A multi-wave multi-time ground penetrating radar system based on cooperative modulation

Abstract

The application discloses a kind of based on the multi-wave multi-time ground penetrating radar system of cooperative modulation, belong to geophysical prospecting field.The system includes cooperative control and signal processing center, multi-waveform cooperative transmitting module, reconfigurable antenna array module and multi-channel cooperative receiving module.Through cooperative control center scheduling, transmit multiple orthogonal or quasi-orthogonal modulation waveforms, use reconfigurable antenna array for multiple spatial sampling, finally the echo data of multi-wave multi-time is cooperatively fused and processed and jointly inverts imaging.The application effectively cooperates the detection depth and resolution, significantly improves the anti-interference ability of system, information acquisition dimension and target identification accuracy, and is suitable for high requirement detection scene in the fields such as geological exploration, civil engineering detection, archaeology etc..

Description

Technical Field

[0001] This invention belongs to the field of geophysical exploration and electronic information technology, specifically relating to a ground-penetrating radar system. More specifically, this invention relates to a multi-wave, multi-pass ground-penetrating radar system that transmits multiple cooperatively modulated waveforms, performs multiple spatial samplings using a reconfigurable antenna array, and processes the data based on cooperative scheduling and joint inversion algorithms. This system aims to solve the challenge of high-precision, high-resolution, and multi-parameter imaging of underground targets in complex environments, and can be widely applied in fields such as geological exploration, urban underground space exploration, non-destructive testing of civil engineering, archaeology, and military security. Background Technology

[0002] Ground-penetrating radar (GPR) technology has made significant progress in hardware integration and data processing over the past few decades. However, facing increasingly complex underground detection needs, its core technological framework still suffers from a series of systemic bottlenecks and theoretical limitations. Traditional GPR systems, whether early time-domain pulse systems or later frequency-modulated continuous wave systems, are generally based on a technical paradigm of single-waveform transmission, fixed array configuration, and independent processing of modules. This paradigm leads to three fundamental contradictions: First, the use of low-frequency signals to achieve greater penetration depths conflicts with the high-frequency bandwidth required for high-resolution capabilities, which cannot be reconciled by signal processing methods alone. Second, the complete electromagnetic characterization of underground media relies on two independent parameters: dielectric constant and conductivity. However, traditional systems can only acquire a single data stream (such as reflection amplitude or two-way travel time) in a single observation, mathematically constituting an underdetermined inversion problem, leading to non-uniqueness of solutions and high uncertainty in interpretation results. Finally, the system's operating mode and parameter configuration are usually preset and static, making it difficult to adapt to the complex and variable underground detection environment with scarce prior information, resulting in significant adaptability defects.

[0003] To overcome these limitations, the industry has explored various improvement approaches. Multi-band synthesis technology expands the equivalent bandwidth by time-division switching of subsystems with different center frequencies, but it is essentially a simple superposition of multiple independent narrowband systems. Maintaining strict synchronization and consistency in the acquisition timing and spatial sampling of data across different frequency bands poses a significant challenge to subsequent data fusion. Multiple-transmitter-multiple-receiver array technology, especially multiple-input-multiple-output architecture, improves data acquisition efficiency and lateral resolution by increasing spatial sampling points. However, its transmitted waveforms often use the same linear frequency modulated signal or pseudo-random sequences with only basic orthogonality. The freedom in waveform design is not fully explored, failing to achieve deep coupling and synergistic effects with the array's spatial beamforming capabilities. While polarimetric detection technology can provide additional vector information dimensions, it is still largely tied to a single carrier frequency, offering limited improvement in the ability to identify complex scatterers. More critically, most of these improvement schemes remain at the level of "hardware superposition" or "local optimization," lacking a top-level collaborative design framework. This results in the overall system performance often falling below the sum of the theoretical potential of each component, exhibiting a diminishing returns phenomenon of "1+1<2."

[0004] These technological limitations manifest as a series of prominent shortcomings in practical engineering applications: In urban environments with strong electromagnetic interference or under high-attenuation geological conditions, the system's signal-to-noise ratio deteriorates sharply; for shallow-buried, small-sized targets (such as underground pipeline joints or early road voids), effective signals are easily annihilated by strong direct wave coupling and surface clutter; when facing complex, non-homogeneous, and anisotropic media, traditional imaging algorithms based on the homogeneous medium assumption produce significant distortions, leading to false anomalies or missed detections of real targets. Furthermore, the intelligence level of existing systems remains limited; the entire process from data acquisition and processing to interpretation still heavily relies on human experience, making it difficult to guarantee the reliability, repeatability, and standardization of detection results. This severely restricts the large-scale, routine application of this technology in high-end scenarios such as infrastructure inspection.

[0005] In summary, the field of ground-penetrating radar (GPR) is at a critical technological turning point. Simply improving single-point performance indicators (such as higher sampling rates, greater transmit power, and more channels) is no longer sufficient to meet the fundamental requirements of next-generation detection missions for deep-penetrating, high-precision integrated detection, multi-parameter quantitative inversion, and adaptive intelligent processing. Therefore, a revolutionary system architecture is urgently needed that, based on the fundamental principles of information theory, collaboratively optimizes waveform design, spatial sampling, and information processing as an organic whole, thereby breaking through the theoretical limitations of traditional paradigms and achieving a leapfrog improvement in detection performance. This invention is proposed precisely to address this pressing technological challenge. Summary of the Invention

[0006] This invention provides a multi-wavelength ground-penetrating radar system and method based on coordinated modulation and intelligent sensing. This system fundamentally solves the inherent contradictions between detection depth, resolution, multi-parameter inversion accuracy and environmental adaptability of traditional ground-penetrating radar by constructing an intelligent detection system that coordinates waveform, spatial and temporal depth.

[0007] The core architecture of this system consists of four major functional modules forming an organic whole. The intelligent collaborative control and processing center, serving as the system's decision-making hub, integrates two core processors: a collaborative scheduling engine and a joint constraint inversion engine. This center employs a Bayesian inference framework to achieve closed-loop optimal control. Its decision criterion is based on the principle of maximizing mutual information in information theory, guiding the entire detection process by minimizing the posterior uncertainty of underground medium parameters. The multi-waveform collaborative transmission module adopts an advanced software-defined radio architecture, capable of dynamically generating a set of transmission waveforms with multi-dimensional orthogonal characteristics in the frequency, code, time, and polarization domains. These waveforms undergo rigorous mathematical optimization design to ensure that their sensitivity functions to dielectric constant and conductivity remain linearly independent in the parameter space, providing sufficient excitation conditions for subsequent multi-parameter decoupling inversion.

[0008] The system's hardware implementation also reflects innovative design. The reconfigurable antenna array module consists of broadband Vivaldi antenna elements, each with a high-speed RF switch network and a high-precision digital phase-shift attenuation network integrated at the back end. This design enables the array to flexibly switch operating modes within microseconds, including seamless transitions from full-aperture wide-beam scanning mode to sub-aperture high-gain focusing mode, as well as dynamic adjustment of polarization. The multi-channel synchronous receiver module employs a high dynamic range acquisition circuit design, coupled with a precise clock synchronization system, to accurately capture complete observation datasets containing waveform, spatial, and temporal three-dimensional indices, providing data assurance for high-precision inversion at the back end.

[0009] In terms of algorithmic innovation, this invention proposes two core mathematical frameworks. The collaborative scheduling algorithm models the detection process as a sequential Bayesian experimental design problem, achieving optimal dynamic allocation of detection resources by solving a waveform-space-temporal joint optimization problem under hardware constraints. This algorithm uses the sequential Monte Carlo method to update the posterior probability distribution of parameters in real time and dynamically adjusts the launch strategy based on stochastic gradient approximation, enabling the system to intelligently transition from a wide-area census mode to a local fine-scale detection mode. The joint constraint inversion method addresses the inherent ill-conditioned nature of geophysical inversion by proposing a regularization framework based on structured priors. This framework innovatively introduces a joint sparse constraint term on top of traditional least-squares data fitting terms and total variational smoothing constraints. It forces the spatial support set of the dielectric constant and conductivity gradient to remain consistent through the mathematical ℓ2,1 mixed norm, effectively utilizing prior knowledge that abrupt changes in subsurface medium properties often exhibit spatial coupling.

[0010] The system employs efficient numerical optimization methods for inversion calculations. The preprocessing alternating direction multiplier method is used to solve large-scale regularized inversion problems, decomposing the complex optimization problem into multiple subproblems that can be solved in parallel. Combined with GPU-accelerated computing technology, the system can achieve fast and accurate inversion calculations for large-scale 3D exploration areas, meeting timeliness requirements in practical engineering applications. The inversion process also integrates an uncertainty quantification module, which can output confidence intervals for parameter estimates, providing a scientific basis for the reliability assessment of the exploration results.

[0011] The technical advantages of this invention are reflected in multiple dimensions. At the physical detection level, the collaborative design of low-frequency penetration waveforms and high-frequency resolution waveforms, combined with multi-band data fusion processing, effectively resolves the contradiction between depth and resolution in traditional systems. At the information acquisition level, by leveraging multi-dimensional orthogonal excitation waveforms and joint constraint inversion algorithms, true multi-parameter quantitative separation imaging of dielectric constant and conductivity is achieved for the first time in a practical system. Regarding anti-interference capabilities, waveform diversity technology and spatiotemporal-frequency three-dimensional joint processing significantly improve the system's target detection performance under strong electromagnetic interference environments. In terms of system intelligence, a fully adaptive autonomous optimization of the detection process is achieved based on a closed-loop optimization architecture.

[0012] In summary, this invention not only provides an innovative hardware system platform, but more importantly, it establishes an intelligent detection methodology that deeply integrates the physical mechanisms of electromagnetic wave propagation, information theory optimization criteria, and artificial intelligence algorithms. This system represents a significant direction in the evolution of ground-penetrating radar technology from traditional instruments to intelligent detection systems, providing a new generation of technical solutions for fields such as precise underground space exploration, infrastructure safety monitoring, and mineral resource exploration. Attached Figure Description

[0013] Figure 1 System overall structure diagram.

[0014] The demonstration features a collaborative control center (including DSP / FPGA) as the core, which connects the data and control flows of multi-waveform transmitting modules (AWGs), reconfigurable antenna arrays (via a switch / phase shifter network), and multi-channel receiving modules (ADCs).

[0015] Figure 2 : Schematic diagram of coordinated modulation waveform.

[0016] son Figure 2 a (Time-Frequency Diagram): Shows the orthogonality of two LFM waveforms in the frequency division, and the distribution of a phase-coded waveform in the time-frequency domain.

[0017] son Figure 2b (Autocorrelation and Cross-correlation Plot): Displays the sharp autocorrelation peaks and low-profile cross-correlation sidelobes of the selected coded waveform, visually illustrating its separability.

[0018] Figure 3 : Schematic diagram of the working mode of a reconfigurable antenna array.

[0019] son Figure 3 a (Wide Beam Scanning Mode): All array elements are excited in phase to form a wide beam.

[0020] son Figure 3 b (Focused Beam Mode): The middle array elements are activated and a focusing phase weight is applied to form a narrow beam pointing at a specific angle.

[0021] Figure 4 Flowchart of Cooperative Signal Processing and Imaging Algorithms

[0022] The flowchart clearly illustrates the complete process from "raw multi-channel data input" to "waveform separation", "target waveform data input", "construction of objective function", "iterative solution" and finally "multi-parameter imaging output".

[0023] Figure 5 Schematic diagram comparing detection performance with traditional systems

[0024] son Figure 5 a (Traditional single-wave radar image): Displays a blurry image, contains false targets, and has unclear layers.

[0025] son Figure 5 b (Fused Data Processing Image): Displays a clear, layered interface, accurately identifies the location and outline of anomalies (such as pipes and cavities), and can display the distribution of ε and σ in different color overlays. Detailed Implementation

[0026] This invention provides a multi-wavelength, multi-stage ground-penetrating radar system based on coordinated modulation. Its core lies in achieving efficient acquisition and fusion of multi-dimensional information through waveform design orthogonality, array reconfigurability, and intelligent coordination of scheduling and algorithms. The following sections describe these modules in detail:

[0027] 1. Detailed design of the multi-waveform collaborative transmission module:

[0028] This module employs a software-defined radio architecture, with a multi-channel high-speed arbitrary waveform generator at its core. The generated cooperative modulation waveform sets must satisfy waveform diversity and receiver separability.

[0029] Waveform orthogonality guarantee scheme:

[0030] Frequency division orthogonality: Generates a set of sub-band waveforms that do not overlap or partially overlap in the frequency spectrum but have significantly different center frequencies. For example, waveform A has a center frequency f.c1 = 100MHz, bandwidth B1 = 80MHz linear frequency modulated wave; waveform B is the center frequency f c2 A linear frequency modulated wave with a frequency response (FQR) of 500 MHz and a bandwidth (B2) of 200 MHz can be initially separated at the receiving end using a bandpass filter.

[0031] Code division orthogonality (CDO): ​​Phase or frequency modulation is performed using coded sequences with good cross-correlation characteristics within the same or similar frequency bands. For example, waveforms C and D both use a carrier with a center frequency of 800MHz, but are modulated using complementary Glaucon codes and Barker codes, respectively, for binary phase modulation. Let the two coded sequences be P. (t) and Q (t) Its ideal autocorrelation function R pp(τ) and R qq(τ) It approaches zero at non-zero delays, and the cross-correlation function R pq(τ) ≈ 0. During reception, interference from the other signal can be effectively suppressed by using their respective matched filters (correlators).

[0032] Time-division orthogonal / quasi-orthogonal: Different waveforms are transmitted in a time-division manner during fast time (within a single detection cycle); or waveform combinations are transmitted alternately during slow time (between multiple detection cycles), separated by time gating.

[0033] Spatial orthogonality: Combining reconfigurable antenna arrays, different waveforms are guided to different spatial angles.

[0034] In practical design, a combination of the above-mentioned orthogonal methods is usually used, such as "frequency division + code division", to maximize the distinguishability between waveforms and ensure effective separation even in the context of strong clutter and noise.

[0035] 2. Specific implementation of the reconfigurable antenna array module:

[0036] This module forms the physical basis for realizing "multi-wave" space transmission and "multiple" space sampling. Its preferred implementation scheme is as follows:

[0037] Structure: The Vivaldi antenna employs a closely spaced Vivaldi horn antenna element to form a uniform linear array or area array. Vivaldi antennas possess ultra-wideband characteristics (e.g., 0.5-3 GHz), capable of covering the frequency bands required for multiple waveforms.

[0038] Reconfigurable Circuitry: Transmit / Receive Switching and Channel Selection: Each antenna element is connected to a single-pole double-throw RF switch at its rear end for switching between transmit and receive states. All switches are controlled uniformly by the FPGA via digital I / O ports.

[0039] Beamforming Network: At the transmitter, an analog phase shifter and attenuator network (or a digital beamforming network, DBF) is used. The FPGA, according to instructions from the collaborative control center, applies specific phase weights φ to the waveform signal of each transmit channel. n And magnitude weight A n This causes the array to form a transmitted beam in a specific direction θ0. The beam pointing is determined by the formula φ. n = -(2π / λ) * d * n * sin(θ0) is determined, where d is the element spacing, λ is the wavelength, and n is the element number.

[0040] Polarization Reconfiguration: Each antenna element can be composed of a pair of orthogonally placed dipoles. By controlling the amplitude and phase difference of the signals fed to the two dipoles, linear, elliptical, or circularly polarized waves can be dynamically generated.

[0041] Example of operating mode: In "scanning mode", the array emits low-frequency probe waves with a wide beam; after the target is detected, it switches to "focusing mode", only activates some array elements (sub-apertures) and adjusts the weights to accurately focus the high-frequency discrimination beam onto the target area.

[0042] 3. Mathematical principles of cooperative scheduling algorithms and joint constraint inversion algorithms:

[0043] Cooperative scheduling algorithm flow:

[0044] Initialization: Input prior information (such as approximate attenuation coefficient of the medium, target size range).

[0045] Waveform-space resource joint optimization model: The objective function is defined as maximizing information entropy or minimizing the Cramer-Rao lower bound, and the constraints are system resources (total transmit power, single cycle duration, number of array elements).

[0046] Decision variables: waveform type combination W, transmission timing T, array weight matrix A.

[0047] The model can be simplified to: maximize I(ε, σ; Y | W, T, A), where I represents the mutual information of the observed data Y with respect to the dielectric parameters (dielectric constant ε, conductivity σ).

[0048] Adaptive Iteration: A Bayesian experimental design is employed. Based on the inversion results (posterior probability distribution) of the previous round of exploration, the understanding of the underground structure is updated, and W, T, and A are dynamically adjusted for the next round or the next time step to minimize parameter uncertainty in the optimal way.

[0049] Joint Constraint Inversion Imaging Algorithm Flow:

[0050] Multi-channel data preprocessing and waveform separation: Echo y of the k-th receiving channelk(t) Using matched filter bank h i(t) (and the i-th transmitted waveform s) i(t) Separate using conjugate matching: r{k,i}(t) = y k(t) ⊗ h i(t) The sub-echo data matrix R corresponding to waveform i is obtained. i .

[0051] Constructing multidimensional data vectors: All R i Arranged by spatial location, waveform index, and time sampling point, a high-dimensional column vector d is constructed.

[0052] Establish a joint forward model: Based on Maxwell's equations, a discretized unified forward modeling operator G is established. This operator relates the subsurface medium parameter model m (containing the ε and σ values ​​of each discrete grid) to all observed data d: d = G(m) + n, where n is noise.

[0053] 4. Construct a joint objective function with multiple physical parameters:

[0054] Φ(m) = ||d - G(m)|| + λ1 TV(ε) + λ2 TV(σ) + λ3 ||Ψ(ε, σ)||1

[0055] The first item is the data fitting term.

[0056] The second and third terms are total variational constraints on ε and σ, which promote piecewise constant solutions and maintain a clear interface.

[0057] The fourth term is the coupling constraint term. Ψ is a joint sparse transformation (such as a joint representation of the gradient domain). By utilizing the physical correlation between ε and σ (such as the interface usually changing simultaneously), it forces the change positions to be consistent, thereby improving the stability and resolution of the inversion.

[0058] 5. Solution and Imaging: The above optimization problem is solved using the alternating direction multiplier method or the split Bregman iterative algorithm, and finally high-resolution ε-distribution images or σ-distribution images are output synchronously to indicate the underground structure and material.

[0059] Matters not covered in this invention are common knowledge.

[0060] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A multi-wave, multiple-stage ground-penetrating radar system based on coordinated modulation, characterized in that, The system includes a collaborative control center, a multi-waveform transmitting module, a multi-waveform receiving module, a power supply, a reconfigurable antenna array module, and a host computer. The collaborative control center is configured to generate system control commands and process and transmit echo data. The multi-waveform transmitting module is connected to the collaborative control center and is configured to generate multiple different transmitting waveforms according to the control commands. The reconfigurable antenna array module is connected to the multi-waveform transmitting module and the multi-waveform receiving module, and is configured to radiate the transmitted waveform and receive the echo signal. Its signal channel can be reconfigured by a switch or a phase-shifting network. The multi-waveform receiving module is configured to receive and preprocess the echo signal, and send the processed data to the collaborative control center or host computer.

2. The system according to claim 1, characterized in that, The multi-waveform collaborative transmission module generates at least two transmission waveforms that have orthogonality or low cross-correlation in the time domain, frequency domain, code domain, or polarization domain to ensure that they can be effectively separated at the receiving end.

3. The system according to claim 2, characterized in that, The transmitted waveform includes, but is not limited to, two or more combinations of linear frequency modulated signals, phase-coded signals, and orthogonal frequency division multiplexing signals; wherein the phase-coded signals are modulated using complementary Glau codes, Barker codes, or m-sequences.

4. The system according to claim 1, characterized in that, The reconfigurable antenna array module includes: Multiple broadband antenna elements are arranged in a uniform linear array or area array; an RF switch network connected to each antenna element is used to control the transmission or reception state of each antenna element; a beamforming network is connected between the RF switch network and the multi-waveform cooperative transmission module and the multi-channel cooperative reception module, the beamforming network including a programmable phase shifter and / or attenuator, used to control the direction, width and polarization of the transmission beam and the reception beam.

5. The system according to claim 4, characterized in that, The broadband antenna element is a Vivaldi antenna, a horn antenna, or a patch antenna; the beamforming network dynamically adjusts the phase weight φ applied to the signals of each channel according to the instructions of the coordinated control and signal processing center. n With amplitude weight A n The phase weight is given by formula φ. n = -(2π / λ) · d· n · sin(θ0) is determined, where λ is the wavelength, d is the element spacing, n is the element number, and θ0 is the target beam pointing angle.

6. The system according to claim 1, characterized in that, The collaborative scheduling algorithm built into the collaborative control and signal processing center is used to perform the following steps: S1.1: Initialize the detection task based on prior information or the results of the previous detection round; S1.2: With the optimization objective of maximizing the information content of the observation data about the underground medium parameters or minimizing the Cramer-Rao lower bound of the parameter estimates, and with the system transmission power, time resources and number of array elements as constraints, the waveform type combination W, the transmission timing T and the array weight matrix A are jointly optimized. S1.3: Output the optimal scheduling strategy to the multi-waveform collaborative transmission module and the reconfigurable antenna array module.

7. The system according to claim 1 or 6, characterized in that, The joint constraint inversion imaging algorithm built into the collaborative control and signal processing center is used to perform the following processing on the received multi-channel echo data: S2.1: Use filters that match each transmitted waveform to separate the echoes and obtain the sub-echo data matrix R_i corresponding to the i-th waveform; S2.2: Arrange all sub-echo data matrices according to spatial, waveform and time dimensions to construct a multidimensional observation data vector d; S2.3: Establish a unified electromagnetic forward modeling operator G such that d = G(m) + n, where m is a parametric model containing the dielectric constant ε and conductivity σ of the medium to be determined, and n is noise; S2.4: Construct an objective function Φ(m) that includes data fitting terms and multi-physical parameter coupling constraints, and solve it using an iterative optimization algorithm to simultaneously invert the dielectric constant distribution ε(x,z) and conductivity distribution σ(x,z).

8. The system according to claim 7, characterized in that, The objective function Φ(m) has the following form: Φ(m) = ‖d - G(m)‖ + λ1 · TV(ε) + λ2 · TV(σ) + λ3 · ‖Ψ(ε, σ)‖1 Where TV(·) represents the total variation regularization term, used to maintain a clear interface; ||Ψ(ε, σ)||1 represents the joint sparse constraint term on the dielectric constant ε and conductivity σ, and λ1, λ2, λ3 are regularization parameters.

9. The system according to claim 1, characterized in that, The system supports at least two operating modes: Wide-area scanning mode: Uses low-frequency broadband waveforms and wide beam configuration for rapid surveying; Focused Fine Mode: High-resolution identification is achieved by using high-frequency or coded waveforms and narrow beam focusing configuration for the region of interest.

10. A ground-penetrating radar detection method based on the system described in any one of claims 1-9, characterized in that, Includes the following steps: S10.1: The Cooperative Control and Signal Processing Center formulates a cooperative scheduling strategy based on the detection scenario. The strategy specifies at least the waveform combination, transmission timing, and array configuration. S10.2: The multi-waveform coordinated transmission module generates corresponding multi-channel transmission signals according to the strategy, and radiates them underground through the reconfigurable antenna array module according to the predetermined beam characteristics; S10.3: The echo is received by the reconfigurable antenna array module and preprocessed by the multi-channel cooperative receiving module to obtain multi-channel echo data; S10.4: The collaborative control and signal processing center calls the joint constraint inversion imaging algorithm to perform waveform separation, data fusion and joint inversion on the multi-channel echo data, and finally outputs the dielectric constant and conductivity distribution image of the underground medium.

Images