Methods for exploring underground structures

Abstract

The present invention relates to a method and apparatus for probing an underground structure, the method comprising the steps of (a) transmitting a probe signal as an electromagnetic wave to the structure, (b) receiving echo signals of the electromagnetic wave from the structure, and (c) processing the echo signals to derive information about the structure. The step (c) of processing the echo signals comprises a step of sampling at a sampling frequency f s The probe signal includes sampling the echo signal at a plurality of discrete frequency components f n,k and multiple frequency components f n,k At least two of the Nyquist zones are in different Nyquist zones. The nth Nyquist zone is (n-1)*f s / 2 to n*f s / 2 frequency range, f s is the sampling frequency, n = 1, 2, .... Furthermore, k = 1...K, where K is the frequency component f n,k The method and apparatus facilitate increasing the data acquisition rate without losing information about the frequency content.

Description

[Technical Field]

[0001] The present invention relates to a method for exploring underground structures and an apparatus for exploring underground structures. [Background technology]

[0002] Electromagnetic waves are commonly used to explore artificial structures and underground structures. Ground-penetrating radar (GPR) is a technology often applied to geophysical exploration and non-destructive testing of concrete structures, and typically uses electromagnetic waves in the 10 MHz to 3.4 GHz range.

[0003] Step frequency continuous wave (SFCW) is a known method for acquiring GPR data. A schematic diagram of the frequency f vs. time t characteristics of the SFCW probe signal is shown in Figure 1a. Instead of transmitting a single broadband pulse or a fixed-frequency continuous probe signal, SFCW uses a continuous wave probe signal with a duration ts that has a frequency that changes over time or between subsequent probe signals, in particular in steps of Δf. The subsequent probe signals constitute a sweep of the probe signal, for example, in the range of 400 MHz to 3.4 GHz. Examples of SFCW methods and suitable apparatus are described in WO2018 / 161183A1.

[0004] Echoes of SFCW electromagnetic waves reflected by underground structures are received by an antenna and processed to obtain information about the underground structures. Such information includes, for example, the presence, location, and / or characteristics of discontinuities in the underground structures, such as reinforcing bars in concrete, defects in building structures, pipes, and geological layers in the soil. In particular, an A-scan, i.e., an amplitude-versus-time plot of echoes received in response to a broadband impulse, can be reconstructed from the echoes of the SFCW probe signal and then evaluated in terms of information about the underground structures.

[0005] The SFCW method is advantageous in terms of data quality, and consequently the quality of the resulting images of subsurface structures, particularly signal-to-noise ratio (SNR) and resolution, while maintaining a large penetration depth of electromagnetic waves, and as a result, it facilitates deep exploration.

[0006] On the other hand, the acquisition speed of the conventional SFCW method is limited; for example, GPR measurements with a resolution of 2 cm can only be achieved if the GPR device does not move faster than the maximum acquisition speed (e.g., 10-20 km / h) relative to the ground. Furthermore, the conventional SFCW method is subject to regulatory limitations because it sweeps a wide frequency range. [Overview of the project]

[0007] Therefore, the overall objective of the present invention is to provide a method and apparatus for exploring underground structures that facilitates the acquisition of high-resolution data, particularly data with a resolution of 50 mm or more, at a high measurement speed, especially 50 to 80 km / h or more. Preferably, such a method and apparatus for exploring underground structures complies with regulatory requirements, such as those of the Federal Communications Commission (FCC).

[0008] The features described below relating to this method also relate to the apparatus, and vice versa. Furthermore, the described features are intended to be disclosed independently of each other and, where reasonable, in combination.

[0009] Methods for exploring underground structures To carry out the above and further objectives of the present invention, which will become more readily apparent from the following description, a method for exploring underground structures, such as artificial structures or subsurfaces, includes the following steps:

[0010] (a) A step of transmitting the probe signal as an electromagnetic wave to the structure. The electromagnetic wave may have a polarization such as linear polarization. The step of transmitting the electromagnetic wave to the structure is performed in particular by an antenna.

[0011] (b) The step of receiving an electromagnetic wave echo signal from the structure. Preferably, a portion of the wave transmitted to the structure is reflected by the internal properties of the structure, which are characterized by changes in electrical properties, such as changes in dielectric constant. A portion of the reflected wave can be received from the structure as an echo signal, i.e., reflected electromagnetic wave, in particular by other means or an antenna.

[0012] (c) A step of processing the echo signal to derive information about the structure. The derived information may relate to the internal features of the structure. These internal features may include, for example, the location and / or gradient of internal boundaries within the structure, the characteristics of scatterers or reflectors, and electrical characteristics at specific locations within the structure, such as the change in dielectric constant. See also the background art section above.

[0013] Step (c) of processing the echo signal is performed at a sampling frequency f s This includes sampling the echo signal at a sampling frequency f. s The signal is then digitized. Sampling is preferably performed by an analog-to-digital converter (ADC). ADCs are typically limited in terms of the maximum possible sampling frequency and analog bandwidth. For example, ADCs with large analog bandwidths, such as 500 MHz or more, are generally complex and expensive, or perform poorly in terms of SNR and output signal resolution.

[0014] Generally, the sampling frequency f s When the signal is sampled, the Nyquist frequency f s Only frequency components up to / 2 can be reliably restored. Above the Nyquist frequency, aliasing occurs, and f s Frequency components with frequencies greater than or equal to / 2 are 0 and f sIt is folded down to the frequency range between / 2. In relation to the limitations of the ADC's capabilities (as described above), the Nyquist frequency limitation makes it difficult to use ultra-wideband (UWB) signals, i.e., signals having a bandwidth of 500 MHz or more in particular, even in GPR applications. To overcome this limitation, the present method has the following features.

[0015] The probe signal includes a plurality of discrete frequency components f n,k and, in particular, the probe signal consists of a plurality of discrete frequency components f n,k In particular, the frequency component f n,k is characterized by its frequency, amplitude, and phase. Alternatively, the amplitude and phase can also be represented as a complex amplitude. Preferably, all frequency components other than the plurality of discrete frequency components f n,k in the probe signal are zero, and in particular, all frequency components other than f n,k have an amplitude that is at most 50%, in particular at most 25%, or at most 10% of the maximum amplitude of f n,k .

[0016] Furthermore, at least two of the plurality of frequency components f n,k are located in different Nyquist zones. The nth Nyquist zone covers the frequency range from (n - 1)*f s / 2 to n*f s / 2, where f s is the sampling frequency and n = 1, 2,.... Strictly speaking, in particular, the nth Nyquist zone does not include its upper limit n*f s / 2. Furthermore, k = 1...K, where K is the number of frequency components in the nth Nyquist zone. K may be zero in some Nyquist zones, and in particular, K is zero in all Nyquist zones above the Nth Nyquist zone where f n,k having the maximum frequency is located.

[0017] In one embodiment, the plurality of discrete frequency components f n,kThis spans the bandwidth of the probe signal. As mentioned above, the bandwidth of the probe signal is wider than the Nyquist zone, and in particular, f s The bandwidth may be wider than / 2. Preferably, the probe signal bandwidth is at least 500 MHz, and in particular at least 1 GHz. The use of such a UWB signal facilitates compliance with regulatory restrictions such as FCC regulations and high-speed data acquisition, as shown below.

[0018] More preferably, the sampling frequency f s The sampling frequency range is 100-500MHz, particularly 200-300MHz, for example, 250MHz. In the latter case, the first Nyquist zone is 0-125MHz. As a result, for bandwidths of 500MHz or 1GHz, four or eight Nyquist zones are required (N=4 or N=8, respectively). Since these sampling frequency and analog input bandwidth values ​​are achievable with conventional ADCs, the implementation of this method is practical and reasonably priced.

[0019] As mentioned earlier, the frequency component f 1,k , that is, f s Only frequency components within the first Nyquist zone with frequencies less than / 2 can be reconstructed without aliasing. In general, the frequency components f of the nth Nyquist zone n,k This is the aliasing frequency component f' in the first Nyquist zone. n,k It is returned to f'. n,k The frequency is expressed by the following formula.

number

[0020] Preferably, f n,k For all n and k, f' n,k These are made to be different from each other. In other words, multiple discrete frequency components f present in the (analog) echo signal. n,k Each of them is (in the digital signal after sampling) another f' n,kUnlike the above, in particular, the aliasing frequency component f' of the distinguishable first Nyquist zone. n,k It is mapped to. In this way, the requirements of the ADC, in particular, f s While maintaining the requirements related to UWB, frequency components from multiple Nyquist zones, representing a wider bandwidth, can be used for exploring subsurface structures. This allows conventional, affordable ADCs to be used with UWB signals. Using UWB signals instead of single-frequency probe signals makes it easier to comply with regulatory restrictions such as FCC regulations.

[0021] remarks The frequency of the frequency component of the echo signal received from the structure is the frequency component f of the probe signal transmitted to the structure. n,k We assume that the frequency is substantially the same as that of . In particular, nonlinear effects such as frequency shifts can usually be ignored.

[0022] f' after sampling n,k A further advantage of the above condition, that it is unambiguous, is the high acquisition speed. Unlike the conventional SFCW method, which transmits one frequency component at a time, it transmits multiple frequency components simultaneously, so the acquisition speed increases by a coefficient equal to several M of the simultaneously transmitted frequency components. This makes it possible to acquire data at speeds of up to, for example, 50 km / h or 80 km / h with a spatial resolution of 50 mm, especially 25 mm or more. Thus, a GPR device implementing the above method can be easily mounted on a vehicle and can acquire GPR data more efficiently.

[0023] In one embodiment, for at least one Nyquist zone, in particular for all Nyquist zones where n=1...N and N is at least 2, K>1, in particular K>10, or K>20. Furthermore, the frequency component f n,kPreferably, the K frequency components are placed in Nyquist zones up to the Nth Nyquist zone, where N > 2. In particular, N is at least 4, or at least 6, or at least 8. If K frequency components are placed in all Nyquist zones up to the Nth Nyquist zone, the number of frequency components becomes N*K, and therefore, in particular, data acquisition becomes up to N*K times faster.

[0024] Furthermore, the probe signal has at least M discrete frequency components f n,k It may include, and M is at least 10, in particular at least 100, or at least 200. If the distribution of frequency components across the first N Nyquist zones is equal, then M is approximately N*K, in particular meaning deviations of up to approximately ±N.

[0025] f n,k Preferred conditions In one embodiment, the frequency component f n,k The elements are arranged at equal intervals in the frequency space, and in particular, adjacent f n,k They are arranged at equal intervals according to the frequency interval (Δf) between them. The frequency interval can be, for example, 0.1 to 100 MHz, and especially 1 to 10 MHz. Frequency component f n,k If the intervals are equal, in particular, simple data processing becomes possible by applying conventional SFCW processing algorithms. n,k If the intervals are equal, separate f' after sampling. n,k To obtain the sampling frequency f s f must not be a multiple of the frequency interval Δf. In general, an appropriate f n,k This is discovered empirically. Examples will be described below with reference to the figures.

[0026] Preferably, the frequency component f n,k For all n and k, the aliased frequency component f' n,k However, using the duration t0 of the probe signal, the minimum distance Δf' between them is set to be at least 1 / t0. n,k Such a condition regarding the minimum distance Δf' between them is, f' n,kThis makes it possible to reliably identify the discrete frequency component f. n,k In particular, its amplitude and phase can be uniquely determined.

[0027] As a typical example of an ADC, the sampling frequency f s =250MHz can be cited. When the duration of the probe signal t0 = 2μs, the aliasing frequency component f' n,k To separate them as distinct frequency components, a minimum distance Δf' of at least 0.5 MHz is required. To shorten the shortest distance Δf', the duration t0 of the probe signal (and therefore the duration of the echo signal) must be increased.

[0028] In one embodiment, the frequency component f n,k For all n and k, the aliased frequency component f' n,k However, they are made to have a minimum distance Δf' from each other that is at least 0.1 MHz, in particular at least 0.4 MHz, or at least 0.7 MHz. As illustrated above, such frequency selection allows for the use of sufficiently short probe signal durations, resulting in high acquisition speeds.

[0029] More preferably, the frequency component f n,k For all n and k, the aliased frequency component f' n,k However, at least f s / (4*M), in particular, if we apply the above assumption that M is approximately N*K, then at least f s The elements are given a minimum distance Δf' between them, which is given by / (4*N*K). Such a condition regarding the minimum distance Δf' is given by f' n,k This means that they are substantially equally spaced within the first Nyquist zone, and the value of the minimum distance Δf' is equal to all M aliasing frequency components f' n,k This is the first Nyquist zone, i.e., from 0 to f s f' when the frequency range up to / 2 is equally spaced n,k This corresponds to 50% of the theoretical distance between them.

[0030] Processing of echo signals In one embodiment, step (c) of processing the echo signal includes performing a spectral analysis of the echo signal, in particular a Fourier analysis. This is preferably performed in the digital domain, i.e., particularly after sampling the (analog) echo signal.

[0031] Preferably, step (c) processing the echo signal involves the aliasing frequency component f' in the echo signal. n,k This involves determining quantities that represent the amplitude and / or phase of each of the following: such quantities are, for example, the amplitude and / or phase of the echo signal directly, or each f'. n,k This can be the complex Fourier coefficients. In particular, the quantities that indicate the amplitude and / or phase of the frequency components are then further processed, for example by a conventional SFCW algorithm, to derive information about the underground structure. However, such quantities are first the (expanded) discrete frequency components f in the echo signal (before sampling). n,k It needs to be attributed to the corresponding frequency.

[0032] Therefore, preferably, step (c) of processing the echo signal is to process the aliased frequency component f' n,k The corresponding discrete frequency component f n,k This includes assigning it to the probe signal. n,k The frequencies are known, and they are different f's. n,k Because it is mapped above, such an attribution is f' n,k and f n,k This is possible given the above relationship between them.

[0033] Frequency component f having an even number n n,k Further steps may be required for this. Preferably, step (c) of processing the echo signal includes conjugating quantities that indicate the amplitude and / or phase of these frequency components. In particular, this is for frequency component f n,kThis involves conjugating the complex Fourier coefficients of by an even number n. Such a conjugate operation is necessary to reverse the conjugate of the quantity during convolution into the first Nyquist zone.

[0034] Non-zero initial phase shift Multiple frequency components f n,k Transmitting probe signals containing echoes directly can result in high peak transmit power. This is because peaks of periodic functions with zero initial phase shift, particularly equally spaced frequencies, are structurally added up over time, generating a high cumulative signal. This "initial phase shift" can be defined as the phase difference between the zero phase of a low-frequency function and the nearest zero phase of a high-frequency function. In equipment for exploring subsurface structures, the front end—that is, the antenna and analog components for processing received signals, the probe signal generator for generating probe signals, and the echo signal processor for processing echo signals—needs to be adapted to handle peak transmit power. Generating high peak transmit power, which is much greater than the average transmit power, is undesirable for signal-to-noise ratio and low intermodulation.

[0035] Therefore, in a preferred embodiment, the frequency component f n,k At least two of the spectral components have a non-zero initial phase shift. In particular, the initial phase shifts of at least two spectral components are such that the maximum amplitude of the probe signal is smaller than the maximum amplitude of a hypothetical probe signal having the (same) frequency components but with zero initial phase shifts for those frequency components, specifically by at least 10%, at least 25%, or at least 50%. This allows the peak transmit power to be kept low, in particular, not significantly higher than the average transmit power, for example, by about 2-3 times. Thus, a good signal-to-noise ratio and low intermodulation can be achieved.

[0036] In reality, a given number of frequency components f n,kAn initial phase shift that satisfies the above conditions can be determined by an empirical approach, such as numerical simulation. The frequency components are added together with a random initial phase shift to generate a test probe signal. Next, the ratio of the maximum value of the test probe signal to the average value of the test probe signal is determined. These two steps are repeated with different random initial phase shifts to create a set of test probe signals. In actual subsurface structure exploration, the test probe signal with the smallest ratio is selected. In this approach, the initial phase shift can be derived from a random number generator but is not subsequently modified in the application. In particular, the initial phase shifts of at least two frequency components can be constant, i.e., identical, for each probe signal.

[0037] Clearly, discrete frequency component f n,k The (known) initial phase shift must be taken into consideration in processing. Preferably, step (c) of processing the echo signal is to process the initial phase shift so that at least two discrete frequency components f n,k This includes correcting the phase.

[0038] Similar to the initial phase shift, discrete frequency components f n,k The frequency is the frequency offset θ n,k It may be offset by different frequency components f n,k Different θ n,k Applying this can shorten the overall period time of the probe signal, i.e., the overall period, compared to the case where the frequency spacing Δf between adjacent frequencies is constant. However, the frequency offset θ n,k It is desirable that the frequency offsets are not too large, for example, separated by an interval of (-0.5, +0.5) × Δf, so that they do not overlap with adjacent frequencies. In particular, all frequency offsets θ n,k is, θ n,k =θ may be the same as =θ.

[0039] Preferably, frequency offset θ n,k This is the same for all probe signals in one cycle. Different θn,k By applying it to different cycles, the overall cycle time can be shortened. A further advantage of applying different frequency offsets to different cycles is the reduction of measurement errors due to spurious signals. Since such errors are different when the frequency offset is changed, such errors can be compensated by averaging over different cycles.

[0040] Sweeping of the probe signal Similar to the conventional SFCW signal, a probe signal having the above characteristics and whose frequency component f n,k changes can be concatenated to achieve a sweep of the probe signal covering a wide frequency range, for example, from 40 MHz to 3 GHz.

[0041] Therefore, in one embodiment, steps (a), (b) and (c) of the above method are repeated while changing the discrete frequency component f n,k in the probe signal. During subsequent iterations, the discrete frequency component f n,k of the probe signal is shifted by only the frequency offset, and this frequency offset may be constant. In particular, the frequency offset can be set to be greater than or equal to the bandwidth.

[0042] Preferably, the probe signal of subsequent iterations extends over an overall bandwidth of at least 1 GHz, particularly at least 2 GHz, or at least 3 GHz. The overall bandwidth can be defined particularly as the frequency range extended by the lowest frequency component and the highest frequency component. In particular, the overall bandwidth can cover frequencies of at least 40 MHz to 3 GHz, and high-resolution and high-penetration-depth radar data can be obtained for many geological and non-destructive testing applications.

[0043] The sweep of the probe signal in one example is three subsequent probe signals, namely, the frequency component f n,k covering frequencies from 40 MHz to 1 GHz for the first probe signal, f n,k covering the range from 1 GHz to 2 GHz for the second probe signal, f covering the range from 2 GHz to 3 GHzn,k It may include a third probe signal having the following characteristics:

[0044] In the case of sweeping the probe signal, step (c) of processing the echo signal preferably includes downmixing the echo signal with the local oscillator signal for at least a portion of the subsequent probe signal. This should be performed in particular before performing spectral analysis of the echo signal, i.e., in the analog domain. According to the values ​​in the above embodiment, the local oscillator signal may have frequencies between 1 GHz and 2 GHz, for example, a first local oscillator at 1 GHz and a second local oscillator at 2 GHz.

[0045] Downmixing the echo signal before sampling can ease the requirements for subsequent components, particularly the ADC. The specified input bandwidth of the ADC does not need to cover the entire bandwidth of the probe signal sweep. However, preferably, the input bandwidth of the ADC covers the bandwidth of one probe signal.

[0046] Device for exploring underground structures A second aspect of the present invention relates to an apparatus for exploring underground structures, comprising the following components. - A probe signal generator configured to generate a probe signal, particularly a probe signal having any of the characteristics described above. Preferably, the probe signal generator includes a digital signal generator and a digital-to-analog converter (DAC). Furthermore, the probe signal generator includes a plurality of discrete frequency components f n,k It is configured to generate a probe signal that includes [the specified element]. Preferably, the frequency resolution of the probe signal generator is 0.1 MHz or less. - A transmitting antenna configured to transmit probe signals as electromagnetic waves to a structure. - A receiving antenna configured to receive electromagnetic wave echo signals from a structure. The transmitting antenna and receiving antenna may each include a low-frequency sub-antenna and a high-frequency sub-antenna. The center frequency of the low-frequency sub-antenna can be, for example, 300-500 MHz, particularly around 380 MHz. The center frequency of the high-frequency sub-antenna is, for example, 1-1.5 GHz, particularly around 1.2 GHz. - An echo signal processing device configured to process echo signals to derive information about a structure. Preferably, the echo signal processor comprises an analog section, a digital section, and a sampling frequency f s It comprises an analog-to-digital converter (ADC) configured to sample the echo signal at a sampling frequency f. s This is less than twice the bandwidth of the probe signal.

[0047] Furthermore, the apparatus is preferably configured to perform the method described above.

[0048] In one embodiment, the echo signal processor includes, in its analog section, a local oscillator and a downmixer configured to downmix the echo signal with the local oscillator signal from the local oscillator. As described above, this has the effect that, in the case of sweeping multiple probe signals, the input bandwidth of the ADC does not need to cover the entire sweep bandwidth, but only the bandwidth of one probe signal.

[0049] Furthermore, the local oscillator may include at least two switchable analog oscillators. In particular, for the first part of the probe signal, the first oscillator among the analog oscillators is used as the local oscillator, for the second part of the probe signal, the second oscillator among the analog oscillators is used as the local oscillator, and for the third part of the probe signal, downmixing is not required, i.e., the downmixer can be omitted. In the case of the low-frequency sub-antenna and the high-frequency sub-antenna, the echo signal received by the high-frequency sub-antenna is downmixed with the local oscillator signal, and the echo signal received by the low-frequency sub-antenna may not require any downmixing at all. By downmixing, for example, the entire bandwidth from 40 MHz to 3 GHz in the above embodiment can be realized with an ADC having an input bandwidth of 1 GHz.

[0050] In one embodiment, the probe signal generator and the echo signal processor are at least partially implemented as a field programmable gate array (FPGA). Preferably, in a part of the signal generator, the FPGA can be configured to generate discrete frequency components f n,k with defined amplitudes and phases in particular. On the side of the echo signal processor, the FPGA is preferably configured to perform spectral analysis of the sampled echo signal and attribution of the corresponding f n,k to f n,k and conjugation of the frequency components. Furthermore, the FPGA can be configured to perform, for example, calibration of the acquired frequency components for the antenna effect and / or spatial averaging over a plurality of sweeps, for example. Since the FPGA can be configured to efficiently perform the above tasks, a high data acquisition speed can be realized. Furthermore, for example, the amount of data required for further processing by a conventional SFCW algorithm and / or transfer via an interface to a remote computing device can be reduced.

[0051] GPR array To further improve the efficiency of data acquisition, particularly to increase lateral coverage, an array of antennas can be formed. In such embodiments, the transmitting antenna comprises multiple transmitting antennas, and the receiving antenna comprises multiple receiving antennas. The transmitting and receiving antennas can be arranged in a staggered pattern laterally with respect to the data acquisition direction, and in particular, one transmitting antenna can always be placed between adjacent receiving antennas. The device can then be configured to operate the multiple transmitting and receiving antennas in succession, for example, as a pair of one transmitting antenna and one receiving antenna at a time, or as a pair of one transmitting antenna and two adjacent receiving antennas at a time.

[0052] Furthermore, it is preferable that at least two transmitting antennas have different polarizations and at least two receiving antennas have different polarizations. In this way, underground structures can be explored using electromagnetic waves with different polarizations, for example, horizontal polarization (HH) and vertical polarization (VV) or cross-polarization (HV, VH), thereby obtaining complementary information about the structure and, ideally, a more complete image of the structure. This is particularly effective for elongated conductive structures such as reinforcing bars and pipes.

[0053] Computer program A third aspect of the present invention relates to a computer program that includes instructions for carrying out the above method. In particular, the computer program may include instructions for causing the device to perform the steps of the above method.

[0054] Other preferred embodiments are described in the dependent claims, along with the following description. [Brief explanation of the drawing]

[0055] The present invention will be better understood and its other objectives will become clear from the following detailed description. Such a description is provided with reference to the accompanying drawings. [Figure 1a]Figure 1a shows a schematic frequency-to-time diagram of a conventional SFCW probe signal, an ultra-wideband probe signal, and an ultra-wideband SFCW probe signal according to an embodiment of the present invention. [Figure 1b] Figure 1b shows a schematic frequency-to-time diagram of a conventional SFCW probe signal, an ultra-wideband probe signal, and an ultra-wideband SFCW probe signal according to an embodiment of the present invention. [Figure 1c] Figure 1c shows a schematic frequency-to-time diagram of a conventional SFCW probe signal, an ultra-wideband probe signal, and an ultra-wideband SFCW probe signal according to an embodiment of the present invention. [Figure 2] Figure 2 shows a schematic power-to-frequency diagram of multiple discrete frequency components and the aliased frequency component after sampling in the echo signal according to the embodiment. [Figure 3] Figure 3 shows a schematic plot of the frequency difference between the aliased frequency component and its vicinity (after aliasing) according to the embodiment. [Figure 4] Figure 4 is a block diagram of the front end of a device for exploring underground structures according to an embodiment. [Figure 5] Figure 5 is a block diagram showing the components and functions of the apparatus according to the embodiment. [Modes for carrying out the invention]

[0056] Figure 1a shows the frequency sweep over time implemented in a conventional SFCW signal. A single-frequency probe signal is transmitted for a certain period of time ts. Subsequently, the frequency of the probe signal increases by a frequency step Δf, as described in the "Background Technology" section.

[0057] Figure 1b shows a typical UWB signal with a frequency f across a bandwidth Bw in frequency space. A UWB signal may contain a finite number of discrete frequency components or it may contain a continuous spectrum of frequencies in frequency space. If the UWB signal does not change over time (as shown in Figure 1b), the bandwidth Bw of the signal is equal to the total bandwidth Bwt. According to the FCC definition, UWB is defined in particular as an antenna transmission in which the bandwidth of the radiated signal exceeds 500 MHz or 20% of the arithmetic center frequency, whichever is smaller. Thus, in particular, every signal with Bw > 500 MHz is a UWB signal.

[0058] Figure 1c shows the sweep of probe signals for each bandwidth Bw, where the sweep covers the entire bandwidth Bwt. Each probe signal preferably has the same duration t0. Furthermore, each probe signal has multiple discrete frequency components f n,k In particular, a finite number of discrete frequency components f n,k This includes these discrete frequency components f n,k In particular, these can be separated by a frequency step Δf in the frequency domain.

[0059] Preferably, the bandwidth Bw of each probe signal in the sweep is at least 500 MHz, i.e., each probe signal is a UWB signal. This has the advantage that the possibility of interference with telecommunications systems is minimized and such signals comply with regulatory restrictions, particularly FCC regulations.

[0060] As an example, the bandwidth of each probe signal may be Bw = 1 GHz. The first swept probe signal may include frequency components from 400 to 1400 MHz, the second probe signal may include frequency components from 1400 to 2400 MHz, and the third probe signal may include frequency components from 2400 to 3400 MHz. In this embodiment, the total bandwidth is Bwt = 3 GHz. Generally, the frequency component f of the subsequent probe signal... n,k The frequency ranges extended by this method may or may not overlap, as in the above embodiment and as shown in Figure 1c, and in particular, they may be directly adjacent to each other.

[0061] Furthermore, in this embodiment, each probe signal has M=240 discrete frequency components f n,k It may include. For equally spaced frequency components, the frequency step Δf is approximately 4 MHz. Such an embodiment of the probe signal, in other words, facilitates the transmission of M=240 frequency components that are subsequently transmitted in a classical SFCW sweep as shown in Figure 1a. This theoretically allows for a data acquisition speed increase of up to M=240 times under the assumption that t0=ts. Conversely, this means that the acquisition speed can be increased to the speed of a car or drone traveling at, for example, 50-80 km / h, compared to the classical SFCW method. However, compared to the conventional SFCW method, each discrete frequency component f n,k It is a requirement that the information, particularly amplitude and / or phase information, can be uniquely derived from the probe signal without loss.

[0062] The above requirements raise a challenge regarding the processing of reflected signals received as echo signals from structures. The majority of the processing of the echo signal involves each frequency component f. n,k It is convenient to derive quantities that indicate the amplitude and / or phase of an echo signal in the digital domain. To transfer the received echo signal from the analog domain to the digital domain, i.e., to digitize the echo signal, sampling is typically performed by an ADC. However, conventional ADCs have a sampling frequency f s And there are limitations on the analog input bandwidth. For example, the maximum sampling frequency and analog input bandwidth of conventional, especially affordable, ADCs are 250 MHz and 1 GHz, respectively. According to the Nyquist theorem, f s From the echo signal sampled by f s Only frequency components up to / 2, i.e., frequency components within the first Nyquist zone of 0-125 MHz in this embodiment, can be uniquely determined. Clearly, UWB signals cannot be digitized without information loss by such conventional ADCs without further measures.

[0063] These problems are solved by the following means according to embodiments of the present invention: Multiple discrete frequency components f of the probe signal n,k As illustrated in Figure 2, the discrete frequency component f is distributed across different Nyquist zones, for example, Nyquist zones 1 to 4. In the illustrated embodiment, the discrete frequency component f n,k These are arranged at equal intervals in the frequency space separated by a frequency step Δf and have equal power. In general, f n,k They may have different powers and / or they may have unequal intervals. In particular, different f n,k The amplitude can be adapted, for example, to compensate for frequency-dependent attenuation in the path of electromagnetic waves traveling through a structure. Also, different f can be used to reduce the ratio of peak power to average power of the probe signal. n,k The phase can be adjusted.

[0064] Figure 2 shows the 13 discrete frequency components f of the probe signal. n,k This shows that n is the number of Nyquist zones and k is the number of frequency components in the nth Nyquist zone. Specifically, the probe signal in Figure 2 has frequency component f 1,1 ,f 1,2 ,f 1,3 ,f 2,1 ,f 2,2 ,f 2,3 ,f 2,4 ,f 3,1 ,f 3,2 ,f 3,3 ,f 4,1 ,f 4,2 ,f 4,3 Includes frequency f s After sampling, higher Nyquist zones, i.e., frequency components where n>1, are folded back to the first Nyquist zone, as shown by the arrows in Figure 2. The folded frequency components, after sampling, are located only in the first Nyquist zone, and f 1,k This means that the frequency does not change during sampling.

[0065] f in the above example s=250MHz, therefore, in the case of the first Nyquist zone of 0-125MHz, this means that the frequency component at 130MHz after sampling is indistinguishable from the frequency component at 120MHz. In general, aliased frequency component f' n,k The above formula applies to this. To avoid losing information, multiple f at the same frequency in the first Nyquist zone n,k It is undesirable for this to be folded back. In other words, the folded frequency component f' n,k They need to be distinguishable. And, for example, the f of the probe signal n,k and the corresponding line breaks f' n,k According to the table, f' n,k The original f of the probe signal n,k It can be uniquely attributed to it.

[0066] frequency component f n,k and sample frequency f s Further conditions for selecting Δf' are, as also see the explanation in Figure 2, any two folded frequency components f' n,k The fact that they differ by at least a minimum distance Δf' in frequency space is due to the fact that spectral analysis of a sampled echo signal, such as Fourier analysis, may only be able to decompose frequency components separated by at least Δf' into distinct frequency components. In particular, only in that case, f' n,k It is possible to uniquely determine the amplitude and / or phase of f. n,k Alternatively, preferred conditions for Δf' are shown above.

[0067] Generally, the discrete frequency component f of the probe signal n,k This can be defined empirically, through trial and error, or based on optimization, for example, by numerical simulation. In the first step, a suitable ADC such as the AD9683-250 is selected, and the analog input bandwidth and sampling frequency f are determined. s Confirm the following: Input bandwidth 1GHz, f s In the above embodiment where =250MHz, the probe signal f n,kThis can be distributed across eight Nyquist zones. In the second step, the minimum frequency distance Δf' after sampling can be defined, for example, based on the desired maximum probe signal duration t0 related to the acquisition rate. The defined Δf' is f' n,k The following conditions are given. In the third step, the number of frequency components M is selected, for example, M = 240. From the bandwidth Bw = 1 GHz and M, an initial frequency step Δf of approximately 4 MHz is set for further optimization. 0 Furthermore, the initial frequency offset f can be derived. O 0 That is, the frequency f1,1 can be selected. Δf 0 and f O 0 By selecting f n,k This is determined. In the fourth step, the corresponding f' n,k f' is calculated via the above formula, n,k It is checked whether the minimum frequency distance Δf' is met. If yes, f n,k This can be used as the probe signal. Otherwise, in the third step, Δf 0 and / or f O 0 Vary the value(s) and repeat the fourth step with the resulting(s) values(s).

[0068] Generally, the above f' n,k and f n,k As can be understood from the relationship, the relationship between Δf' and Δf is nonlinear. Therefore, as above, Δf 0 and / or f O 0 By changing f, we can find the appropriate f through a brute-force approach over multiple iterations. n,k It may be necessary to explore.

[0069] Figure 3 shows the frequency components f with M = 240. n,k An example of a suitable set is shown. Figure 3 shows each aliasing frequency component f' on the vertical axis. n,kThis indicates the distance to the nearest next adjacent frequency component. In this case, two folded f' n,k The minimum distance between them, Δf', is approximately 0.4 MHz. The fundamental (unfolded) frequency component f n,k is parameter f O The sampling frequency is determined by =2MHz and Δf=3.8MHz, and the sampling frequency is f s = 250MHz.

[0070] Figure 4 is a block diagram of the front end of a device for exploring underground structures according to an embodiment. This device is a GPR array comprising N transmitting antennas, for example, TX1, TX2, TX3 if N=3, and N receiving antennas, for example, RX1, RX2, RX3. Generally, the device may have only one transmitting antenna and one receiving antenna, or it may have multiple transmitting and receiving antennas, for example, in the range of 20 to 50 antennas. Also, the number of transmitting antennas and the number of receiving antennas do not necessarily coincide. Preferably, the multiple transmitting and receiving antennas are offset laterally from each other, i.e., offset laterally with respect to the acquisition direction. In this way, by multiplexing across different transmitting and receiving antenna pairs TXn-RXn using multiplexers MUX1 and MUX2, etc., multiple measurement lines can be measured pseudo-simultaneously.

[0071] Furthermore, the device may be equipped with low-frequency sub-antennas, such as RX1 and TX1, with a center frequency of 300-500 MHz in particular. The device may also be equipped with high-frequency sub-antennas, such as RX2 and TX2, with a center frequency of 1-1.5 GHz in particular. Thus, the device is suitable for transmitting probe signals with a total bandwidth Bwt of at least 3 GHz.

[0072] To obtain more information about underground structures, it is preferable to use electromagnetic waves with different polarizations. Therefore, at least two transmitting antennas should have different polarizations, for example, TX1 should be horizontally polarized and TX3 vertically polarized, and at least two receiving antennas should have different polarizations, for example, RX1 should be horizontally polarized and RX3 vertically polarized.

[0073] In the apparatus shown in Figure 4, f n,k The probe signal, including the probe signal, is generated by the FPGA and converted from digital to analog by a DAC, such as the AD9163, which can generate a signal bandwidth of up to 1 GHz within a range of up to 6 GHz. The DAC includes an integrated numerically controlled oscillator (NCO) and a mixer and is configured to generate the probe signal over a full bandwidth of, for example, 3 GHz. The transmitting components from the FPGA to MUX1 constitute the probe signal generator.

[0074] In contrast, an ADC, such as the AD9683-250, may have an input bandwidth of 1 GHz. Therefore, all frequencies above 1 GHz need to be mixed down. For this purpose, an RX mixing stage is added to the circuit between the receiving antenna RXN and the ADC. The RX mixing stage includes a mixer MIX and is configured to downmix the echo signal, for example, from above 1 GHz to below 1 GHz.

[0075] In a preferred embodiment (as shown in Figure 4), the RX mixing stage includes two switches SW1 and SW2 for transmitting the echo signal via mixer MIX or for shortcutting mixer MIX. For probe / echo signals with frequencies within the ADC input bandwidth, mixer MIX can be shortcutted. This is particularly the case for echo signals received by low-frequency sub-antennas. For probe / echo signals with frequencies exceeding the ADC input bandwidth, the echo signal can be downmixed by mixer MIX. This is particularly the case for echo signals received by high-frequency sub-antennas.

[0076] The echo signal, digitized by the ADC, is further processed by the FPGA, as described later with reference to Figure 5. The receiving end, i.e., the components between MUX2 and the FPGA, constitute the echo signal processor.

[0077] As shown in Figure 4, preferably, further electronic components may be present in the analog section of the device. On the transmitting side, i.e., between the DAC and the transmitting antenna TXN, a first bandpass filter FILT1 and a first amplifier AMP1 may be configured to shape and amplify the probe signal. On the receiving side, i.e., between the receiving antenna RXN and the ADC, a second (and third) amplifier AMP2 (AMP3) and a second (and third) bandpass filter FILT2 (FILT3) can do the same for the echo signal. Also, an adjustable attenuator ATT may be present between the receiving antenna RXN and the mixer MIX, in particular between the second bandpass filter FILT2 and the mixer MIX. The attenuator ATT preferably keeps the level of the echo signal constant. Furthermore, the transmitting and receiving antennas can be shortcutted via a calibration path CAL between the multiplexers MUX1 and MUX2, making antenna calibration easier, and in particular, allowing for later compensation of the antenna's effects in echo signal processing in the FPGA.

[0078] Figure 5 is a block diagram showing the components and functions of a similar device. In particular, Figure 5 shows a different embodiment of the device in Figure 4. On the transmitting side connected to the transmitting antenna TX, the FPGA uses an inverse fast Fourier transform (IFFT) to obtain the baseband discrete frequency component f from the corresponding amplitude coefficients and phase coefficients. b n,k It generates the amplitude coefficient and phase coefficient, as well as the baseband discrete frequency component f. b n,k This may be the same for all probe signals in the sweep. Depending on the frequency range of the probe signal, the baseband discrete frequency component f b n,kThe signal from the NCO is upmixed with the DAC's integrated mixer, iMIX. The NCO is controlled by a finite state machine (FSM) in the FPGA, such as a counter, and is configured to change the frequency of the NCO's signal so that a sweep of the subsequent probe signal is generated, for example, as schematically shown in Figure 1c.

[0079] In the receiver side of the device shown in Figure 5, the mixer MIX is configured to downmix the echo signal from the receiving antenna RX with the local oscillator signal generated by the local oscillator LO. In this case, LO is also controlled by a finite state machine FSM in the FPGA. In particular, the receiving FSM and the transmitting FSM are synchronized. The receiving FSM processes the echo signal within the baseband frequency range, specifically the baseband frequency component f within the input bandwidth of the ADC. b n,k It is configured to downmix to the corresponding frequency range.

[0080] Further processing is performed on the FPGA for the digitized echo signal (after the ADC). Generally, further processing of the echo signal is performed, at least partially, on the central processing unit CPU, and in particular on the CPU of a remote device, such as a personal computer, laptop computer, or tablet computer. In the block diagram of Figure 5, the functions of the blocks grouped as a library LIB are executed on the CPU. The interface between the FPGA of the device and the CPU of the remote device can be implemented via a wireless connection. Different function assignments to the FPGA and CPU may generally differ from those shown in Figure 5. Preferably, the amount of data obtained from the echo signal is reduced by processing in the FPGA, in particular, before the data is sent to the CPU of the remote device.

[0081] As the first step in digital signal processing, specifically on an FPGA, a Fast Fourier Transform (FFT) or similar spectral analysis is performed on the echo signal. As a result, each f' n,kA (complex-valued) quantity representing the amplitude and phase is obtained. As the second step, f' n,k f of the probe signal n,k Associated with the above, each even number of f f n,k The quantities representing amplitude and phase are conjugated (symbol CON in Figure 5). In the third step, the calibration data is loaded from memory MEM, and f n,k It can be used to calibrate quantities that indicate the amplitude and phase (symbol CAL in Figure 5). The calibration data may be related to antenna calibration, for example, as shown by the symbol CAL in Figure 4. As a fourth step, spatial average (symbol in Figure 5) The JPEG0007866041000002.jpg57) process is executed, which increases the signal-to-noise ratio (SNR).

[0082] The weighting, zero-padding, and, in particular, the additional steps of performing the inverse Fast Fourier Transform (IFFT), which can be performed on a CPU, are the same as in the conventional SFCW method. A description of these steps is provided, for example, in WO2018 / 161183A1. As a result, a conventional A scan is received, and this A scan is different f n,k Due to the bandwidth covered and the sweep of the probe signal, it simultaneously possesses a large penetration depth, good depth resolution, and high SNR.

[0083] Due to the specific characteristics of the probe signal, particularly the discrete frequency components fn and k, as described above, the acquisition speed of GPR surveys can be significantly improved while maintaining the advantages of the conventional SFCW method as described above. In particular, an acquisition speed of at least 80 km / h can be achieved with a spatial resolution of 50 mm or more, or 25 mm or more with a high-frequency sub-antenna. [Configuration 1] A method for exploring underground structures, (a) A step of transmitting the probe signal as an electromagnetic wave to the structure, (b) The step of receiving the electromagnetic wave echo signal from the structure, (c) A step of processing the echo signal in order to derive information relating to the structure, Includes, Step (c) of processing the echo signal involves a sampling frequency (f s ) includes sampling the echo signal, The probe signal has multiple discrete frequency components (f n,k ) including, The plurality of frequency components (f n,k At least two of these are located in different Nike zones. The nth Nikest zone is (n-1)*f s / 2 to n*f s It covers a frequency range up to / 2, f s is the sampling frequency, where n=1,2,... k=1...K, where K is the frequency component (f) of the aforementioned n Nyquist zone. n,k The number of ) method. [Configuration 2] The frequency component (f) of the aforementioned n Nyquist zone n,k ) is the reflection frequency component (f') of the first Nyquist zone. n,k ) is folded over,

number

Claims

1. A method for exploring underground structures, (a) The step of transmitting the probe signal as an electromagnetic wave to the structure, (b) The step of receiving the electromagnetic wave echo signal from the structure, (c) The step of processing the echo signal in order to derive information relating to the structure, Includes, Step (c) of processing the echo signal involves a sampling frequency (f ｓ ) includes sampling the echo signal, The probe signal has multiple discrete frequency components (f ｎ，ｋ ) including, The plurality of discrete frequency components (f ｎ，ｋ At least two of these are located in different Nike zones. The nth Nikest zone is (n-1) * f ｓ / 2 to n*f ｓ It covers a frequency range up to / 2, f ｓ is the sampling frequency, where n = 1, 2, ... k = 1 ...K is the frequency component (f) of the aforementioned Nyquist zone. ｎ，ｋ ) is the number of, In step (a) above, the plurality of discrete frequency components are transmitted simultaneously. Steps (a), (b), and (c) are repeated while varying the discrete frequency components (f n, k) of the probe signal. During subsequent iterations, the discrete frequency components (f n, k) of the probe signal are shifted by a frequency offset. method.

2. The frequency component (f ｎ，ｋ ) of the n-th Nyquist zone is folded above the folded frequency component (f' ｎ，ｋ ) of the first Nyquist zone, [Math 1] And, The discrete frequency component (f ｎ，ｋ ) is the aliased frequency component (f') for all n and k. ｎ，ｋ ) are made to be different from each other, In particular, the frequency component (f ｎ，ｋ ) are equally spaced in frequency space. The method according to claim 1.

3. At least one Nikest zone, in particular, n = 1...N, for all Nikest zones where N is at least 2, K > 1, in particular K > 10, or K > 20. The method according to claim 1.

4. The discrete frequency component (f ｎ，ｋ ) are placed in the Nikest zones up to the Nth Nikest zone, where N > 2, In particular, N is at least 4, or at least 6, or at least 8. The method according to claim 1.

5. The discrete frequency component (f ｎ，ｋ ) is the aliased frequency component (f') for all n and k. ｎ，ｋ ) is at least 1 / t ０ They have a minimum distance (Δf') from each other, and t ０ This is set to be the duration of the probe signal. The method according to claim 2.

6. The discrete frequency component (f ｎ，ｋ ) is the aliased frequency component (f') for all n and k. ｎ，ｋ ) are such that they have a minimum distance (Δf') from each other that is at least 0.1 MHz, and in particular at least 0.4 MHz. The method according to claim 2.

7. The probe signal has at least M discrete frequency components (f ｎ，ｋ ) including, M is at least 10, in particular at least 100, or at least 200. In particular, M is approximately N*K and / or In particular, the discrete frequency component (f ｎ，ｋ ) is the aliased frequency component (f') for all n and k. ｎ，ｋ ) but at least f ｓ / (4*M), in particular, at least f ｓ They are made to have a minimum distance (Δf') from each other that is / (4*N*K). The method according to claim 2.

8. The plurality of discrete frequency components (f) in the probe signal ｎ，ｋ All frequency components other than those listed are zero. The method according to claim 1.

9. The plurality of discrete frequency components (f ｎ，ｋ ) spans the bandwidth (Bw) of the probe signal, The bandwidth (Bw) of the probe signal is wider than the Nyquist zone, and in particular, f ｓ Wider than / 2, In particular, the bandwidth (Bw) of the probe signal is at least 500 MHz, and especially at least 1 GHz. The method according to claim 1.

10. The sampling frequency (f ｓ ) is 100-500 MHz, especially 200-300 MHz. The method according to claim 1.

11. Step (c) of processing the echo signal includes performing spectral analysis of the echo signal, in particular Fourier analysis. The method according to claim 1.

12. Step (c) of processing the echo signal is to process the aliased frequency component f' in the echo signal. ｎ，ｋ This includes determining the quantities that represent the respective amplitudes and / or phases of, In particular, step (c) of processing the echo signal includes the aliased frequency component (f' ｎ，ｋ ) corresponds to the discrete frequency component (f ｎ，ｋ This includes attributing to ) and / or, In particular, step (c) of processing the echo signal includes an even number of n frequency components (f ｎ，ｋ This includes conjugating the quantities that represent the amplitude and / or phase of ) The method according to claim 2.

13. The discrete frequency component (f ｎ，ｋ At least two initial phase shifts of the probe signal are such that the maximum amplitude of the probe signal has the same discrete frequency component (f ｎ，ｋ ) has the discrete frequency component (f ｎ，ｋ The initial phase shift of the virtual probe signal is smaller than the maximum amplitude of the virtual probe signal at which the initial phase shift is zero, and in particular, it is smaller by at least 10%, at least 25%, or at least 50%. The method according to claim 1.

14. In particular, the frequency offset is greater than or equal to the bandwidth (Bw), and / or In particular, the subsequent repeated probe signal spans a full bandwidth (Bwt) of at least 1 GHz, in particular at least 2 GHz, or at least 3 GHz, and in particular, the full bandwidth (Bwt) covers frequencies from at least 40 MHz to 3.44 GHz. The method according to claim 9.

15. The step (c) of processing the echo signal includes, at least a portion of the subsequent probe signal, in particular, downmixing the echo signal with the local oscillator signal before performing the spectral analysis of the echo signal, In particular, the local oscillator signal has a frequency of 1 to 2 GHz. The method according to claim 11.

16. A device for exploring underground structures, A probe signal generator configured to generate a probe signal, A transmitting antenna configured to transmit the probe signal as an electromagnetic wave to the structure, A receiving antenna configured to receive the electromagnetic wave echo signal from the aforementioned structure, An echo signal processor configured to process the echo signal in order to derive information relating to the aforementioned structure, Equipped with, The apparatus is configured to perform the method according to any one of claims 1 to 15. Device.

17. The transmitting antenna (TX) and the receiving antenna (RX) each include a low-frequency sub-antenna and a high-frequency sub-antenna, In particular, the center frequency of the low-frequency sub-antenna is 300 to 500 MHz, and / or In particular, the center frequency of the high-frequency sub-antenna is 1 to 1.5 GHz. The apparatus according to claim 16.

18. The probe signal generator comprises a digital signal generator and a digital-to-analog converter (DAC). The apparatus according to claim 16.

19. The probe signal generator has multiple discrete frequency components (f ｎ，ｋ It is configured to generate a probe signal that includes ) In particular, the frequency resolution of the probe signal generator is 0.1 MHz or less, and / or In particular, the probe signal generator and the echo signal processor are implemented, at least in part, as field-programmable gate arrays (FPGAs). The apparatus according to claim 16.

20. The echo signal processor has a sampling frequency (f ｓ The system comprises an analog section, a digital section, and an analog-to-digital converter (ADC) configured to sample the echo signal, In particular, the sampling frequency (f ｓ ) is less than twice the bandwidth (Bw) of the probe signal, and / or, In particular, the echo signal processor includes, in its analog section, a local oscillator (LO) and a downmixer configured to downmix the echo signal with the local oscillator signal from the local oscillator (LO), wherein the local oscillator (LO) comprises at least two switchable analog oscillators. The apparatus according to claim 16.

21. The aforementioned transmitting antenna (TX) comprises a plurality of transmitting antennas (TX1, TX2, TXN), The receiving antenna (RX) comprises a plurality of receiving antennas (RX1, RX2, RXN), The device is configured to operate the plurality of transmitting antennas and receiving antennas in succession, with one transmitting antenna and one receiving antenna pair operating at a time. In particular, at least two transmitting antennas have different polarizations and / or In particular, at least two receiving antennas have different polarizations. The apparatus according to claim 16.

22. A computer program element comprising an instruction causing the apparatus according to claim 16 to perform a step of the method according to claim 1.

Images