .

[0089] Furthermore, calculate the sensitive point density of each slope section according to the spatial coordinate set of the deformation-sensitive area obtained in real time to reflect the deformation sensitivity of each slope section: , where A k is the slope section, P is the sensitive point count, A rea is the sectional area.

[0090] When the following conditions are met simultaneously, the determination of deteriorated stability is triggered: 1. The current displacement acceleration , is the distribution reference value of the sensitive points of the entire slope); 2. The inclination change rate ; 3. The lag time parameter ; 4. The characteristic frequency falls into the dangerous frequency band ; S45. Substitute the displacement acceleration overlimit , the inclination trend slope and the activation characteristic frequency of the structural plane into the following formula to calculate the comprehensive instability index: ; In the formula, is the critical frequency threshold related to density, which linearly increases with ρ k increasing.

[0091] S48. Input the slope stability evaluation index set into the augmented reality display terminal for visualizing the virtual-real superposition of the slope deformation field and the three-dimensional geological model, and automatically generate multi-level warning instructions linked to the highway operation management system.

[0092] Furthermore, as Figure 8 shown, step S48 includes: S481. Synchronize the real-time slope stability evaluation index with the augmented reality display terminal to achieve the spatial registration and superposition display of the slope deformation field and the three-dimensional geological model.

[0093] S482. Generate multi-level warning signals including at least one of speed limit control, lane closure, and emergency support instructions according to the comprehensive instability index level.

[0094] S483. Encode the multi-level warning signals into traffic control protocol data packets and establish a linkage with the gate control unit, information board display unit, and maintenance scheduling platform of the highway operation management system through the 5G-V2X communication module.

[0095] In yet another specific embodiment, map the three-dimensional deformation vector in the sub-millimeter-level displacement data stream into a displacement field heat map encoded with color gradients, quantify the comprehensive instability index I f into risk iso-surfaces with different transparencies, and convert the spatial coordinate set of the deformation-sensitive area into spatial anchor point markers.

[0096] Determine the spatial reference coordinate system of the radar array and the slope real scene through Beidou / GNSS positioning data and inertial measurement unit data, align the structural plane features of the phase consistency radar image and the BIM geological model using the feature point matching algorithm, and update the spatial coverage area of the displacement field heat map in the augmented reality scene in real time.

[0097] In the augmented reality display terminal, superimpose the displacement field heat map on the slope real scene video stream with a spatial resolution of 0.5 - 2m, render the instability risk level at the corresponding geological horizon according to the risk equivalent surface transparency parameter, and apply a pulse warning special effect to the spatial anchor point marker, with the special effect frequency and the characteristic frequency f c Maintain a 1:3 harmonic relationship.

[0098] When the displacement acceleration is detected a curr Exceed the current dynamic grading threshold : If , generate a level three warning instruction, trigger the highway information board to issue a speed limit of 40 km / h and a slope monitoring reminder.

[0099] If , generate a level two warning instruction, start the drone inspection and close the outer lane.

[0100] If , generate a level one warning instruction, implement full-section closure and activate the emergency support system.

[0101] Finally, encode the warning instruction into a traffic control protocol data packet that meets the standard, and establish a linkage with the gate control unit, information board display unit, and maintenance dispatching platform of the highway operation management system through the 5G-V2X communication module.

[0102] Specifically, taking the Yu-Kun Expressway as an example, the multi-level warning linkage effect shown in Table 4 is obtained: Table 4 Multi-level warning linkage data table

[0103] In addition, the embodiment of the present invention provides a dynamic beam radar monitoring and linkage warning system for highway layered slopes. The system is used to execute the above method, including: 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, synchronously acquiring phase-coherent echo signals from the slope surface and internal structure, and uploading them to the edge computing unit. The radar hardware unit includes a continuous wave radar array, comprising a MIMO architecture consisting of multiple transmitting units and multiple receiving units; a beam control module connected to the continuous wave radar array via a high-speed bus, and configured to control the radar array's transmitting units to transmit frequency-modulated signals according to the beam incidence angle adjustment sequence and scanning parameter constraints for each scanning cycle of the continuous wave radar array as specified in the layered scanning strategy, and synchronously adjust the beam synthesis direction of the receiving units to match the spatial coverage requirements of each scanning layer.

[0104] The edge computing unit is connected to the radar hardware unit and includes a signal processing module, a deformation solution module and an early warning module. The signal processing module is used to extract multi-threshold scattering points from the phase-coherent echo signal, combine the analysis and processing of the prior constraint model constructed by the acquired slope geometric features and the geological vegetation recognition network optimized by transfer learning, and eliminate the interference signals caused by vehicle passage to obtain a set of spatiotemporal coherent scattering points; the deformation solution module is used to input the set of spatiotemporal coherent scattering points into the preset deformation solution 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 time series evolution characteristics, and perform spatiotemporal correlation analysis on the displacement data stream with the acquired layered inclination time series data and meteorological and hydrological data. The early warning module is used to update the slope stability assessment index in real time through a dynamic baseline, and link with the augmented reality display terminal to automatically generate multi-level early warning instructions.

[0105] The system also includes a multi-source interface unit (MIU) for accessing data from radar-visual fusion sensors and meteorological and hydrological equipment, and linking it with the highway operations management system via the 5G-V2X communication protocol. The MIU provides a radar-visual fusion data channel, supporting calibration data from laser displacement meters and fiber Bragg 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 edge computing unit exchange data via a data bus, while the edge computing subunit communicates with the early warning platform unit, ensuring end-to-end transmission latency of ≤10ms.

[0106] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, system, or computer program product. Therefore, the present invention can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0107] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications to these embodiments after learning the basic creative concepts. Therefore, the present invention should be construed to include the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0108] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalent technologies, the present invention should also include these modifications and variations.

Claims

1. A dynamic beam radar monitoring and linkage warning method for highway layered slopes, 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 the highway slope, synchronously acquiring phase-coherent echo signals from the slope surface and internal structure. The phase-coherent echo signal is subjected to multi-threshold scattering point extraction. This is processed using a priori 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 eliminated to obtain a set of spatiotemporal coherent scattering points. The spatiotemporal coherent scattering point set is input into the preset deformation solution pipeline, which simultaneously performs differential interferometry, adaptive atmospheric disturbance correction, and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and time-series evolution characteristics. The displacement data stream is analyzed in temporal and spatial correlation with the acquired layered inclination time series data and meteorological and hydrological data. The slope stability assessment index is updated in real time through the dynamic baseline, and is linked with the augmented reality display terminal to automatically generate multi-level early warning instructions.

2. The dynamic beam radar monitoring and linkage warning method for highway layered slopes according to claim 1, characterized in that Using a continuous wave radar array with adjustable beam angle, the layered geological structure of the highway slope is scanned layer by layer, and phase-coherent echo signals of the slope surface and internal structure are simultaneously acquired, including: 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 rock and soil layer is dynamically generated and a beamforming network is constructed. Combined with the preset scanning parameter constraints, the beam incidence angle adjustment sequence for each scanning cycle of the continuous wave radar array is determined. Based on the scanning parameter constraints, the radar array's transmit / receive unit spatial arrangement parameters and operating frequency band combination are initialized to construct the hardware configuration template corresponding to each scanning layer. Control the continuous wave radar array to switch the transmit beam angle according to the beam incidence angle adjustment sequence. When switching to each scanning layer, the hardware configuration template corresponding to the current layer is loaded. Based on the dielectric constant of the current scanning layer and the target detection depth, the amplitude and phase distribution weights of the beamforming network are adjusted in real time to ensure that the spatial resolution of the radar array and the electromagnetic wave penetration depth change adaptively with the scanning layer. During the layered scanning process, the parameter combination including the transmission power, pulse repetition frequency and coherent integration time is dynamically adjusted based on the signal-to-noise ratio spectrum characteristics of the echo signal, so that the signal phase stability of each scanning layer is maintained within the preset stability range; The received multi-level echo signals are orthogonally separated, and the phase-consistent radar image of the slope surface and internal structure is reconstructed through the digital beamforming matrix, and a phase-coherent echo signal set that meets the preset differential interferometry measurement requirements is output.

3. The dynamic beam radar monitoring and linkage warning method for layered slopes facing expressways according to claim 2, characterized in that 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 rock and soil layer is dynamically generated and a beamforming network is constructed. Combined with the preset scanning parameter constraints, the beam incidence angle adjustment sequence for each scanning cycle of the continuous wave radar array is determined, including: Based on the imported geological survey data of highway slopes, the dielectric constant distribution and layer thickness parameters of each rock and soil layer are extracted; According to the dielectric constant distribution and layer thickness parameters of each rock and soil layer, a hierarchical scanning strategy function is constructed with the joint optimization objective of minimizing the inter-layer phase error and the total scanning time. The inter-layer phase error is quantified by the sum of squares of the phase delays of the electromagnetic wave propagation paths in each layer and the deviation from the corresponding threshold, and the total scanning time is calculated by accumulating the product of the beam switching frequency and the dwell time; Configure the phase stability error threshold, the maximum interval of adjacent beam incident angles, and the upper limit of the single-cycle scanning time as hard constraint conditions in the hierarchical scanning strategy optimization function, and solve through the constrained optimization algorithm to obtain the beam incident angle adjustment sequence that satisfies all constraints; Synchronously construct a beamforming network that matches the layered structure, and dynamically initialize the amplitude-phase distribution weights of the beamforming network according to the dielectric constant distribution of each rock and soil layer.

4. The dynamic beam radar monitoring and linkage warning method for highway layered slopes according to claim 2, characterized in that Perform orthogonal separation on the received multi-level echo signals, reconstruct the phase-consistent radar image of the slope surface and internal structure through the digital beamforming matrix, and output the phase-coherent echo signal set that meets the preset differential interferometry requirements, including: According to the dip angle and layer thickness parameters of the slope layered geological structure extracted from the imported geological exploration data, assign an orthogonal time-domain coding sequence and a spatial-domain beam pointing code set to each scanning layer; Construct a composite control matrix for spatio-temporal joint coding based on the time-domain orthogonal coding sequence and the spatial-domain beam pointing code set; Perform cross-correlation decoupling operations on the multi-level echo signals received by the radar array through the composite control matrix, separate the spatio-temporal coupling components of the echo signals of different layers, and generate the baseband signal channels of each scanning layer; Perform spatial-domain selective enhancement processing on each baseband signal channel using the spatial-domain beam pointing code set and suppress adjacent layer interference; Reconstruct the three-dimensional complex scattering characteristic distribution data of each scanning layer based on the enhanced signals, and eliminate the phase distortion caused by the multi-path effect through phase gradient consistency analysis; Perform full-aperture phase plane fitting and coherent accumulation on the data with eliminated distortion to generate a spatially continuous phase-consistent radar image; Extract the scatter point set that meets the preset spatial continuity criterion from the radar image sequence, and output the phase-coherent echo signal set.

5. The dynamic beam radar monitoring and linkage warning method for highway layered slopes as described in claim 1, wherein Perform multi-threshold scatter point extraction on the phase-coherent echo signals, analyze and process them through the prior constraint model constructed from the obtained slope geometric features and the geological vegetation recognition network optimized by transfer learning, and exclude the interference signals caused by vehicle passage, to obtain the spatio-temporal coherent scatter point set, including: According to the slope angle, layer thickness parameters, and structural plane spatial orientation parameters of the slope extracted from the imported geological exploration data, construct a three-dimensional slope prior model that includes scatter point spatial density constraints, displacement vector continuity constraints, and structural plane strike and radar viewing direction orientation constraints; Based on the joint criterion of the amplitude deviation index and the phase stability threshold, screen candidate scatter points from the phase-coherent echo signals, and use the spatial density constraint of the three-dimensional slope prior model to perform layer membership tests on the candidate scatter points to eliminate outliers that deviate from the layered geological features of their respective layers; Input the scatter points that have passed the layer membership test into the pre-constructed geological vegetation recognition network optimized by transfer learning, and output the geological attribute classification results and confidence scores of each scatter 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 transmission signal wavelength, vehicle three-dimensional attitude, and scattering point azimuth angle. Each scattering center of the model is jointly characterized by a complex reflection coefficient, a time-varying radial distance, and an acceleration frequency modulation term. The time-varying radial distance includes the initial distance, the radial velocity, and the quadratic motion component of the acceleration, and a rectangular window function is introduced to limit the effective illumination time window of each scattering point; Generate the time-frequency distribution characteristic spectrum of the dynamic interference signal through the moving target scattering model, extract the time-varying scattering intensity envelope curve and the frequency modulation rate change curve from the time-frequency distribution characteristic spectrum, and establish a vehicle interference feature library; Based on the vehicle interference feature library and the geological attribute classification results, perform cross-validation, and identify and filter out the transient interference signals caused by vehicle passage according to the preset vehicle interference judgment criteria to obtain a spatio-temporally coherent scattering point set; among them, the vehicle interference judgment criteria include: the scattering intensity envelope is positively correlated with the vehicle moving speed and the envelope width matches the lane occupancy time, the frequency modulation rate change curve conforms to the non-linear characteristics in the acceleration / braking mode, and the geological attribute classification result of the target area is a non-rock-soil structure body.

6. The dynamic beam radar monitoring and linkage warning method for highway layered slopes according to claim 1, characterized in that Input the spatio-temporally coherent scattering point set into the preset deformation solution pipeline, and synchronously perform differential interferometric measurement, adaptive atmospheric disturbance correction, and slope structure parameter inversion to generate a displacement data stream containing three-dimensional deformation vectors and temporal evolution characteristics, including: Receive the spatio-temporally coherent scattering point set through the deformation solution pipeline pre-deployed on the edge computing node, perform spatio-temporal reference unification processing, and establish standardized observation data registered with the slope geographic coordinate system; Based on the observation data, perform in parallel in the deformation solution pipeline: select scattering points of adjacent time phases for phase difference calculation to generate an interference phase map covering the entire slope area; construct an atmospheric delay correction surface based on the meteorological sensor data deployed on the slope to compensate for the spatial variation error of the interference phase map to obtain a residual phase field; call the elastic modulus and Poisson's ratio of the preset geotechnical mechanics database, and calculate the deformation gradient field of each rock and soil layer using the displacement-strain conversion relationship; Perform multi-source data fusion on the interference phase map, the residual phase field, and the deformation gradient field, and use the confidence-weighted adaptive least squares algorithm to solve the three-dimensional deformation vector; Analyze the statistical characteristics of the deformation vector through a time series sliding window, extract the time series evolution characteristics including the surface displacement rate, acceleration, and deep slip trend parameters of the slope, and organize them in time series to generate a sub-millimeter-level displacement data stream.

7. The dynamic beam radar monitoring and linkage warning method for layered slopes facing expressways according to any one of claims 1-6, characterized in that Perform spatio-temporal correlation analysis on the displacement data stream, the obtained hierarchical dip angle time series data, and the meteorological and hydrological data, update the slope stability evaluation index in real time through a dynamic baseline, and link with the augmented reality display terminal to automatically generate multi-level warning instructions, including: Decompose the displacement data stream into frequency bands to extract the first deformation component, the second deformation component, and the third deformation component, and synchronously perform time domain alignment processing on the hierarchical dip angle time series data collected by the dip angle sensors deployed on the slope to generate a joint observation sequence of deformation and dip angle; The obtained rainfall data is converted into a surface runoff intensity distribution map of the slope through spatial interpolation, and the pore water pressure gradient field is inverted based on the obtained groundwater level monitoring data, and the change rate of the pore water pressure gradient within a preset time period is calculated; Calculate the time-delay coupling intensity between the first deformation component and the change rate of the pore water pressure gradient to obtain the delay time parameter of the displacement lagging behind the 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 activation characteristic frequency of the structural plane by detecting the frequency-domain coherence between the third deformation component and the time-series data of the layered dip angle; Based on the joint observation sequence, establish an evaluation baseline including the displacement acceleration threshold, the inclination change rate threshold, the lag time threshold, and the activation dangerous frequency band of the structural plane; Calculate the sensitive point density of each slope partition based on the spatial coordinate set of the deformation-sensitive area, and dynamically adjust the evaluation baseline according to the sensitive point density. Among them, the displacement acceleration threshold floats proportionally with the increase of the sensitive point density, the inclination change rate threshold tightens linearly with the increase of the sensitive point density, the lag time threshold is corrected based on the sensitive point density, and the activation dangerous frequency band of the structural plane expands adaptively with the increase of the density; When multiple conditions are met, including the displacement acceleration being greater than the adjusted displacement acceleration threshold, the inclination change rate being greater than the adjusted inclination change rate threshold, the delay time parameter being less than the adjusted lag time threshold, and the activation characteristic frequency of the structural plane being within the adjusted activation dangerous frequency band of the structural plane, trigger the determination of the deterioration of slope stability; Based on the displacement over-limit amount, the inclination change trend slope, and the activation characteristic frequency of the structural plane obtained in real time when the determination is triggered, combined with the evaluation baseline, calculate the comprehensive instability index through non-linear weighting, and generate a slope stability evaluation index set: Input the slope stability evaluation index set into the augmented reality display terminal for visualizing the virtual-real superposition of the slope deformation field and the three-dimensional geological model, and automatically generate multi-level warning instructions linked to the highway operation management system.

8. The dynamic beam radar monitoring and linkage warning method for highway layered slopes according to claim 7, characterized in that Input the slope stability evaluation index set into the augmented reality display terminal for visualizing the virtual-real superposition of the slope deformation field and the three-dimensional geological model, and automatically generate multi-level warning instructions linked to the highway operation management system, including: Synchronize the real-time slope stability evaluation index with the augmented reality display terminal to achieve the spatial registration and superposition display of the slope deformation field and the three-dimensional geological model; Generate multi-level warning signals including at least one of speed limit control, lane closure, and emergency support instructions according to the comprehensive instability index level; Encode the multi-level warning signals into traffic control protocol data packets, and establish a linkage with the gate control unit, the information board display unit, and the maintenance dispatching platform of the highway operation management system through the 5G-V2X communication module.

9. A dynamic beam radar monitoring and linkage warning system for highway layered slopes, characterized in that, The system is used to execute the method described in any one of claims 1-8, including: A radar hardware unit for implementing layered scanning of the layered geological structure of the highway slope according to the layered scanning strategy issued by the edge computing unit, synchronously obtaining the phase coherence echo signals of the slope surface and internal structure, and uploading them to the edge computing unit; The edge computing unit is connected to the radar hardware unit and includes a signal processing module, a deformation solution module and an early warning module. The signal processing module is used to extract multi-threshold scattering points from the phase-coherent echo signal, combine the analysis and processing of the prior constraint model constructed by the acquired slope geometric features and the geological vegetation recognition network optimized by transfer learning, and eliminate the interference signals caused by vehicle passage to obtain a set of spatiotemporal coherent scattering points; the deformation solution module is used to input the set of spatiotemporal coherent scattering points into the preset deformation solution 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 time series evolution characteristics, and perform spatiotemporal correlation analysis on the displacement data stream with the acquired layered inclination time series data and meteorological and hydrological data. The early warning module is used to update the slope stability assessment index in real time through a dynamic baseline, and link with the augmented reality display terminal to automatically generate multi-level early warning instructions.

10. The dynamic beam radar monitoring and linkage warning system for layered slopes facing highways according to claim 9, characterized in that, Radar hardware unit, including: 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 through a high-speed bus. It is used to control the radar array's transmitting units 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 synchronously adjust the beam synthesis direction of the receiving units to match the spatial coverage requirements of each scanning layer.

Images