Tunnel multi-element lining quality information detection vehicle and tunnel lining information detection method

Abstract

The present application relates to the technical field of tunnel detection, and particularly relates to a tunnel multi-element lining quality information detection vehicle and a tunnel lining information detection method, in the tunnel multi-element lining quality information detection vehicle, the automobile chassis can walk along the tunnel ground; the lining multifunctional detection arm is arranged on the automobile chassis, and the end of the lining multifunctional detection arm away from the automobile chassis is provided with a radar monitoring module capable of walking along the tunnel wall surface; the inverted arch telescopic arm is arranged at the bottom of the automobile chassis, and the inverted arch telescopic arm is provided with at least two radar monitoring modules capable of walking along the tunnel ground; the radar monitoring module is used for detecting tunnel multi-element lining information; the information detection analysis module is electrically connected with the radar monitoring module, and the information detection analysis module obtains lining disease information according to the detection data analysis of the radar monitoring module. The present application can comprehensively evaluate the tunnel lining quality according to the lining disease information, and realize detection of different positions, and has the advantages of compact structure, convenient and flexible detection.

Description

Technical Field

[0001] This invention relates to the field of tunnel inspection technology, and in particular to a tunnel lining quality information inspection vehicle and a tunnel lining information inspection method. Background Technology

[0002] Tunnel lining is a crucial component of tunnel structure, and its quality directly impacts the safety and durability of the tunnel. By inspecting lining quality information, defects and problems can be detected promptly, effectively preventing accidents such as collapses and leaks caused by lining quality issues, and ensuring that the tunnel does not experience safety incidents due to lining quality problems during operation.

[0003] Traditional lining quality inspection typically involves manual handheld ground-penetrating radar mounted on an aerial work platform. During operation, personnel must continuously hold the radar device to maintain its contact with the tunnel contour, resulting in high labor intensity, low efficiency, and low accuracy. Existing mechanized lining quality inspection equipment mostly uses one or more inspection arms to perform single-line quality inspections using ground-penetrating radar technology, which cannot provide a comprehensive assessment of lining quality and has significant limitations. Furthermore, the inspection arms of existing mechanized lining quality inspection equipment are usually large and heavy, lacking flexibility and making it difficult for the equipment to turn or maneuver within the tunnel.

[0004] Therefore, it is necessary to provide a new vehicle for detecting the quality information of tunnel lining and a method for detecting tunnel lining information to solve the above-mentioned technical problems. Summary of the Invention

[0005] The main objective of this invention is to provide a tunnel lining quality information detection vehicle and a tunnel lining information detection method, aiming to solve the problem that existing detection equipment cannot comprehensively evaluate the lining quality and has significant limitations.

[0006] To achieve the above objectives, the present invention proposes a tunnel multi-layer lining quality information detection vehicle, comprising a vehicle chassis, a multi-functional lining detection arm, a radar monitoring module, an invert arch telescopic arm, and an information detection and analysis module. The vehicle chassis is capable of traveling along the tunnel surface. The multi-functional lining detection arm is mounted on the vehicle chassis, and the radar monitoring module, capable of traveling along the tunnel wall, is located at one end of the multi-functional lining detection arm away from the vehicle chassis. The invert arch telescopic arm is located at the bottom of the vehicle chassis, and at least two radar monitoring modules capable of traveling along the tunnel surface are mounted on the invert arch telescopic arm. The radar monitoring module is used to detect tunnel multi-layer lining information. The information detection and analysis module is electrically connected to the radar monitoring module, and the information detection and analysis module can analyze the detection data from the radar monitoring module to obtain lining defect information.

[0007] Optionally, the lining multifunctional detection arm includes a primary rotary base, a primary boom section, a secondary boom section, a tertiary boom section, a secondary rotary base, a tertiary rotary base, and a quaternary boom section. The primary rotary base is mounted on the vehicle chassis and is capable of rotating along a horizontal plane. The primary boom section, the secondary boom section, and the tertiary boom section are connected sequentially, and the primary boom section is connected to the primary rotary base. The primary boom section is capable of rotating between a horizontal and a vertical position. The secondary boom section is capable of rotating relative to the primary boom section, and the tertiary boom section is capable of extending and retracting along the axial direction of the secondary boom section. The secondary rotary base is connected to the tertiary boom section and is capable of rotating along a horizontal plane. The tertiary rotary base is mounted on the secondary rotary base and is capable of rotating along a vertical plane. The first end of the quaternary boom section is hinged to the tertiary rotary base, and the second end is connected to the radar monitoring module. The quaternary boom section is capable of multi-stage extension and retraction.

[0008] Optionally, the lining multi-functional inspection arm further includes a suspended platform arm, a four-stage rotary seat, a suspended platform body, a concrete multi-functional non-destructive testing instrument, and an anchor cable non-destructive testing instrument. The suspended platform arm is hinged to the three-stage boom section. The four-stage rotary seat is located at the end of the suspended platform arm away from the three-stage boom section, and the four-stage rotary seat can rotate along the horizontal plane. The suspended platform body is mounted on the four-stage rotary seat. The concrete multi-functional non-destructive testing instrument and the anchor cable non-destructive testing instrument are both mounted on the suspended platform body. The concrete multi-functional non-destructive testing instrument is used to test the strength and quality of the concrete, and the anchor cable non-destructive testing instrument is used to detect anchor defects.

[0009] Optionally, the radar monitoring module includes a radar antenna, rollers, a leveling mechanism, and a buffer mechanism. The radar antenna is mounted on the leveling mechanism and makes rolling contact with the tunnel wall via the rollers. The radar antenna is used to detect multi-element lining information. The leveling mechanism is mounted on the buffer mechanism. The buffer mechanism is respectively mounted on the fourth-stage boom or the invert arch telescopic boom, and the buffer mechanism buffers the leveling mechanism.

[0010] Optionally, the leveling mechanism includes a support structure, a mounting plate, an adjusting plate, a slider, and a floating drive component. The support structure is disposed between the mounting plate and the adjusting plate. The support structure includes a connecting rod and two cross-link groups. The two cross-link groups are spaced apart. Each cross-link group includes a first link and a second link that are hinged in a cross configuration. The two ends of the first link are respectively connected to the first link of the other cross-link group through a connecting rod. The two ends of the second link are respectively connected to the second link of the other cross-link group through a connecting rod.

[0011] The mounting plate is provided with slide rails on the side facing the support structure and on both sides of the adjusting plate; a slider is slidably disposed in each slide rail; the first end of the first connecting rod is hinged to the slider on the mounting plate, the second end of the first connecting rod is hinged to the adjusting plate, the first end of the second connecting rod is hinged to the slider on the side of the adjusting plate facing the support structure, and the second end of the second connecting rod is hinged to the mounting plate; the first end of the floating drive is hinged to the slider on the side of the adjusting plate away from the support structure, and the second end is hinged to a connecting rod.

[0012] Optionally, the leveling mechanism further includes a guide post, a supporting elastic element, and a pressure sensor. One end of the guide post is connected to the mounting plate via the pressure sensor, and the other end slides through the leveling plate. The supporting elastic element is sleeved on the guide post, and both ends of the supporting elastic element are connected to the mounting plate and the leveling plate, respectively. The pressure sensor is electrically connected to the information detection and analysis module.

[0013] Optionally, the buffer mechanism includes a mounting base, a rotating shaft, and buffer elastic elements. The mounting base is hinged to the adjusting plate via the rotating shaft. A plurality of buffer elastic elements are evenly distributed between the mounting base and the adjusting plate.

[0014] Optionally, the vehicle chassis is provided with a vertical plate, which is located at the front end of the lining multi-functional detection arm along the tunnel travel direction; the tunnel multi-dimensional lining quality information detection vehicle also includes a lidar, an inertial navigation system, and multiple sets of linear array cameras mounted on the vertical plate. The lidar is located on the central axis of the vehicle chassis and is electrically connected to the information detection and analysis module. The lidar is used to scan the tunnel to obtain point cloud data of the tunnel's full cross-section contour; the multiple sets of linear array cameras are arranged along the vertical ring and are symmetrically arranged with the lidar as the center. The linear cameras are electrically connected to the information detection and analysis module and are used to acquire high-definition visual images of the tunnel's full cross-section; the inertial navigation system is electrically connected to the information detection and analysis module and is used to acquire preliminary position information of the tunnel contour.

[0015] In addition, the present invention also provides a method for detecting tunnel lining information, which uses a tunnel multi-element lining quality information detection vehicle as described above to detect lining defects. The lining defect information includes the location information and classification and grading information of surface defects and the location information and classification and grading information of internal defects. The method is characterized by the following steps:

[0016] The vehicle, equipped with inertial navigation, lidar, and a linear array camera, is driven through the tunnel to be inspected to obtain a tunnel outline model and a high-definition image of the entire tunnel cross section.

[0017] The information detection and analysis module is used to identify the lining surface defects of the tunnel under inspection in the full-section high-definition appearance image of the tunnel, and obtain the location information of the lining surface defects in the tunnel outline model. Then, the defects are classified and graded to obtain the classification and grading information of the lining surface defects.

[0018] The information detection and analysis module plans several internal quality inspection channels based on the classification and grading information of the lining surface defects. Among them, at least one internal quality inspection channel is planned for the arch crown area, at least one internal quality inspection channel is planned for each of the left and right arch waist areas, and at least two internal quality inspection channels are planned for the invert arch area.

[0019] Select an internal quality inspection channel as the internal quality inspection channel based on the classification information of the lining surface defects.

[0020] The vehicle, designed for detecting the quality of multi-layered tunnel lining, passes through the tunnel to be inspected. Simultaneously, the radar monitoring module on the vehicle chassis and the radar monitoring module on the multi-functional lining inspection arm detect the internal quality of the tunnel to be inspected, obtaining the geological radar signal spectrum of the tunnel.

[0021] The information detection and analysis module is used to identify internal defects in the lining of the tunnel under inspection in the ground-penetrating radar signal spectrum, and obtain the location information of the internal defects in the tunnel outline model. Then, the internal defects are classified and graded to obtain the classification and grading information of the internal defects in the lining.

[0022] Optionally, the information detection and analysis module includes an information detection and analysis model, and the method for constructing the information detection and analysis model specifically includes:

[0023] ① Construct a database of apparent defects in lining and a database of internal defects in lining; specifically:

[0024] The lining appearance defect database is obtained by classifying and grading the defect images in the initial appearance defect database according to the preset classification and grading logic of the lining appearance defects; wherein: the initial appearance defect database includes N defect images under different scenarios and working conditions.

[0025] The initial internal defect database is obtained by classifying and classifying the ground radar signal spectra of the lining defects according to the preset classification and grading logic. The initial internal defect database includes M ground radar signal spectra.

[0026] ② The initial model was trained using the database of apparent defects in lining and the database of internal defects in lining to obtain a lining defect identification model;

[0027] ③ Based on the lining defect identification model, a big data comparison mechanism is introduced to obtain an information detection and analysis model, specifically:

[0028] High-definition visual images of the entire tunnel section or ground-penetrating radar signal spectrum maps are used as inputs to the lining defect identification model;

[0029] The apparent defects or internal defects of the lining output by the lining defect identification model are used as features to be detected.

[0030] Calculate the cosine similarity between the feature to be detected and the corresponding defect feature vectors of each type in the lining appearance defect database or the lining internal defect database.

[0031] The feature to be detected is assigned to the category corresponding to the highest cosine similarity, and then classified according to the corresponding pre-screening and grading logic to obtain an information detection and analysis model that can output the classification and grading information of the feature to be detected.

[0032] Optionally, a self-learning mechanism is introduced based on the information detection and analysis model. This mechanism is activated to update the information detection and analysis model after each tunnel detection operation at a preset distance is completed. Specifically:

[0033] When a tunnel detection operation is completed at a preset distance, semi-supervised annotation is used to annotate and correct the features to be detected for which all cosine similarities are lower than the preset similarity threshold.

[0034] The labeled features to be detected are added to the corresponding lining surface defect database and lining internal defect database. The lining defect recognition model is then retrained based on the updated lining surface defect database and lining internal defect database to obtain the updated information detection and analysis model.

[0035] Optionally, the defect images in the lining appearance defect database are classified and graded according to a preset classification and grading logic for lining appearance defects. Specifically, the classification and grading logic for lining appearance defects is as follows:

[0036] The Sobel operator is used to calculate the gray-level gradient value of each defect image data. If the gray-level gradient value along the defect extension direction is greater than or equal to the set gradient threshold and the number of consecutive pixels of the defect is greater than or equal to the set pixel threshold, or the ratio of the length of the defect to the maximum width is greater than or equal to the set ratio threshold and the maximum number of pixels in the width direction is less than or equal to the set width threshold, then the defect image is a crack defect. Crack defects are divided into three levels: minor, moderate and severe, according to the length of the crack defect.

[0037] The local gray-level variance of the defect image is calculated using a sliding window. If the average gray-level variance of the defect area within the sliding window is ≥B and the gray-level value change of two adjacent pixels is ≥C within the preset pixel statistics window, then the defect image is a pitted defect. Based on the area of ​​the pitted defect, the crack defect is divided into three levels: slight, moderate, and severe.

[0038] Calculate the average gray value of the defect area. If the average gray value of the defect area ≤ D, or the proportion of edge pixels of the defect area to the total pixels of the defect area ≤ F, or there is an anti-photon area with a gray value ≥ G and the area of the anti-photon area ≥ H in the defect area, then the defect image is a water leakage defect; and classify the crack defect into three levels: minor, general, and severe according to the area of the water leakage defect.

[0039] Optionally, classify and grade the ground penetrating radar signal spectrograms in the lining internal defect database according to the preset classification logic and grading logic of the lining internal defects, specifically:

[0040] Calculate the maximum signal amplitude of the defect area;

[0041] Judge: If the maximum signal amplitude A of the defect area ≥ 2.8A0, and the root mean square energy E of the signal ≥ 2.3E0, and the number of recognizable complete sine wave cycles in the defect area ≥ 3 and the coefficient of variation of the adjacent wave peak spacing ≤ I, and the rising edge time of the signal from the normal area to the defect area ≤ J, then the lining internal defect is a cavity defect; and classify it into three levels: minor, general, and severe according to the area of the cavity defect; where: A0 is the reference signal amplitude; E0 is the reference root mean square energy;

[0042] If the maximum signal amplitude A of the defect area satisfies 1.4A0 ≤ A < 2.8A0, and the root mean square energy E of the signal satisfies 1.1E0 ≤ E < 2.3E0, and the number of recognizable complete sine wave cycles in the defect area ≥ 2 and the coefficient of variation of the adjacent wave peak spacing ≤ K, and the rising edge time of the signal from the normal area to the defect area satisfies T1 < T < T2 and the time width corresponding to the transition area ≥ T3, then the lining internal defect is a void defect; and classify it into three levels: minor, general, and severe according to the area of the void defect; <s

[0043] If the maximum signal amplitude A of the defect area satisfies 1.0A0 ≤ A < 1.4A0, and the root mean square energy E of the signal satisfies 1.05E0 ≤ E < 1.1E0, and there is no complete sine wave cycle in the defect area and the coefficient of variation of the amplitude of the signal wave peak / valley ≥ L, and the attenuation time of the signal ≥ M, then the lining internal defect is an incompact defect; and classify it into three levels: minor, general, and severe according to the area of the incompact defect.

[0044] Optionally, the tunnel lining information detection method further includes: using a concrete multi-functional non-destructive detector and a bolt and cable non-destructive detector to perform non-destructive detection on the tunnel to be inspected to obtain the defect information of the concrete and the bolts.

[0045] Optionally, using a concrete multi-functional non-destructive detector and a bolt and cable non-destructive detector to perform non-destructive detection on the tunnel to be inspected to obtain the defect information of the concrete and the bolts, specifically including:

[0046] ① Randomly determine the test area of ​​the tunnel lining surface to be inspected, and determine the specific coordinate positions of several test points on the test area based on the location information of the lining defects.

[0047] ② The information detection and analysis module plans the automatic movement path of the main body of the suspended basket on the tunnel multi-layer lining quality information detection vehicle according to the specific coordinate position of the point to be measured, so as to move the suspended basket to the area to be measured;

[0048] ③ Use a four-stage boom to move the main body of the suspended platform to the test point positions in the test area;

[0049] ④ Use a multi-functional concrete non-destructive testing instrument and an anchor cable non-destructive testing instrument to perform non-destructive testing on the concrete and anchor at the test point to obtain defect information of the concrete and anchor.

[0050] In this invention, the multi-functional lining detection arm can adjust the position of the radar monitoring module on the tunnel wall and, driven by the vehicle chassis, moves along the tunnel wall to detect multi-dimensional lining information. The inverted arch telescopic arm can extend and retract along the width of the vehicle chassis to adjust the distance between the two radar monitoring modules on the inverted arch telescopic arm. The vehicle chassis, through the inverted arch telescopic arm, moves the two radar monitoring modules along the tunnel floor to detect multi-dimensional lining information on the tunnel floor. The information detection and analysis module can analyze the detection data from the radar monitoring module to obtain lining defect information, enabling operators to comprehensively evaluate the tunnel lining quality based on the lining defect information. It also allows for detection at different locations and has the advantages of compact structure and convenient, flexible detection. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0052] Figure 1 This is a schematic diagram of the structure of the tunnel multi-layer lining quality information detection vehicle in an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the structure of the lining multifunctional detection arm in an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of the radar monitoring module in an embodiment of the present invention;

[0055] Figure 4 for Figure 3A sectional view;

[0056] Figure 5 This is a schematic diagram of the installation of the inverted arch telescopic arm and the radar monitoring module in an embodiment of the present invention;

[0057] Figure 6 This is a schematic diagram of the installation of the lidar and the linear array camera in an embodiment of the present invention;

[0058] Figure 7 This is a schematic diagram of the operation of the tunnel lining quality information detection vehicle for lining inspection of the tunnel under inspection in an embodiment of the present invention;

[0059] Figure 8 This is a schematic diagram of the operation of a tunnel multi-layer lining quality information detection vehicle performing non-destructive testing in an embodiment of the present invention.

[0060] Explanation of icon numbers:

[0061] 1. Automobile chassis; 1.1 Vertical plate; 1.2 Covering parts; 2. Lining multi-functional inspection arm; 2.1 First-stage slewing base; 2.2 First-stage boom section; 2.3 Second-stage boom section; 2.4 Third-stage boom section; 2.5 Second-stage slewing base; 2.6 Third-stage slewing base; 2.7 Fourth-stage boom section; 2.8 Suspended basket arm; 2.9 Fourth-stage slewing base; 2.10 Suspended basket body; 3. Radar monitoring module; 3.1 Radar antenna; 3.2 Rollers; 3.3 Leveling mechanism; 3.3.1 Support structure; A 1. Connecting rod; A2. Cross linkage assembly; 3.3.2. Mounting plate; 3.3.3. Adjusting plate; 3.3.4. Slider; 3.3.5. Floating drive component; 3.3.6. Slide rail; 3.3.7. Guide column; 3.3.8. Support elastic component; 3.3.9. Pressure sensor; 3.3.10. Limiting plate; 3.4. Buffer mechanism; 3.4.1. Mounting base; 3.4.2. Rotating shaft; 3.4.3. Buffer elastic component; 4. Inverted arch telescopic arm; 5. LiDAR; 6. Linear scan camera.

[0062] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0064] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0065] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0066] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0067] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0068] This invention proposes a multi-element tunnel lining quality information detection vehicle and a tunnel lining information detection method, aiming to solve the problem that existing detection equipment cannot comprehensively evaluate the lining quality and has significant limitations.

[0069] Example 1

[0070] like Figure 1As shown, the tunnel multi-layer lining quality information inspection vehicle includes a vehicle chassis 1, a multi-functional lining inspection arm 2, a radar monitoring module 3, an invert arch telescopic arm 4, and an information detection and analysis module. The vehicle chassis 1 can travel along the tunnel surface. The multi-functional lining inspection arm 2 is mounted on the vehicle chassis 1, and a radar monitoring module 3 that can travel along the tunnel wall is located at the end of the multi-functional lining inspection arm 2 away from the vehicle chassis 1. The invert arch telescopic arm 4 is located at the bottom of the vehicle chassis 1, and at least two radar monitoring modules 3 that can travel along the tunnel surface are provided on the invert arch telescopic arm 4. The radar monitoring module 3 is used to detect tunnel multi-layer lining information. The information detection and analysis module is electrically connected to the radar monitoring module 3, and the information detection and analysis module can analyze the detection data of the radar monitoring module 3 to obtain lining defect information. In actual operation, the multi-functional lining detection arm 2 can adjust the position of the radar monitoring module 3 on the tunnel wall, and under the action of the vehicle chassis 1, it drives the radar monitoring module 3 to move along the tunnel wall to detect multi-dimensional lining information of the tunnel wall; and the inverted arch telescopic arm 4 can extend and retract along the width direction of the vehicle chassis 1 to adjust the distance between the two radar monitoring modules 3 on the inverted arch telescopic arm 4. The vehicle chassis 1 drives the two radar monitoring modules 3 to move along the tunnel ground through the inverted arch telescopic arm 4 to detect multi-dimensional lining information of the tunnel ground; the information detection and analysis module can analyze the detection data of the radar monitoring module 3 to obtain lining defect information, so that the operators can make a comprehensive assessment of the tunnel lining quality based on the lining defect information, and realize the detection of different positions. It has the advantages of compact structure and convenient and flexible detection.

[0071] See also Figure 2The lining multi-functional inspection arm 2 includes a primary swivel base 2.1, a primary boom section 2.2, a secondary boom section 2.3, a tertiary boom section 2.4, a secondary swivel base 2.5, a tertiary swivel base 2.6, and a quaternary boom section 2.7. The primary swivel base 2.1 is mounted on the vehicle chassis 1 and can rotate along the horizontal plane. The primary boom sections 2.2, 2.3, and 2.4 are connected sequentially, and the primary boom section 2.2 is connected to the primary swivel base 2.1. The primary boom section 2.2 can rotate in both horizontal and vertical positions. The secondary boom 2.3 can rotate relative to the primary boom 2.2, and the tertiary boom 2.4 can extend and retract along the axial direction of the secondary boom 2.3; the secondary slewing seat 2.5 is connected to the tertiary boom 2.4, and the secondary slewing seat 2.5 can rotate along the horizontal plane; the tertiary slewing seat 2.6 is set on the secondary slewing seat 2.5, and the tertiary slewing seat 2.6 can rotate along the vertical plane; the first end of the quaternary boom 2.7 is hinged to the tertiary slewing seat 2.6, and the second end is connected to the radar monitoring module 3, and the quaternary boom 2.7 can perform multi-stage extension and retraction. The first-stage slewing unit 2.1 can achieve 360° rotation, the first-stage boom 2.2 can achieve 90° pitch, the second-stage boom 2.3 can achieve 120° pitch relative to the first-stage boom 2.2, and the third-stage boom 2.4 can telescopically extend along the axis of the second-stage boom 2.3. The multi-functional tunnel lining inspection arm 2 is controlled by an electrical control system and a hydraulic system using both electronic and hydraulic control methods, enabling manual and wireless remote control operation. The second-stage slewing unit 2.5 and the third-stage slewing unit 2.6 can achieve rotation in the horizontal and vertical planes of the fourth-stage boom. The multi-functional tunnel lining inspection arm 2 is equipped with angle sensors, rotary encoders, and wire sensors. When used in conjunction with the lidar 5 in the multi-dimensional tunnel lining information detection and analysis system, it can achieve automatic one-key positioning of the multi-functional tunnel lining inspection arm 2. Through the linkage of multiple boom sections, the radar monitoring module 3 on the multi-functional tunnel lining inspection arm 2 can achieve multi-position adjustment, which is beneficial for comprehensive detection of tunnel lining.

[0072] Furthermore, the lining multi-functional inspection arm 2 also includes a suspended platform arm 2.8, a four-stage swivel base 2.9, a suspended platform body 2.10 (2.10), a concrete multi-functional non-destructive testing instrument, and an anchor cable non-destructive testing instrument. The suspended platform arm 2.8 is hinged to the three-stage boom section 2.4; the four-stage swivel base 2.9 is located at the end of the suspended platform arm 2.8 away from the three-stage boom section 2.4, and the four-stage swivel base 2.9 can rotate along the horizontal plane; the suspended platform body 2.10 is set on the four-stage swivel base 2.9; the concrete multi-functional non-destructive testing instrument and the anchor cable non-destructive testing instrument are both set on the suspended platform body 2.10. The concrete multi-functional non-destructive testing instrument is used to test the strength and quality of the concrete, and the anchor cable non-destructive testing instrument is used to test anchor defects. The suspended platform boom 2.8 is hinged at the end of the third-stage boom section 2.4, allowing for a 90° pitch relative to the third-stage boom section 2.4. A slewing seat is located at its end, enabling the suspended platform to rotate 360° via a speed reducer. The suspended platform serves as an auxiliary working platform, equipped with a multi-functional concrete non-destructive testing instrument and an anchor bolt non-destructive testing instrument. Operators can use the suspended platform body 2.10 to approach the tunnel contour and inspect lining information such as concrete strength and quality, and anchor bolt defects. The inspection data is transmitted back to the multi-dimensional lining information detection and analysis system. The system performs data analysis; the secondary slewing and connecting seat is used to connect the tertiary boom 2.4 and the lining inspection arm. The connection between the secondary slewing seat 2.5 and the tertiary boom 2.4 is equipped with a secondary slewing seat 2.5 that can rotate 360° in the horizontal direction, and the connection between the secondary slewing seat 2.6 and the lining inspection arm is equipped with a tertiary slewing seat 2.6 that can rotate 360° in the vertical direction. Through the combined action of the multi-stage boom, the radar monitoring module 3 at the front end of the suspended platform and the lining inspection arm can be positioned without blind spots along the circumferential contour of the tunnel. It has the advantages of compact structure and convenient and flexible detection.

[0073] In this embodiment, see reference Figures 3 to 5 The radar monitoring module 3 includes a radar antenna 3.1, rollers 3.2, a leveling mechanism 3.3, and a buffer mechanism 3.4. The radar antenna 3.1 is mounted on the leveling mechanism 3.3 and rolls against the tunnel wall via the rollers 3.2. The radar antenna 3.1 is used to detect multi-element lining information. The leveling mechanism 3.3 is mounted on the buffer mechanism 3.4. The buffer mechanism 3.4 is mounted on the fourth-stage boom 2.7 or the invert arch telescopic boom 4, and it buffers the leveling mechanism 3.3. The radar monitoring module 3 is equipped with several rollers 3.2, allowing it to move smoothly along the tunnel wall. The radar antenna 3.1 on the radar monitoring module 3 can transmit the detection data back to the multi-element lining information detection and analysis system in real time for data analysis and display. The leveling mechanism 3.3 and the buffer mechanism 3.4 on the radar monitoring module 3 ensure that the radar monitoring module 3 remains in contact with the tunnel wall during the detection process, guaranteeing the detection quality.

[0074] Specifically, the leveling mechanism 3.3 includes a support structure 3.3.1, a mounting plate 3.3.2, an adjusting plate 3.3.3, a slider 3.3.4, and a floating drive component 3.3.5. The support structure 3.3.1 is disposed between the mounting plate 3.3.2 and the adjusting plate 3.3.3. The support structure 3.3.1 includes a connecting rod A1 and two cross-link groups A2, which are spaced apart. Each cross-link group A2 includes a first link and a second link that are hinged in a cross configuration. The two ends of the first link are connected to the first link of the other cross-link group A2 via a connecting rod A1, and the two ends of the second link are connected to the second link of the other cross-link group A2 via a connecting rod A1. Mounting plate 3.3 .2 Slide rails 3.3.6 are provided on one side facing the support structure 3.3.1 and on both sides of the adjusting plate 3.3.3; a slider 3.3.4 is slidably arranged in each slide rail 3.3.6. The first end of the first connecting rod is hinged to the slider 3.3.4 on the mounting plate 3.3.2, and the second end of the first connecting rod is hinged to the adjusting plate 3.3.3. The first end of the second connecting rod is hinged to the slider 3.3.4 on the side of the adjusting plate 3.3.3 facing the support structure 3.3.1, and the second end of the second connecting rod is hinged to the mounting plate 3.3.2. The first end of the floating drive component 3.3.5 is hinged to the slider 3.3.4 on the side of the adjusting plate 3.3.3 away from the support structure 3.3.1, and the second end is hinged to a connecting rod A1. The leveling mechanism 3.3 also includes a guide post 3.3.7, a supporting elastic element 3.3.8, and a pressure sensor 3.3.9. One end of the guide post 3.3.7 is connected to the mounting plate 3.3.2 via the pressure sensor 3.3.9, and the other end slides through the leveling plate 3.3.3. The supporting elastic element 3.3.8 is sleeved on the guide post 3.3.7, and both ends of the supporting elastic element 3.3.8 are connected to the mounting plate 3.3.2 and the leveling plate 3.3.3, respectively. The pressure sensor 3.3.9 is electrically connected to the information detection and analysis module.

[0075] In this embodiment, the leveling mechanism 3.3 further includes a floating bracket and a limiting plate 3.3.10. Both ends of the floating bracket are slidably mounted on the leveling plate 3.3.3 via sliders 3.3.4. Limiting plates 3.3.10 are provided in the slide rails 3.3.6 corresponding to the two sliders 3.3.4 connected to the floating bracket. A floating drive component 3.3.5 is connected between the floating bracket and a connecting rod A1. When the radar monitoring module 3 is not in operation, the floating drive component 3.3.5 extends, and the sliders 3.3.4 on the floating bracket slide to the end away from the limiting plate 3.3.10. The floating drive component 3.3.5 pushes the two cross linkage groups A2 to retract via the connecting rod A1. At this time, the leveling mechanism 3.3 retracts and locks, preventing damage to the radar antenna 3.1 caused by the self-movement of the leveling mechanism 3.3 in the non-operating state. When the radar monitoring module 3 is in operation, the floating drive component 3.3.5 retracts, and the floating bracket... The frame moves the slider 3.3.4 to abut against the limit plate 3.3.10, and the two cross linkage groups A2 extend. At this time, the leveling mechanism 3.3 unfolds and is in an active state. When one side roller 3.2 contacts the tunnel contour, the contact force will compress the supporting elastic element 3.3.8 and move it along the guide post 3.3.7. Due to the uneven load, the first and second links in the cross linkage group A2 will move along the slide rail 3.3.6 respectively, so that the other side roller 3.2 can smoothly fit the tunnel contour until the force is balanced. A pressure sensor 3.3.9 is installed at the upper end of the guide post 3.3.7. When the supporting elastic element 3.3.8 is compressed along the guide sleeve, it will transmit the pressure value to the multi-element lining information detection and analysis system in real time. When the pressure value exceeds the maximum set value, an alarm will be triggered and the system will stop to avoid damage to the radar antenna 3.1. When the pressure value is lower than the minimum set value, an alarm will be triggered. At this time, the radar antenna 3.1 is not in contact with the tunnel contour and the detection fails. In this embodiment, the floating drive component 3.3.5 is a floating hydraulic cylinder, and the supporting elastic component 3.3.8 is a supporting spring.

[0076] Furthermore, the buffer mechanism 3.4 includes a mounting base 3.4.1, a rotating shaft 3.4.2, and buffer elastic elements 3.4.3. The mounting base 3.4.1 is hinged to the adjusting plate 3.3.3 via the rotating shaft 3.4.2. Multiple evenly distributed buffer elastic elements 3.4.3 are provided between the mounting base 3.4.1 and the adjusting plate 3.3.3. When the radar monitoring module 3 continuously advances along the tunnel axis and synchronously detects, the four buffer elastic elements 3.4.3 and the hinge shaft on the buffer mechanism 3.4 can drive the radar detection device to rotate slightly and provide buffering to avoid damage to the radar detection device due to impact. In this embodiment, the buffer elastic element 3.4.3 is a buffer spring.

[0077] In addition, see Figure 6The vehicle chassis 1 is equipped with a vertical plate 1.1, which is located at the front end of the lining multi-functional inspection arm 2 along the tunnel travel direction. The tunnel multi-layer lining quality information inspection vehicle also includes a lidar 5, an inertial navigation system, and multiple sets of linear array cameras 6 mounted on the vertical plate 1.1. The lidar 5 is located on the central axis of the vehicle chassis 1 and is electrically connected to the information detection and analysis module. The lidar 5 is used to scan the tunnel to obtain point cloud data of the tunnel's full cross-section contour. The multiple sets of linear array cameras 6 are arranged in a vertical ring and are symmetrically arranged with the lidar 5 as the center. The linear cameras are electrically connected to the information detection and analysis module and are used to acquire high-definition appearance images of the tunnel's full cross-section. The inertial navigation system is electrically connected to the information detection and analysis module and is used to acquire preliminary position information of the tunnel contour. The lidar 5 can perform a 360° circular scan, transmitting the entire tunnel cross-section contour data to the multi-dimensional lining information detection and analysis system. Multiple linear array cameras 6 are arranged circumferentially along the vehicle frame, enabling real-time acquisition of high-definition surface images of the entire tunnel cross-section during inspection. These images are then transmitted to the multi-dimensional lining information detection and analysis system for analysis of lining surface defects. The inertial navigation system enables real-time and precise positioning of the inspection vehicle, which, in conjunction with the lidar 5, allows for precise location of lining defects. The multi-dimensional lining information detection and analysis system can display more than 10 types of lining information in real time, including lining quality, lining defects, concrete strength and quality, lining surface defects, and anchor bolt defects. It can also copy and transmit this information to a designated platform.

[0078] In this embodiment, the vehicle chassis 1 is also provided with a cover 1.2. The electrical control system, hydraulic system and multi-layer lining information detection and analysis system are located inside the cover 1.2 and are protected by the cover 1.2.

[0079] In this embodiment, the information detection and analysis module adopts a multi-level architecture of "hardware layer - data layer - analysis layer - application layer". Each layer works together to realize the whole process management from information collection to decision output. It has the functions of identification, detection, processing and transmission. Its processing results can guide the optimization of lining quality detection strategy, which can effectively improve detection efficiency, reduce safety risks and save maintenance costs.

[0080] Example 2

[0081] See also Figure 7 and Figure 8 This embodiment provides a method for detecting tunnel lining information. The method uses a multi-element tunnel lining quality information detection vehicle as described above to detect lining defects. The lining defect information includes the location and classification information of surface defects as well as the location and classification information of internal defects. The method includes the following steps:

[0082] The inertial navigation system (INS), lidar 5, and linear array camera 6 are activated. The vehicle, designed for detecting the quality of tunnel lining, is driven through the tunnel to be inspected, acquiring a tunnel contour model and a high-resolution image of the entire tunnel cross-section. Specifically, during vehicle movement, the INS measures the linear acceleration and angular velocity of the equipment in real time, and, combined with the initial position and attitude, calculates the displacement and rotation changes through double integration to obtain preliminary position information of the tunnel contour. LiDAR 5 uses point cloud feature extraction to correct motion distortion during INS positioning and to model the tunnel contour, thereby achieving accurate positioning and modeling of the tunnel contour by integrating dynamic motion calculation with static environmental perception. During vehicle movement, the linear array camera 6 collects tunnel contour information line by line based on the relative motion along the tunnel contour, and uses a stacking method to construct a two-dimensional image to obtain a high-resolution image of the entire tunnel cross-section.

[0083] The information detection and analysis module is used to identify the lining defects of the tunnel under inspection in the full-section high-definition appearance image of the tunnel, and obtain the location information of the lining defects in the tunnel outline model. Then, the defects are classified and graded to obtain the classification and grading information of the lining defects. In this embodiment, the full-section high-definition appearance image of the tunnel needs to be filtered to eliminate image noise and enhance defect features before being input into the information detection and analysis module for identification.

[0084] The information detection and analysis module plans several internal quality inspection channels based on the classification and grading information of the lining surface defects. Among them, at least one internal quality inspection channel is planned for the arch crown area, at least one internal quality inspection channel is planned for each of the left and right arch waist areas, and at least two internal quality inspection channels are planned for the invert arch area.

[0085] Based on the classification information of surface defects in the lining, select an internal quality inspection channel as the internal quality inspection channel; by comparing the number of serious defects on the route, accurately recommend priority inspection channels to improve inspection efficiency.

[0086] The vehicle for detecting the quality of multi-layer lining in tunnels passes through the tunnel to be inspected. At the same time, the radar monitoring module 3 on the chassis 1 and the radar monitoring module 3 on the multi-functional lining inspection arm 2 detect the internal quality of the tunnel to be inspected and obtain the ground radar signal spectrum of the tunnel to be inspected. The ground radar signal spectrum is obtained by processing the electromagnetic wave reflection signal after noise suppression, gain adjustment and background removal.

[0087] An information detection and analysis module is used to identify internal defects in the tunnel lining of the tunnel under inspection in the ground-penetrating radar signal spectrum, and to obtain the location information of the internal defects in the tunnel outline model. Then, classification and grading information of the internal defects are obtained. Specifically, based on the differences in the propagation characteristics of electromagnetic waves in different media, the internal defects are automatically classified and graded according to the differences in the characteristics of reflected signals. For example, voids, cavities, and non-compact defects are classified by using the extremely high, relatively high, and relatively low amplitude and disorder of the reflected signals; the area of ​​the defect is determined by the lateral extension length of the reflected signal (the continuous range of reflection anomalies in the radar profile) and the longitudinal thickness (calculated by the electromagnetic wave propagation time: thickness = (propagation time × electromagnetic wave propagation speed in the medium) / 2), and automatic grading is performed according to the size of the defect.

[0088] In this embodiment, the information detection and analysis module includes an information detection and analysis model, and the method for constructing the information detection and analysis model specifically includes:

[0089] ① Construct a database of apparent defects in lining and a database of internal defects in lining; specifically:

[0090] The lining appearance defect database is obtained by classifying and grading the defect images in the initial appearance defect database according to the preset classification and grading logic of lining appearance defects. The initial appearance defect database includes N defect images from different scenes and working conditions; in this embodiment, N is 20000, and the images of the same type comprehensively cover various scenes (such as strong direct light, weak light and shadow, etc.) and various working conditions (dust coverage, water stains, etc.). The rich image data provides sufficient samples for the model to learn appearance defect features.

[0091] The initial internal defect database is obtained by classifying and classifying the ground-penetrating radar signal spectra of the lining defects according to the preset classification and grading logic. The initial internal defect database includes M ground-penetrating radar signal spectra. In this embodiment, M is 50000.

[0092] ② The initial model was trained using the database of apparent defects in lining and the database of internal defects in lining to obtain a lining defect identification model;

[0093] ③ Based on the lining defect identification model, a big data comparison mechanism is introduced to obtain an information detection and analysis model: The system compares with the pre-stored big data of lining surface defects and combines a self-learning mechanism to classify lining surface defects based on image grayscale differences. At the same time, defect grading is achieved through pixel-to-physical size conversion; or the internal defects of the lining are classified based on the vibration characteristics of electromagnetic wave reflection signals. At the same time, the area of ​​internal defects of the lining is calculated through the lateral extension length and longitudinal thickness of electromagnetic wave reflection signals, and defect grading is performed based on the area of ​​internal defects of the lining.

[0094] Specifically:

[0095] High-definition visual images of the entire tunnel section or ground-penetrating radar signal spectrum maps are used as inputs to the lining defect identification model;

[0096] The apparent defects or internal defects of the lining output by the lining defect identification model are used as features to be detected.

[0097] Calculate the cosine similarity between the feature to be detected and the corresponding defect feature vectors of each type in the lining appearance defect database or the lining internal defect database.

[0098] The feature to be detected is assigned to the category corresponding to the highest cosine similarity, and then classified according to the corresponding pre-screening and grading logic to obtain an information detection and analysis model that can output the classification and grading information of the feature to be detected.

[0099] In this embodiment, a self-learning mechanism is introduced based on the information detection and analysis model. Specifically, the self-learning mechanism is activated to update the information detection and analysis model every time a preset distance of tunnel detection is completed.

[0100] When a tunnel detection operation is completed at a preset distance, the features to be detected that have a cosine similarity lower than a preset similarity threshold are labeled and corrected using a semi-supervised labeling method; in this embodiment, the preset similarity threshold is 85%;

[0101] The labeled features to be detected are added to the corresponding lining surface defect database and lining internal defect database. The lining defect recognition model is then retrained based on the updated lining surface defect database and lining internal defect database to obtain the updated information detection and analysis model.

[0102] In this embodiment, the classification and grading of lining surface defects and internal defects can be manually modified. The system will then perform self-learning on the manually modified data to optimize the defect classification and grading model parameters, providing a more accurate basis for subsequent detection processes.

[0103] In this embodiment, the defect images in the lining appearance defect database are classified and graded according to a preset classification and grading logic for lining appearance defects. The specific classification and grading logic for lining appearance defects is as follows:

[0104] When the grayscale gradient of an image changes abruptly and the shape is continuous along the length direction, the apparent defect of the lining is a crack defect. Specifically, the following conditions are met: the grayscale gradient value of each defect image data is calculated using the Sobel operator. If the grayscale gradient value along the defect extension direction is greater than or equal to a set gradient threshold and the number of consecutive pixels in the defect is greater than or equal to a set pixel threshold, or the ratio of the length to the maximum width of the defect is greater than or equal to a set ratio threshold and the maximum number of pixels in the width direction is less than or equal to a set width threshold, then the defect image is a crack defect. Crack defects are divided into three levels: minor, moderate, and severe, based on the length of the crack defect. In this embodiment, the gradient threshold is set to 20, the pixel threshold is set to 30, the ratio threshold is set to 10:1, and the width threshold is set to 5.

[0105] When the grayscale fluctuation is large, fluctuates drastically, and the grayscale variance increases significantly, the lining surface defect is classified as a pitted defect. Specifically, it meets the following criteria: the local grayscale variance of the defect image is calculated using a sliding window. If the average grayscale variance of the defect area within the sliding window is ≥B and the grayscale value change range between two adjacent pixels is ≥C within a preset pixel statistics window, then the defect image is a pitted defect. The crack defect is then classified into three levels: slight, moderate, and severe, based on the area of ​​the pitted defect. Wherein, the grayscale value change range is the number of times the grayscale difference between two adjacent pixels is ≥30. In this embodiment, the sliding window size is 3×3 pixels, the pixel statistics window size is 10×10 pixels, and C is a grayscale value of 30.

[0106] When the grayscale value is abnormal and there are reflective areas or excessively blurred grayscale at the boundaries, the apparent defect of the lining is a water leakage defect, specifically satisfying the following conditions: Calculate the average grayscale value of the defect area. If the average grayscale value of the defect area is ≤D, or the proportion of edge pixels of the defect area to the total pixels of the defect area is ≤F, or there is a reflective sub-region with a grayscale value ≥G and the area of ​​the reflective sub-region is ≥H, then the defect image is a water leakage defect. Based on the area of ​​the water leakage defect, the crack defect is divided into three levels: minor, moderate, and severe. In this embodiment, D is a grayscale value of 30, F is 10%, G is a grayscale value of 220, and H is 50 square pixels.

[0107] In this embodiment, the ground-penetrating radar signal spectrum of the lining internal defect database is classified and graded according to preset classification and grading logic for internal lining defects. Specifically:

[0108] Calculate the maximum signal amplitude in the defect area;

[0109] judge:

[0110] When the electromagnetic wave reflection signal presents a violent vibration with strong amplitude and high energy, and the waveform of the reflection signal is complete and the boundary is clear, the internal defect of the lining is a cavity defect, specifically satisfying: if the maximum signal amplitude A in the defect area ≥ 2.8A0, and the root mean square energy E of the signal ≥ 2.3E0, and the number of recognizable complete sine wave cycles in the defect area ≥ 3 and the coefficient of variation of the adjacent wave peak spacing ≤ I, and the rising edge time of the signal transitioning from the normal area to the defect area ≤ J, then the internal defect of the lining is a cavity defect; and it is divided into three levels of minor, general, and severe according to the area of the cavity defect; where: A0 is the reference signal amplitude; E0 is the reference root mean square energy; in this embodiment, I is 10%, and J is 0.6 μs;

[0111] When the electromagnetic wave reflection signal shows a continuous vibration with medium amplitude, the reflection waveform is relatively continuous but the energy is lower than that of the cavity defect, and there is a certain fuzzy transition area at the reflection interface, the internal defect of the lining is a debonding defect, specifically satisfying: if the maximum signal amplitude A in the defect area satisfies 1.4A0 ≤ A < 2.8A0, and the root mean square energy E of the signal satisfies 1.1E0 ≤ E < 2.3E0, and the number of recognizable complete sine wave cycles in the defect area ≥ 2 and the coefficient of variation of the adjacent wave peak spacing ≤ K, and the rising edge time of the signal transitioning from the normal area to the defect area satisfies T1 < T < T2 and the time width corresponding to the transition area ≥ T3, then the internal defect of the lining is a debonding defect; and it is divided into three levels of minor, general, and severe according to the area of the debonding defect; in this embodiment, K is 20%, T1 is 0.6 μs, T2 is 1.6 μs, and T3 is 1 μs;

[0112] When the electromagnetic wave reflection signal presents a chaotic vibration with low amplitude and low energy, the waveform has no obvious pattern and the reflection interface is fuzzy, and the overall signal energy is slightly higher than that of the normal dense area but there is no significant mutation, the internal defect of the lining is an incompact defect, specifically satisfying: if the maximum signal amplitude A in the defect area satisfies 1.0A0 ≤ A < 1.4A0, and the root mean square energy E of the signal satisfies 1.05E0 ≤ E < 1.1E0, and there is no complete sine wave cycle in the defect area and the coefficient of variation of the amplitude of the signal wave peak / valley ≥ L, and the attenuation time of the signal ≥ M, then the internal defect of the lining is an incompact defect; and it is divided into three levels of minor, general, and severe according to the area of the incompact defect. In this embodiment, L is 25%, and M is 2.2 μs.

[0113] In this embodiment, the tunnel lining information detection method further includes: using a multi-functional concrete non-destructive testing instrument and an anchor cable non-destructive testing instrument to perform non-destructive testing on the tunnel to be inspected to obtain defect information of concrete and anchors, and selecting non-destructive testing on the multi-dimensional lining information detection and analysis system; specifically: personnel enter the suspended platform, select a detection channel on the suspended platform operation display screen, and the system realizes automatic path planning of the suspended platform to reach the area to be tested based on the positioning information; the position of the detection arm is accurately positioned by manual adjustment, and after positioning is completed, the operator uses the multi-functional concrete non-destructive testing instrument and the anchor cable non-destructive testing instrument to complete the detection of concrete strength and quality and anchor defects. The detection data is transmitted to the multi-dimensional lining information detection and analysis system in real time to realize the display of concrete and anchor information (concrete strength and quality, structural dimensions, anchor length, number of anchors, etc.), defect identification (crack location, crack depth, grouting density, anchor defects, etc.) and positioning. Specific implementation method: Utilizing the correlation between the acoustic parameters (sound velocity, sound duration, amplitude, frequency, waveform, etc.) of ultrasonic waves propagating in concrete and anchor bolts and internal defects, different types and severity of defects are identified and classified; specifically including:

[0114] ① Randomly determine the test area of ​​the tunnel lining surface to be inspected, and determine the specific coordinate positions of several test points on the test area based on the location information of the lining defects.

[0115] ② The information detection and analysis module plans the automatic movement path of the main body of the suspended basket on the tunnel multi-layer lining quality information detection vehicle according to the specific coordinate location of the point to be measured, so as to move the suspended basket to the area to be measured;

[0116] ③ Use a four-stage boom to move the main body of the suspended platform to the test point positions in the test area;

[0117] ④ Use a multi-functional concrete non-destructive testing instrument and an anchor cable non-destructive testing instrument to perform non-destructive testing on the concrete and anchor at the test point to obtain defect information of the concrete and anchor.

[0118] In this embodiment, the self-learning process of the information detection and analysis model is as follows:

[0119] ① Offline pre-training: The initial model was trained using the constructed apparent defect library and internal defect library. Simultaneously, to further expand the diversity of the dataset, CycleGAN was used to generate virtual samples. CycleGAN can achieve image style transfer between two domains without paired data. For example, for images of apparent defects in tunnel lining, normal lining images can be converted into virtual images with different defect styles, such as generating virtual images of cracks of different shapes and sizes, and various degrees of water leakage. These virtual samples, along with real-labeled samples, were used for model training, enabling the model to learn a wider range of defect features.

[0120] ② Online Incremental Learning: The system automatically triggers a model update mechanism every 500m of tunnel inspection. For newly discovered defect samples (with a similarity value of less than 85% to existing samples), a semi-supervised annotation method is used. First, the current model is used to make preliminary predictions for the new samples, generating pseudo-labels. Then, the pseudo-labels are manually reviewed and corrected. For samples that are difficult to judge, expert knowledge and annotation tools are used to accurately label them. The manually modified labeled data is added to the training set, and the model is retrained so that the model can learn the characteristics of newly emerging defects in a timely manner. For example, when a new non-compact defect is detected in the lining, after the model makes a preliminary prediction of its type, it is manually checked against the GPR (Ground Ground Radar) signal spectrum. If the prediction is inaccurate, the label is corrected, and the corrected sample is then used for model updates.

[0121] Information detection and analysis model parameter optimization methods:

[0122] ① Dual-loop optimization mechanism – Inner loop: Precision-Recall curves are calculated based on the confusion matrix to evaluate the model's classification performance at different thresholds. The CNN convolutional kernel size is dynamically adjusted according to the curve's changes. For example, if the model's recall rate for a certain type of defect (such as small cracks) is found to be low, the convolutional kernel size is appropriately reduced to enhance the model's ability to extract detailed features; if the model's accuracy for large-area defects is not high, the convolutional kernel size is increased to improve its ability to grasp overall features. By continuously adjusting the convolutional kernel size, the model achieves a better balance in detecting different defect types.

[0123] ② Dual-loop optimization mechanism – Outer loop: Reinforcement learning is used to optimize the AHP (Analytic Hierarchy Process) weight coefficients. Detection accuracy and maintenance cost are weighted as the reward function, encouraging the model to minimize unnecessary maintenance costs caused by false detections while maintaining high detection accuracy. During optimization, the weights of the reward function are reasonably set based on human experience in assessing the severity of different defects and maintenance costs. For example, defects that seriously affect tunnel safety (such as large cavities inside the lining) are given higher weights for detection accuracy; for minor defects (such as fine surface pitting), the weight of maintenance cost is appropriately balanced. The optimal AHP weight coefficients are continuously explored through reinforcement learning algorithms, making the model's decisions more aligned with actual engineering needs.

[0124] ③ Optimization Cycle: A full-parameter optimization is performed every 5km of tunnel to ensure the model's F1-score is ≥ 0.92. During the full-parameter optimization process, all parameters of the model are readjusted, including the weights and biases of the convolutional neural network, as well as various hyperparameters. Simultaneously, the direction and scope of parameter adjustments are determined based on manual analysis of previous detection data and model performance. For example, if the model is found to be ineffective in detecting internal defects, targeted adjustments are made to the model parameters during the full-parameter optimization to comprehensively improve model performance and ensure it maintains a high detection capability in complex tunnel environments.

[0125] Since the tunnel lining information detection method includes the tunnel multi-element lining quality information detection vehicle as described above, it possesses all the beneficial effects of the aforementioned tunnel multi-element lining quality information detection vehicle, which will not be elaborated upon here.

[0126] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A method for detecting tunnel lining information, comprising using a multi-element tunnel lining quality information detection vehicle to obtain lining defect information, the lining defect information including location information and classification and grading information of surface defects and location information and classification and grading information of internal defects, characterized in that, The tunnel lining quality information detection vehicle includes a car chassis (1), a multi-functional lining detection arm (2), a radar monitoring module (3), an invert arch telescopic arm (4), an information detection and analysis module, a lidar (5) mounted on a vertical plate (1.1), an inertial navigation system, and multiple linear array cameras (6). The method for detecting tunnel lining information includes the following steps: Turn on the inertial navigation, lidar (5) and linear array camera (6), drive the tunnel multi-layer lining quality information detection vehicle through the tunnel to be inspected, and obtain the tunnel outline model and the high-definition appearance image of the entire tunnel section respectively; The information detection and analysis module is used to identify the lining surface defects of the tunnel under inspection in the full-section high-definition appearance image of the tunnel, and obtain the location information of the lining surface defects in the tunnel outline model. Then, the defects are classified and graded to obtain the classification and grading information of the lining surface defects. The information detection and analysis module plans several internal quality inspection channels based on the classification and grading information of the lining surface defects. Among them, at least one internal quality inspection channel is planned for the arch crown area, at least one internal quality inspection channel is planned for each of the left and right arch waist areas, and at least two internal quality inspection channels are planned for the invert arch area. Select an internal quality inspection channel as the internal quality inspection channel based on the classification information of the lining surface defects. The vehicle is driven through the tunnel to be inspected. At the same time, the radar monitoring module (3) on the chassis (1) and the radar monitoring module (3) on the lining multi-functional inspection arm (2) are used to detect the internal quality of the tunnel to be inspected and obtain the geological radar signal spectrum of the tunnel to be inspected. The information detection and analysis module is used to identify the internal defects of the tunnel lining in the ground-penetrating radar signal spectrum, and obtain the location information of the internal defects of the lining in the tunnel outline model. Then, the internal defects of the lining are classified and graded to obtain the classification and grading information of the internal defects of the lining. The information detection and analysis module includes an information detection and analysis model, and the method for constructing the information detection and analysis model specifically includes: ① Construct a database of apparent defects in lining and a database of internal defects in lining; specifically: The lining appearance defect database is obtained by classifying and grading the defect images in the initial appearance defect database according to the preset classification and grading logic of the lining appearance defects; wherein: the initial appearance defect database includes N defect images under different scenarios and working conditions. The initial internal defect database is obtained by classifying and classifying the ground radar signal spectra of the lining defects according to the preset classification and grading logic. The initial internal defect database includes M ground radar signal spectra. ② The initial model was trained using the database of apparent defects in lining and the database of internal defects in lining to obtain a lining defect identification model; ③ Based on the lining defect identification model, a big data comparison mechanism is introduced to obtain an information detection and analysis model, specifically: High-definition visual images of the entire tunnel section or ground-penetrating radar signal spectrum maps are used as inputs to the lining defect identification model; The apparent defects or internal defects of the lining output by the lining defect identification model are used as features to be detected. Calculate the cosine similarity between the feature to be detected and the corresponding defect feature vectors of each type in the lining appearance defect database or the lining internal defect database. The feature to be detected is assigned to the category corresponding to the highest cosine similarity, and then classified according to the corresponding pre-screening and grading logic to obtain an information detection and analysis model that can output the classification and grading information of the feature to be detected.

2. The tunnel lining information detection method as described in claim 1, characterized in that, A self-learning mechanism is introduced based on the information detection and analysis model. This mechanism is activated to update the information detection and analysis model after each tunnel detection operation at a preset distance. Specifically: When a tunnel detection operation is completed at a preset distance, semi-supervised annotation is used to annotate and correct the features to be detected for which all cosine similarities are lower than the preset similarity threshold. The labeled features to be detected are added to the corresponding lining surface defect database and lining internal defect database. The lining defect recognition model is then retrained based on the updated lining surface defect database and lining internal defect database to obtain the updated information detection and analysis model.

3. The tunnel lining information detection method as described in claim 2, characterized in that, The defect images in the lining appearance defect database are classified and graded according to the preset classification and grading logic of lining appearance defects. The specific classification and grading logic of lining appearance defects is as follows: The Sobel operator is used to calculate the gray-level gradient value of each defect image data. If the gray-level gradient value along the defect extension direction is greater than or equal to the set gradient threshold and the number of consecutive pixels in the defect is greater than or equal to the set pixel threshold, or the ratio of the length to the maximum width of the defect is greater than or equal to the set ratio threshold and the maximum number of pixels in the width direction is less than or equal to the set width threshold, then the defect image is a crack defect. Crack defects are classified into three levels: minor, moderate, and severe according to the length of the crack defect. The local gray-level variance of the defect image is calculated using a sliding window. If the average gray-level variance of the defect area within the sliding window is ≥B and the gray-level value change of two adjacent pixels is ≥C within the preset pixel statistics window, then the defect image is a pitted defect. Based on the area of ​​the pitted defect, the crack defect is divided into three levels: slight, moderate, and severe. Calculate the average gray value of the defect area. If the average gray value of the defect area is ≤D, or the proportion of edge pixels of the defect area to the total pixels of the defect area is ≤F, or there is a reflective sub-region with a gray value ≥G and the area of ​​the reflective sub-region is ≥H, then the defect image is a water leakage defect. Based on the area of ​​the water leakage defect, the crack defect is divided into three levels: minor, general, and severe.

4. The tunnel lining information detection method as described in claim 3, characterized in that, The ground-penetrating radar signal spectrum of the lining internal defect database is classified and graded according to the preset classification and grading logic of internal lining defects. Specifically: Calculate the maximum signal amplitude in the defect area; Judgment: If the maximum signal amplitude A of the defect area ≥ 2.8A0, and the root mean square energy E of the signal ≥ 2.3E0, and the number of recognizable complete sine wave cycles in the defect area ≥ 3 and the coefficient of variation of the adjacent wave crest spacing ≤ I, and the rising edge time of the signal transitioning from the normal area to the defect area ≤ J, then the internal defect of the lining is a cavity defect; and it is divided into three levels of minor, general, and severe according to the area of the cavity defect; where: A0 is the reference signal amplitude; E0 is the reference root mean square energy; If the maximum signal amplitude A of the defect area satisfies 1.4A0 ≤ A < 2.8A0, and the root mean square energy E of the signal satisfies 1.1E0 ≤ E < 2.3E0, and the number of recognizable complete sine wave cycles in the defect area ≥ 2 and the coefficient of variation of the adjacent wave crest spacing ≤ K, and the rising edge time of the signal transitioning from the normal area to the defect area satisfies T1 < T < T2 and the time width corresponding to the transition area ≥ T3, then the internal defect of the lining is a void defect; and it is divided into three levels of minor, general, and severe according to the area of the void defect; If the maximum signal amplitude A of the defect area satisfies 1.0A0 ≤ A < 1.4A0, and the root mean square energy E of the signal satisfies 1.05E0 ≤ E < 1.1E0, and there is no complete sine wave cycle in the defect area and the coefficient of variation of the amplitude of the signal wave crest / trough ≥ L, and the attenuation time of the signal ≥ M, then the internal defect of the lining is an incompact defect; and it is divided into three levels of minor, general, and severe according to the area of the incompact defect.

5. The tunnel lining information detection method as described in claim 4, characterized in that, The tunnel lining information detection method described above further includes: using a concrete multi-functional non-destructive detector and a bolt and cable non-destructive detector to perform non-destructive detection on the tunnel to be inspected to obtain the defect information of the concrete and bolts.

6. The tunnel lining information detection method as described in claim 5, characterized in that, Using a concrete multi-functional non-destructive detector and a bolt and cable non-destructive detector to perform non-destructive detection on the tunnel to be inspected to obtain the defect information of the concrete and bolts, specifically including: ①. Randomly determine the待测区间 on the surface of the tunnel lining to be inspected, and determine the specific coordinate positions of several待测点 according to the positioning information of the lining disease information in the待测区间; ②. The information detection and analysis module plans the automatic movement path of the hanging basket main body (2.10) on the tunnel multi-lining quality information detection vehicle according to the specific coordinate positions of the待测点 to move the hanging basket to the待测区间; ③. Use the four-stage arm section (2.7) to move the hanging basket main body (2.10) to the position of the待测点 in the待测区间 respectively; ④. Use the concrete multi-functional non-destructive detector and the bolt and cable non-destructive detector to perform non-destructive detection on the concrete and bolts at the position of the待测点 to obtain the defect information of the concrete and bolts.

7. A tunnel lining quality information detection vehicle, used to detect lining defect information using the tunnel lining information detection method as described in any one of claims 1 to 6, characterized in that, The vehicle chassis (1) is capable of traveling along the tunnel floor; the lining multi-functional detection arm (2) is mounted on the vehicle chassis (1), and the end of the lining multi-functional detection arm (2) away from the vehicle chassis (1) is provided with a radar monitoring module (3) capable of traveling along the tunnel wall; the arch telescopic arm (4) is mounted on the bottom of the vehicle chassis (1), and the arch telescopic arm (4) is provided with at least two radar monitoring modules (3) capable of traveling along the tunnel floor; the radar monitoring module (3) is used to detect tunnel multi-layer lining information; the information detection and analysis module is electrically connected to the radar monitoring module (3), and the information detection and analysis module can analyze the lining defect information based on the detection data of the radar monitoring module (3).

8. The tunnel multi-layer lining quality information detection vehicle as described in claim 7, characterized in that, The lining multifunctional inspection arm (2) includes a primary rotary seat (2.1), a primary boom section (2.2), a secondary boom section (2.3), a tertiary boom section (2.4), a secondary rotary seat (2.5), a tertiary rotary seat (2.6), and a quaternary boom section (2.7). The primary rotary seat (2.1) is mounted on the vehicle chassis (1) and is capable of rotating along a horizontal plane. The primary boom section (2.2), the secondary boom section (2.3), and the tertiary boom section (2.4) are connected sequentially, and the primary boom section (2.2) is connected to the primary rotary seat (2.1). The primary boom section (2.2) is capable of rotating in both horizontal and vertical positions. The secondary boom (2.3) can rotate relative to the primary boom (2.2), and the tertiary boom (2.4) can extend and retract along the axial direction of the secondary boom (2.3). The secondary slewing seat (2.5) is connected to the tertiary boom (2.4), and the secondary slewing seat (2.5) can rotate along the horizontal plane. The tertiary slewing seat (2.6) is disposed on the secondary slewing seat (2.5), and the tertiary slewing seat (2.6) can rotate along the vertical plane. The first end of the quaternary boom (2.7) is hinged to the tertiary slewing seat (2.6), and the second end is connected to the radar monitoring module. The quaternary boom (2.7) can perform multi-stage extension and retraction.

9. The tunnel multi-layer lining quality information inspection vehicle as described in claim 8, characterized in that, The lining multi-functional testing arm (2) also includes a suspended basket arm (2.8), a four-stage rotating seat (2.9), a suspended basket body (2.10), a concrete multi-functional non-destructive testing instrument, and an anchor cable non-destructive testing instrument. The suspended basket arm (2.8) is hinged to the three-stage boom section (2.4). The four-stage rotating seat (2.9) is located at the end of the suspended basket arm (2.8) away from the three-stage boom section (2.4), and the four-stage rotating seat (2.9) can rotate along the horizontal plane. The suspended basket body (2.10) is located on the four-stage rotating seat (2.9). The concrete multi-functional non-destructive testing instrument and the anchor cable non-destructive testing instrument are both located on the suspended basket body (2.10). The concrete multi-functional non-destructive testing instrument is used to test the strength and quality of concrete, and the anchor cable non-destructive testing instrument is used to test anchor defects.

10. The tunnel multi-layer lining quality information detection vehicle as described in claim 9, characterized in that, The radar monitoring module (3) includes a radar antenna (3.1), a roller (3.2), a leveling mechanism (3.3), and a buffer mechanism (3.4). The radar antenna (3.1) is mounted on the leveling mechanism (3.3). The radar antenna (3.1) rolls and contacts the tunnel wall through the roller (3.2). The radar antenna (3.1) is used to detect multi-element lining information. The leveling mechanism (3.3) is mounted on the buffer mechanism (3.4). The buffer mechanism (3.4) is mounted on the fourth-stage boom (2.7) or the invert arch telescopic boom (4). The buffer mechanism (3.4) buffers the leveling mechanism (3.3).

11. The tunnel multi-layer lining quality information inspection vehicle as described in claim 10, characterized in that, The leveling mechanism (3.3) includes a support structure (3.3.1), a mounting plate (3.3.2), an adjusting plate (3.3.3), a slider (3.3.4), and a floating drive component (3.3.5). The support structure (3.3.1) is disposed between the mounting plate (3.3.2) and the adjusting plate (3.3.3). The support structure (3.3.1) includes a connecting rod (A1) and two cross linkage groups (A2). The two cross linkage groups (A2) are spaced apart. Each cross linkage group (A2) includes a first link and a second link that are hinged in a cross configuration. The two ends of the first link are connected to the first link of the other cross linkage group (A2) through a connecting rod (A1), and the two ends of the second link are connected to the second link of the other cross linkage group (A2) through a connecting rod (A1). The mounting plate (3.3.2) has slide rails (3.3.6) on one side facing the support structure (3.3.1) and on both sides of the adjusting plate (3.3.3); a slider (3.3.4) is slidably disposed within each slide rail (3.3.6). The first end of the first connecting rod is hinged to the slider (3.3.4) on the mounting plate (3.3.2), and the second end of the first connecting rod is hinged to the adjusting plate (3.3.3). The first end of the connecting rod is hinged to the slider (3.3.4) on the side of the adjusting plate (3.3.3) facing the support structure (3.3.1), and the second end of the second connecting rod is hinged to the mounting plate (3.3.2); the first end of the floating drive (3.3.5) is hinged to the slider (3.3.4) on the side of the adjusting plate (3.3.3) away from the support structure (3.3.1), and the second end is hinged to a connecting rod (A1).

12. The tunnel multi-layer lining quality information inspection vehicle as described in claim 11, characterized in that, The leveling mechanism (3.3) further includes a guide post (3.3.7), a supporting elastic element (3.3.8), and a pressure sensor (3.3.9). One end of the guide post (3.3.7) is connected to the mounting plate (3.3.2) via the pressure sensor (3.3.9), and the other end slides through the leveling plate (3.3.3). The supporting elastic element (3.3.8) is sleeved on the guide post (3.3.7), and both ends of the supporting elastic element (3.3.8) are connected to the mounting plate (3.3.2) and the leveling plate (3.3.3), respectively. The pressure sensor (3.3.9) is electrically connected to the information detection and analysis module.

13. The tunnel multi-layer lining quality information detection vehicle as described in claim 12, characterized in that, The buffer mechanism (3.4) includes a mounting base (3.4.1), a rotating shaft (3.4.2), and buffer elastic elements (3.4.3). The mounting base (3.4.1) is hinged to the adjusting plate (3.3.3) via the rotating shaft (3.4.2). A plurality of buffer elastic elements (3.4.3) are evenly distributed between the mounting base (3.4.1) and the adjusting plate (3.3.3).

14. The tunnel multi-layer lining quality information inspection vehicle as described in any one of claims 7 to 13, characterized in that, A vertical plate (1.1) is provided on the vehicle chassis (1), and the vertical plate (1.1) is set at the front end of the lining multi-functional detection arm (2) along the tunnel travel direction; the lidar (5) is located on the central axis of the vehicle chassis (1), and the lidar (5) is electrically connected to the information detection and analysis module. The lidar (5) is used to scan the tunnel to obtain the tunnel full-section contour point cloud data; multiple sets of line array cameras (6) are arranged in a ring along the vertical plate (1.1), and the multiple sets of line array cameras (6) are symmetrically arranged with the lidar (5) as the center. The line array cameras (6) are electrically connected to the information detection and analysis module, and the line array cameras (6) are used to acquire high-definition appearance images of the tunnel full-section; the inertial navigation system is electrically connected to the information detection and analysis module, and the inertial navigation system is used to acquire preliminary position information of the tunnel contour.

Images