Appartus and method for detecting and identifying foreign matter in shallow stratum on basis of distributed acoustic wave sensing

Abstract

An apparatus for detecting and identifying foreign matter in a shallow stratum on the basis of distributed acoustic wave sensing. The apparatus comprises: a vibration-sensitive optical cable (1), a distributed acoustic wave sensing demodulator (2), a seismic source system (3), a vibration data processing unit (4), a velocity structure inversion unit (5) and an artificial intelligence positioning and identification unit (6), wherein when performing detection and identification, the distributed acoustic wave sensing demodulator (2) transmits a vibration signal to the vibration data processing unit (4) for preprocessing, so as to acquire a waveform signal; the velocity structure inversion unit (5) inverts a shallow stratum underground velocity structure on the basis of the waveform signal, so as to acquire the shallow stratum underground velocity structure; the location of foreign matter in a shallow stratum is identified by means of the underground velocity structure and an abnormal waveform; and the type of the foreign matter in the shallow stratum is identified by means of the artificial intelligence positioning and identification unit (6), image noise reduction and object segmentation. In the present application, a vibration signal is acquired by using a distributed optical cable acoustic wave sensing method, velocity imaging is performed on a shallow stratum velocity structure, and the positioning of shallow stratum underground foreign matter and the identification of the type of the foreign matter are realized by utilizing artificial intelligence. The present application further relates to a method for detecting and identifying foreign matter in a shallow stratum on the basis of distributed acoustic wave sensing.

Description

Shallow layer foreign body detection and identification device and method based on distributed acoustic wave sensing Technical Field

[0001] The present invention relates to the field of underground foreign body detection and identification, and in particular to a device and method for detecting and identifying shallow stratum foreign bodies based on distributed acoustic wave sensing. Background Art

[0002] Shallow underground foreign objects refer to objects or features found in the shallow soil or underground structures that are different from the surrounding environment. These objects or features include rock masses, underground garbage, pollutants, water sources, and underground pipelines. Shallow underground foreign objects can affect building stability, increase construction costs, protect environmental safety, damage underground pipeline facilities, and threaten the safe progress of construction. Underground foreign object detection is typically performed using geophysical methods, but the accuracy and depth of various current detection systems need to be improved. Therefore, to ensure the smooth progress of construction projects and reduce additional costs and delays, it is crucial to develop methods for detecting shallow underground foreign objects.

[0003] The main methods currently used for detecting underground foreign objects include magnetic detection, electromagnetic detection, geological radar detection, and radiometric detection. Magnetic detection is based on the influence of underground foreign objects on the geomagnetic field. These foreign objects usually have magnetic properties different from those of the surrounding geological materials. When there are foreign objects with different magnetic properties, they will distort the surrounding geomagnetic field, resulting in geomagnetic anomalies. By measuring the strength and direction of the geomagnetic field, magnetic surveys can detect and locate underground foreign objects. The principle of electromagnetic detection of underground foreign objects is based on the response of underground materials to electromagnetic fields. When electromagnetic waves pass through underground media, their propagation speed and direction are affected by the conductivity and dielectric constant of the medium. Underground foreign objects usually cause changes in electromagnetic parameters, thereby generating electromagnetic anomalies. By measuring the characteristics of electromagnetic wave propagation underground, including the amplitude and phase of the electromagnetic field, as well as the frequency response, electromagnetic methods can detect and locate underground foreign objects.

[0004] Both magnetic and electromagnetic methods can detect and locate foreign objects in shallow underground layers, but each has its own limitations. Magnetic detection can only detect magnetic foreign objects or geological bodies that exist underground. It is less effective for detecting non-magnetic materials. Furthermore, magnetic data is easily affected by artificial magnetic fields on the ground, requiring data correction during detection. Interpretation of electromagnetic detection data is relatively complex, requiring consideration of the distribution of the conductivity and dielectric constant of the underground medium. Furthermore, electromagnetic detection is not suitable for geological environments with high conductivity or high dielectric constant. The selection of the appropriate method for both methods depends on the specific geological conditions and the characteristics of the exploration object.

[0005] Geological radar detection is a method of detecting underground structures and foreign objects by emitting high-frequency electromagnetic waves and receiving their reflected signals. During operation, the geological radar system emits electromagnetic waves that penetrate the surface and interact with different materials or foreign objects underground. When electromagnetic waves encounter underground material interfaces, gaps or foreign objects, they are reflected, refracted and absorbed. These reflected signals are received and used to generate underground images or profiles. By analyzing the characteristics of the reflected signal such as time delay, intensity and frequency, the nature, depth and location of underground foreign objects can be identified. However, the geological radar method requires point-by-point measurement, which is less efficient in surveys of large underground areas, has a limited detection range, and is difficult to process due to the large amount and complexity of the data generated by the geological radar.

[0006] Radiometry is a method of detecting underground foreign matter by using rays or particles to penetrate underground media and detecting their attenuation. It is done by emitting rays or particles (such as gamma rays, neutrons) into the ground. These rays are absorbed or scattered when passing through different underground materials, and then form a specific energy spectrum or intensity distribution on the surface or on the detector. By analyzing the interaction of rays or particles in the underground medium, the location, density, composition and other information of underground foreign matter can be determined. Radiometry requires the use of radioactive isotopes or particle beams, which poses a radiation safety risk; data processing is complex, and the interpretation of results may be complicated. In addition, radiometry is sensitive to the specific physical properties of the underground medium and may not be able to detect non-radioactive foreign matter.

[0007] In recent years, distributed acoustic sensing technology has rapidly developed due to its advantages such as strong environmental adaptability, resistance to electromagnetic interference, ease of installation, and high data acquisition density. It has been widely used in geophysical sensing fields such as earthquake monitoring, oil and gas exploration, and pipeline intrusion detection. Conventional underground foreign object detection methods generally suffer from low resolution, poor environmental adaptability, complex data processing, and are unable to accurately locate and identify foreign objects in shallow layers.

[0008] How to identify the types of shallow underground foreign objects, improve the positioning accuracy and type identification capabilities of shallow underground foreign objects, and reduce potential geological risks are technical problems that need to be solved urgently.

[0009] Summary of the Invention

[0010] Purpose of the invention: In response to the technical problems existing in the prior art, the present invention proposes a device and method for detecting and identifying shallow foreign objects based on distributed acoustic wave sensing. For shallow underground foreign objects in geological engineering, a vibration-sensitized optical cable is pre-buried in the shallow layer, and an active source signal is emitted through a seismic source system. A highly sensitive, high-precision and high-spatial-resolution distributed optical cable acoustic wave sensing method is used to obtain vibration signals for detection, and velocity imaging is performed on the shallow velocity structure along the optical cable, thereby improving the accuracy and resolution of shallow foreign object detection. Combined with artificial intelligence, the positioning of shallow underground foreign objects and the identification of their types are realized.

[0011] Technical solution: The shallow-layer foreign body detection and identification device based on distributed acoustic wave sensing of the present invention includes a vibration-sensitized optical cable, a distributed acoustic wave sensor demodulator, a source system, a vibration data processing unit, a velocity structure inversion unit, and an artificial intelligence positioning and identification unit;

[0012] The vibration-sensitized optical cable has a fiber core, which is covered with a vibration-sensitized cladding. A vibration-low-loss Bingham gel is filled between the vibration-sensitized cladding and the sheath, and optical cable-sand coupling enhancement fins are added to the sheath.

[0013] The vibration-sensitizing optical cable is connected to the distributed acoustic wave sensor demodulator; the distributed acoustic wave sensor demodulator collects the vibration data of the optical cable and transmits it to the vibration data processing unit; the vibration data processing unit is connected to the velocity structure inversion unit; and the artificial intelligence positioning and identification unit is connected to the vibration data processing unit and the velocity structure inversion unit.

[0014] The method for detecting and identifying foreign objects in shallow layers based on distributed acoustic wave sensing of the present invention comprises the following steps:

[0015] (1) Wrapping the fiber core with a vibration-sensitizing cladding, filling the space between the vibration-sensitizing cladding and the sheath with a vibration-low-loss Bingham gel, and adding optical cable-sand coupling enhancement fins on the sheath to make a vibration-sensitizing optical cable;

[0016] (2) Shallowly bury the vibration-sensitized optical cable in the detection area and check the line of the vibration-sensitized optical cable and the cable-soil coupling;

[0017] (3) Connecting the vibration-sensitized optical cable to the distributed acoustic wave sensor demodulator and setting the sampling parameters of the distributed acoustic wave sensor demodulator;

[0018] (4) The active source signal is transmitted through the source system, and the distributed acoustic wave sensor demodulator collects the vibration data of the vibration-sensitized optical cable at a set time;

[0019] (5) The distributed acoustic wave sensor demodulator transmits the collected vibration data to the vibration data processing unit, which pre-processes the vibration data and displays the waveform changes of the vibration signal in the optical cable over time. The optical cable channel where the shallow foreign object is located is obtained through the abnormal waveform, and the plane position of the underground foreign object is determined.

[0020] (6) The vibration data processing unit transmits the pre-processed vibration signal to the velocity structure inversion unit to perform shallow underground velocity structure inversion, obtain the shallow underground velocity structure of the detection area, analyze the ground depth of the shallow foreign body based on the abnormal shear wave velocity, and determine the three-dimensional coordinates of the shallow foreign body;

[0021] (7) The artificial intelligence positioning and identification unit receives the data images from the vibration data processing unit and the velocity structure inversion unit, uses the shallow layer foreign body image denoising method to denoise the data images, and performs binary semantic segmentation on the obtained denoised images to separate the shallow layer foreign bodies from the background of the denoised images along the boundaries; uses a deep learning algorithm to train a deep learning model for target detection based on the object contours and elastic wave response characteristics of different shallow layer foreign bodies, and identifies the types of shallow layer foreign bodies in the segmented images.

[0022] In step (1), when optical cable-sand coupling enhancement fins are added to the sheath, the type of optical cable-sand coupling enhancement fins and the spacing of the optical cable-sand coupling enhancement fins are determined according to the density of the rock and soil medium and the soil.

[0023] In step (6), the velocity structure inversion unit performs shallow underground velocity structure inversion using a full waveform imaging method of a spectral element method based on the vibration signal of the vibration data processing unit to obtain the shallow underground velocity structure of the detection area.

[0024] In step (6), the vibration data processing unit transmits the preprocessed vibration signal to the velocity structure inversion unit, uses the middle section channel of the optical cable as a virtual source, performs cross-correlation calculation with the channel in the area where the abnormal waveform exists, and uses the phase weighted superposition method to superimpose the cross-correlation results to obtain the cross-correlation function between the detection areas; uses the phase shift method to extract the surface wave dispersion curve of the underground structure of the detection area, and uses the random sampling algorithm of the Monte Carlo method to invert the underground velocity structure of the detection abnormal area through multiple iterations.

[0025] In step (7), the artificial intelligence positioning and recognition unit uses a bilateral filtering method to filter the noisy image in the underground foreign body waveform velocity image obtained by the vibration data processing unit and the velocity structure inversion unit, suppressing the interference of the image noise on the detection and boundary recognition of shallow foreign body targets; the denoised image is annotated along the edge of the shallow foreign body in the image using the LabelMe tool to obtain a label map, and an image detection dataset and a semantic segmentation dataset are established.

[0026] In step (7), the artificial intelligence positioning and recognition unit uses the YOLO-V4 network in deep learning to detect shallow-layer foreign objects in the image using the established image detection dataset and semantic segmentation dataset. The established deep learning model for target detection is trained by inputting the velocity structure and imaging features of different types of shallow-layer foreign objects to identify the types of shallow-layer foreign objects.

[0027] In step (7), the image denoising, semantic segmentation and precise identification of shallow foreign bodies based on artificial intelligence underground structure surface wave imaging are used to identify the types of shallow foreign bodies in the segmented image.

[0028] In step (1), the material of the vibration-sensitizing cladding is determined by the optical cable material and the rock and soil medium and the density-characteristic doping material.

[0029] In step (4), a rare earth giant magnetostrictive source excitation system is used to excite the active source signal.

[0030] Working principle: The present invention adopts a distributed vibration-sensitizing optical cable containing a vibration low-loss Bingham body-filled gel and a vibration-sensitizing cladding, and adds optical cable-sand coupling enhancement fins. The optical cable-sand coupling enhancement fins are arranged on the surface of the vibration-sensitizing optical cable at a certain interval to improve the optical cable's perception of underground vibration signals, study the vibration mechanism of the optical cable and different rock and soil bodies and foreign objects of different scales, achieve high-resolution inversion, and use a deep learning algorithm to identify the types of shallow foreign objects in the segmented image according to the object contours and elastic wave response characteristics of different shallow foreign objects.

[0031] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0032] (1) The present invention combines the vibration-sensitized optical cable with the distributed optical cable acoustic wave sensing method, adds a vibration low-loss Bingham body filling gel and a vibration-sensitized cladding material, thereby significantly improving the measurement sensitivity of the distributed optical cable acoustic wave sensing, and thus more accurately obtaining the surface rock and soil characteristics within the range of the vibration signal from the active source to the optical cable, thereby improving the accuracy of locating and identifying foreign objects in shallow underground layers.

[0033] (2) The present invention adopts a high-frequency, high-power rare earth giant magnetostrictive ultrasonic source excitation system and a full-waveform imaging method based on the spectral element method to image the underground three-dimensional velocity structure along the optical cable, thereby obtaining higher-resolution and accurate underground structure information, and then locating foreign objects in shallow underground layers.

[0034] (3) The present invention adopts image denoising, semantic segmentation and precise identification of shallow foreign bodies based on artificial intelligence underground structure surface wave imaging to identify the types of shallow foreign bodies in the segmented images, providing more accurate information and safety guarantees for ground construction projects and shallow stratum safety.

[0035] (4) The present invention combines artificial intelligence algorithms with image noise reduction and target segmentation methods to realize the identification of types of shallow underground foreign objects, improves the positioning accuracy and type identification ability of shallow underground foreign objects, and efficiently explores and manages geological resources, reduces potential geological risks, and ensures the safety and feasibility of geological engineering design. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] FIG1 is a schematic structural diagram of a device for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to the present invention;

[0037] FIG2 is a schematic structural diagram of a vibration-sensitized optical cable according to the present invention;

[0038] FIG3 is a waveform diagram of underground foreign matter detected by the present invention;

[0039] FIG4 is a diagram of underground velocity anomalies of shallow stratum foreign bodies detected by the present invention;

[0040] FIG5 is a diagram showing the foreign body waveform learning and recognition by the artificial intelligence positioning and recognition unit of the present invention. DETAILED DESCRIPTION

[0041] As shown in Figure 1, the present invention's distributed acoustic sensing-based shallow-subsurface foreign object detection and identification device includes a vibration-sensitized optical cable 1, a distributed acoustic sensor demodulator 2, a source system 3, a vibration data processing unit 4, a velocity structure inversion unit 5, and an artificial intelligence positioning and identification unit 6. Source system 3 is a source excitation system, and velocity structure inversion unit 5 is a high-resolution velocity structure inversion unit.

[0042] The vibration-sensitizing optical cable 1 includes a fiber core 1-1, and the fiber core 1-1 is wrapped with a vibration-sensitizing cladding 1-2. A vibration low-loss Bingham filling gel 1-3 is filled between the vibration-sensitizing cladding 1-2 and the sheath 1-4. Optical cable-sand coupling enhancement fins 1-5 are added to the sheath 1-4 of the vibration-sensitizing optical cable 1 to enhance the vibration sensitivity of the sensing optical cable and the optical cable-sand coupling.

[0043] The vibration-sensitizing optical cable 1 is connected to a distributed acoustic sensor demodulator 2. The distributed acoustic sensor demodulator 2 collects vibration data from the optical cable and transmits it to a vibration data processing unit 4. This vibration data is preprocessed based on the vibration signals collected by the distributed acoustic sensor demodulator 2 to remove electromagnetic interference from related instruments. The waveform of the underground foreign object is then filtered using a waveform mutation method to determine its location. The vibration data processing unit 4 is connected to a velocity structure inversion unit 5. This unit inverts the underground velocity structure using the preprocessed waveform signals. It distinguishes abnormal wave velocities caused by foreign objects based on similar wave velocities within the same underground medium, thereby locating the underground foreign object. An artificial intelligence positioning and identification unit 6 is connected to the vibration data processing unit 4 and the velocity structure inversion unit 5. It uses bilateral filtering to filter noise from the acquired waveform velocity images of underground foreign objects. A deep learning model for target detection is trained based on the velocity structure and imaging characteristics of different types of shallow foreign objects. This allows for accurate identification of shallow foreign objects through image denoising, semantic segmentation, and precise identification of shallow foreign object types based on artificial intelligence-based underground structure surface wave imaging.

[0044] The method for detecting and identifying foreign objects in shallow layers based on distributed acoustic wave sensing of the present invention comprises the following steps:

[0045] (1) The optical cable cladding uses a vibration-sensitizing cladding 1-2, which is wrapped around the fiber core 1, and a vibration-low-loss Bingham filling gel 1-3 is filled between the fiber core 1-1 and the sheath 1-4 of the sensing optical cable. Optical cable-sand coupling enhancement fins 1-5 are added to the optical cable at a set interval to enhance the optical cable's ability to receive vibration, thereby manufacturing a vibration-sensitizing optical cable 1;

[0046] (2) In the detection area, the vibration-sensitized optical cable 1 is shallowly buried in the detection area on the surface of the ground, and the line integrity and cable-soil coupling of the vibration-sensitized optical cable are checked at the same time;

[0047] (3) Connect the vibration-sensitized optical cable 1 to the distributed acoustic wave sensor demodulator 2, check whether the optical cable channel is unobstructed, and set the sampling parameters of the distributed acoustic wave sensor demodulator 2;

[0048] (4) The active source signal is emitted by the source system 3, and the distributed acoustic wave sensor demodulator 2 collects the vibration data of the vibration-sensitized optical cable 1 at a set time;

[0049] (5) The distributed acoustic wave sensor demodulator 2 transmits the collected vibration data to the vibration data processing unit 4; the vibration data processing unit pre-processes the vibration data and displays the waveform changes of the vibration signal in the optical cable over time. The optical cable channel where the shallow foreign object is located is obtained through the abnormal waveform, and the plane position of the underground foreign object is determined.

[0050] (6) The vibration data processing unit 4 transmits the pre-processed vibration signal to the velocity structure inversion unit 5, performs shallow underground velocity structure inversion, obtains the shallow underground velocity structure of the detection area, analyzes the ground depth of the shallow foreign body based on the abnormal shear wave velocity, and determines the three-dimensional coordinates of the shallow foreign body;

[0051] (7) The artificial intelligence positioning and identification unit 6 receives the data image from the vibration data processing unit 4 and the velocity structure inversion unit 5, uses the shallow layer foreign body image denoising method to denoise the data image, and performs binary semantic segmentation on the obtained denoised image to separate the shallow layer foreign body from the background of the denoised image along the boundary; uses a deep learning algorithm to train a deep learning model for target detection based on the object contours and elastic wave response characteristics of different shallow layer foreign bodies, and identifies the types of shallow layer foreign bodies in the segmented image.

[0052] Wherein, in step (1), the vibration low-loss Bingham body filling gel 1-1 is filled between the fiber core and the optical cable sheath, which not only protects the optical cable from being damaged due to large strain, but also transmits small-scale vibration strain.

[0053] In step (1), the material of the vibration-sensitizing cladding is determined based on the optical cable material used and the rock and soil medium and density-characteristic doping material.

[0054] In step (1), the type and spacing of the optical cable-sand coupling enhancement fins are determined according to the different rock and soil media and different soil densities of the detection site, so as to expand the contact area between the optical cable and the surrounding rock and soil, enhance the cable-soil coupling, and improve the micro-vibration sensing efficiency.

[0055] In step (2), in this embodiment, the vibration-sensitized optical cable 1 is buried 0.2 m below the ground surface. During the burial process of the optical cable, the integrity of the optical cable line is checked and the backfill quality is guaranteed, thereby ensuring that the optical cable and the soil maintain cable-soil coupling.

[0056] In step (3), the connection between the vibration-sensitizing optical cable and the distributed acoustic sensor demodulator is checked and ensured, and the demodulator channel spacing, sampling frequency and other related parameters are designed according to the detection range and detection depth.

[0057] In step (4), a rare earth giant magnetostrictive source excitation system is used to excite active source signals, and source signals of relevant levels are excited according to the detection area.

[0058] In step (5), the vibration data processing unit pre-processes the collected vibration data to show the waveform changes of the vibration signal at each position of the optical cable over time, and determines the plane position of the underground foreign object by identifying the position of the optical cable channel where the abnormal waveform is located.

[0059] In step (6), the velocity structure inversion unit uses the full waveform imaging method of the spectral element method to perform shallow underground velocity structure inversion based on the vibration signal of the vibration data processing unit to obtain the shallow underground velocity structure of the detection area. Since the propagation speed of waves in a uniform medium is consistent, when the vibration wave propagates through the underground foreign body, it will change the underground velocity structure in the area where the foreign body is located, thereby identifying abnormal underground velocity, determining the depth of the underground foreign body, and then determining the three-dimensional coordinates of the shallow foreign body.

[0060] In step (7), the artificial intelligence positioning and recognition unit uses bilateral filtering to filter the noisy images in the underground foreign body waveform velocity images obtained by the vibration data processing unit and the velocity structure inversion unit, suppressing the interference of the image noise on the detection and boundary recognition of shallow foreign body targets. The denoised image is annotated along the edges of the shallow foreign body in the image using the LabelMe tool to obtain a label map, thereby establishing an image detection dataset and a semantic segmentation dataset.

[0061] In step (7), the artificial intelligence positioning and recognition unit uses the YOLO-V4 network in deep learning, and uses the established image detection dataset and semantic segmentation dataset to detect shallow-layer foreign body targets in the image. By inputting the velocity structure and imaging characteristics of different types of shallow-layer foreign bodies, the deep learning model is trained to identify the types of shallow-layer foreign bodies.

[0062] Example

[0063] The present invention's shallow-layer foreign object detection and identification device and method based on distributed acoustic wave sensing was used to locate and identify shallow-layer foreign objects at a specific site. The underground soil structure of the test site was as follows: silt at 0-2 meters, sandy soil at 2-5 meters, clay at 5-11 meters with a groundwater layer, and clay rock below 11 meters.

[0064] The method for detecting and identifying foreign objects in shallow layers based on distributed acoustic wave sensing of the present invention comprises the following steps:

[0065] (1) As shown in Figure 2, a vibration low-loss Bingham body filling gel 1-1 is added between the core and the sheath of the traditional sensing optical cable, and a vibration-sensitizing cladding 1-2 is wrapped around the core to transmit small-scale vibration strain. In addition, optical cable-sand coupling enhancement fins 1-3 are added to the optical cable at a spacing of 1m to enhance the cable-soil coupling, thereby improving the optical cable's perception gain effect on underground vibration signals.

[0066] (2) The total length of the vibration-sensitized optical cable 1 is 3 km. It is buried in the detection area in a rectangular array with a length of 800 m and a width of 700 m. The burial depth is 0.2 m. After burial, the integrity of the optical cable line is checked to ensure the quality of the backfill process and to ensure good coupling between the optical cable and the soil.

[0067] (3) Connect the vibration-sensitized optical cable 1 to the distributed acoustic wave sensor demodulator 2 to check whether the optical cable channel is unobstructed. Set the demodulator sampling frequency to 100 Hz, the channel spacing to 5 m, and the number of channels to 600.

[0068] (4) The active source signal is emitted by the rare earth giant magnetostrictive source excitation system, and the distributed acoustic wave sensor demodulator collects 24-hour optical cable vibration data;

[0069] (5) The distributed acoustic wave sensor demodulator transmits the collected vibration data to the vibration data processing unit for preprocessing and waveform extraction. The channel position is displayed based on the abnormal waveform. As shown in FIG3 , it is determined that the underground foreign object is below the plane position of channel 71, that is, there is an underground foreign object below 355 m from the end of the optical cable.

[0070] (6) The vibration data processing unit 4 transmits the pre-processed vibration signal to the velocity structure inversion unit. Using the middle channel of the optical cable as a virtual source, it sequentially performs cross-correlation calculations with the channels in the area where the abnormal waveform exists. The phase-weighted superposition method is used to superimpose the cross-correlation results to obtain the cross-correlation function between the detection areas. Since distributed acoustic wave sensor data is more accurate in collecting phase data, the phase shift method is used to extract the surface wave dispersion curve of the underground structure in the detection area. Then, based on the relevant information of the detection area, the Monte Carlo method random sampling algorithm is used to invert the underground velocity structure of the detection abnormal area through multiple iterations. As shown in Figure 4, there is an abnormal slowdown of underground velocity at a depth of 7m to 7.5m from the surface, indicating the presence of underground foreign matter at this depth.

[0071] (7) The artificial intelligence positioning and identification unit 6 receives the data images from the vibration data processing unit 4 and the velocity structure inversion unit 5, as shown in Figure 5. The shallow layer foreign body image denoising method based on the artificial intelligence algorithm performs binary semantic segmentation on the obtained denoised image, separates the shallow layer foreign body from the background of the denoised image along the boundary, and uses a deep learning algorithm to train a deep learning model for target detection based on the object contours and elastic wave response characteristics of different shallow layer foreign bodies, and identifies the types of shallow layer foreign bodies in the segmented image.

[0072] The planar position of the foreign object is determined based on the planar abnormal waveform obtained by the vibration data processing unit, and the underground depth of the underground foreign object is determined based on the velocity structure inversion unit, thereby determining the three-dimensional coordinate position of the shallow underground foreign object; a deep learning algorithm is used to train the deep learning model of target detection by training the object contours and elastic wave response characteristics of different shallow foreign objects, and then judgment is made based on the data obtained by the vibration data processing unit and the velocity structure inversion unit, and then the type of shallow underground foreign objects is identified.

Claims

1. A shallow stratum foreign body detection and identification device based on distributed acoustic wave sensing, characterized in that: It includes a vibration-sensitized optical cable (1), a distributed acoustic wave sensor demodulator (2), a seismic source system (3), a vibration data processing unit (4), a velocity structure inversion unit (5) and an artificial intelligence positioning and identification unit (6); The vibration-sensitized optical cable (1) has a fiber core (1-1) inside, a vibration-sensitized cladding (1-2) outside the fiber core (1-1), a vibration-low-loss Bingham filling gel (1-3) is filled between the vibration-sensitized cladding (1-2) and the sheath (1-4), and an optical cable-sand coupling enhancement fin (1-5) is added to the sheath (1-4); The vibration-sensitizing optical cable (1) is connected to a distributed acoustic wave sensor demodulator (2); the distributed acoustic wave sensor demodulator (2) collects optical cable vibration data and transmits the data to a vibration data processing unit (4); the vibration data processing unit (4) is connected to a velocity structure inversion unit (5); and the artificial intelligence positioning identification unit (6) is connected to the vibration data processing unit (4) and the velocity structure inversion unit (5).

2. A method for detecting and identifying foreign objects in shallow layers based on distributed acoustic wave sensing, characterized in that: The method is implemented by the shallow stratum foreign body detection and identification device based on distributed acoustic wave sensing as described in claim 1, and the method comprises the following steps: (1) Wrapping a vibration-sensitizing cladding (1-2) outside the fiber core (1), filling a vibration-low-loss Bingham body filling gel (1-3) between the vibration-sensitizing cladding (1-2) and the sheath (1-4), and adding an optical cable-sand coupling enhancement fin (1-5) on the sheath (1-4) to prepare a vibration-sensitizing optical cable (1); (2) shallowly burying the vibration-sensitizing optical cable (1) in the detection area, and inspecting the line of the vibration-sensitizing optical cable (1) and the cable-soil coupling; (3) connecting the vibration-sensitized optical cable (1) to the distributed acoustic wave sensor demodulator (2), and setting sampling parameters of the distributed acoustic wave sensor demodulator (2); (4) an active seismic source signal is emitted through the seismic source system (3), and the distributed acoustic wave sensor demodulator (2) collects vibration data of the vibration-sensitized optical cable (1) at a set time; (5) The distributed acoustic wave sensor demodulator (2) transmits the collected vibration data to the vibration data processing unit (4), and the vibration data processing unit (4) pre-processes the vibration data and displays the waveform change of the vibration signal in the optical cable over time, and obtains the optical cable channel where the foreign object in the shallow layer is located through the abnormal waveform, thereby determining the plane position of the foreign object in the underground; (6) The vibration data processing unit (4) transmits the pre-processed vibration signal to the velocity structure inversion unit (5), performs shallow underground velocity structure inversion, obtains the shallow underground velocity structure of the detection area, analyzes the ground depth of the shallow foreign body according to the abnormal shear wave velocity, and determines the three-dimensional coordinates of the shallow foreign body; (7) The artificial intelligence positioning and identification unit (6) receives the data image from the vibration data processing unit (4) and the velocity structure inversion unit (5), uses the shallow foreign body image denoising method to denoise the data image, and performs binary semantic segmentation on the obtained denoised image to separate the shallow foreign body from the background of the denoised image along the boundary; uses a deep learning algorithm to train a deep learning model for target detection based on the object contours and elastic wave response characteristics of different shallow foreign bodies, and identifies the types of shallow foreign bodies in the segmented image.

3. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2 is characterized in that: In step (1), when optical cable-sand coupling enhancement fins are added to the sheath (1-4), the type of optical cable-sand coupling enhancement fins and the spacing of the optical cable-sand coupling enhancement fins are determined according to the density of the rock medium and the soil.

4. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2 is characterized in that: In step (6), the velocity structure inversion unit (5) performs shallow underground velocity structure inversion based on the vibration signal of the vibration data processing unit (4) using a full waveform imaging method of the spectral element method to obtain the shallow underground velocity structure of the detection area.

5. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2 is characterized in that: In step (6), the vibration data processing unit (4) transmits the pre-processed vibration signal to the velocity structure inversion unit (5), uses the middle section channel of the optical cable as a virtual source, performs cross-correlation calculation with the channel in the area where the abnormal waveform exists, and uses the phase weighted superposition method to superimpose the cross-correlation results to obtain the cross-correlation function between the detection areas; uses the phase shift method to extract the surface wave dispersion curve of the underground structure of the detection area, and uses the random sampling algorithm of the Monte Carlo method to invert the underground velocity structure of the detection abnormal area through multiple iterations.

6. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2 is characterized in that: In step (7), the artificial intelligence positioning and identification unit uses a bilateral filtering method to filter the noisy images in the underground foreign body waveform velocity images obtained by the vibration data processing unit and the velocity structure inversion unit, thereby suppressing the interference of the image noise on the detection and boundary recognition of shallow foreign body targets; the denoised image is annotated along the edges of the shallow foreign bodies in the image using the LabelMe tool to obtain a label map, and an image detection dataset and a semantic segmentation dataset are established.

7. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 6 is characterized in that: In step (7), the artificial intelligence positioning and recognition unit uses the YOLO-V4 network in deep learning, and uses the established image detection dataset and semantic segmentation dataset to detect shallow-layer foreign objects in the image. By inputting the velocity structure and imaging features of different types of shallow-layer foreign objects, the established deep learning model for target detection is trained to identify the types of shallow-layer foreign objects.

8. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2 is characterized in that: In step (7), the image denoising, semantic segmentation and precise identification of shallow foreign bodies in the segmented image are used based on artificial intelligence underground structure surface wave imaging to identify the types of shallow foreign bodies.

9. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2 is characterized in that: In step (1), the material of the vibration-sensitizing cladding is determined by the optical cable material and the rock and soil medium and the material doped with density characteristics.

10. The method for detecting and identifying foreign objects in shallow strata based on distributed acoustic wave sensing according to claim 2, characterized in that: In step (4), a rare earth giant magnetostrictive seismic source excitation system is used to excite active seismic source signals.

Images