Tunneling boulder cooperative detection method and computer equipment

Abstract

The application discloses a tunnel boulder cooperative detection method and computer equipment, wherein the method comprises the following steps: S1, using broadband vibration noise as a natural sound source in the HSP near-field detection stage, and collecting reflected sound wave signals by a signal collection module; S2, performing wavelet packet decomposition on the reflected sound wave signals; S3, performing a travel time tomography algorithm processing, generating an HSP three-dimensional wave velocity distribution model in front of a shield tunneling machine, and marking a potential boulder area; S4, according to the potential boulder area position marked by the HSP detection, positioning and emitting coded sound wave signals in the specified direction; S5, receiving the reflected coded sound wave signals, and calculating a shear wave splitting angle deviation and an amplitude attenuation coefficient; and S6, generating a boulder space probability distribution map in front of the shield tunneling machine based on a Bayesian probability fusion algorithm. The application provides a feasible way for realizing accurate advanced detection of boulders in the shield tunneling process.

Description

Technical Field

[0001] This invention relates to the field of tunnel boulder detection technology, and more specifically to a collaborative method and computer equipment for detecting tunnel boulders. Background Technology

[0002] In recent years, with the continuous development of infrastructure construction, underground projects spanning cities and towns have gradually increased, especially the demand for long-distance tunnel projects such as water diversion tunnels, intercity rail transit, and energy transmission pipelines. Compared with the traditional drill and blast method, the shield tunneling method, due to its advantages such as strong construction continuity, minimal disturbance to the strata, and adaptability to complex geological conditions, occupies an important position in the construction of long-distance underground projects and has become the core construction technology for large-diameter tunnel excavation. However, during shield tunneling, the complex hydrogeological environment remains an important factor affecting construction safety and efficiency. Considering that the strata of the hilly and mountainous areas of southeastern coastal China are widely distributed with extremely strong granite boulders, the presence of these boulders may cause problems such as cutterhead tool damage, obstruction of advance, deviation of tunneling posture, and instability of the tunnel face. In severe cases, there is even a risk of sudden surge at the working face and blockage of the slurry system. Therefore, accurate detection of the location of boulders before construction is crucial to ensuring construction safety and progress.

[0003] Existing methods for detecting isolated boulders, such as ground-penetrating radar, borehole exploration, multidimensional micro-motion, and cross-hole CT, have limited effectiveness in practice due to problems such as low efficiency, high cost, poor timeliness, and susceptibility to working environment constraints. Furthermore, some engineering projects require advanced tunnel detection technology, and surface boulder detection cannot accurately locate the position of deeply buried boulders. Summary of the Invention

[0004] The present invention provides a method and computer equipment for collaborative detection of isolated boulders in tunnels that effectively improves the reliability and resolution of isolated boulder detection, and can solve at least one of the above-mentioned technical problems.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for collaborative detection of isolated boulders in tunnels includes the following steps:

[0007] S1. The broadband vibration noise generated by the shield machine cutterhead cutting the rock and soil during tunnel excavation is used as the natural sound source for the HSP near-field detection stage. The reflected sound wave signal is collected by the signal acquisition module arranged behind the cutterhead.

[0008] S2. In the signal processing module, wavelet packet decomposition is performed on the acquired reflected sound wave signal to extract the characteristic reflected sound wave components of the boulder.

[0009] S3. The extracted isolated rock feature reflected sound waves are processed by the travel time tomography algorithm to generate a three-dimensional HSP wave velocity distribution model in front of the tunnel boring machine and mark the potential isolated rock areas.

[0010] S4. Based on the orientation of the potential boulder area marked by the HSP detection, control the array of dipole acoustic wave detection modules arranged in the shield ring of the tunnel boring machine to position and transmit coded acoustic wave signals in the specified direction.

[0011] S5. The dipole acoustic wave detection module array receives the reflected coded acoustic wave signal and calculates the transverse wave splitting angle deviation and amplitude attenuation coefficient.

[0012] S6. Combining the HSP three-dimensional wave velocity analysis model and the dipole acoustic wave amplitude attenuation coefficient, a probability distribution map of the isolated rock space in front of the tunnel boring machine is generated based on the Bayesian probability fusion algorithm.

[0013] Furthermore, in S1, the signal acquisition module is a triaxial accelerometer installed behind the cutterhead of the tunnel boring machine. The triaxial accelerometer has 6 sets, which are evenly distributed along the circumference of the flange of the cutterhead of the tunnel boring machine. The angle between two adjacent triaxial accelerometers and the center of the flange is 60° to ensure that the vibration signal of the entire cross section is covered.

[0014] Furthermore, in S2, the frequency band dynamics of wavelet packet decomposition should be adapted to the tunneling parameters of the tunnel boring machine during the HSP near-field detection phase.

[0015] Furthermore, in S4, the dipole acoustic wave detection module is a dipole acoustic wave transducer pre-embedded inside the shield shell of the tunnel boring machine. There are 4 groups of dipole acoustic wave transducers, which are evenly distributed along the circumference of the shield shell of the tunnel boring machine. The angle between two adjacent dipole acoustic wave transducers and the center of the shield shell of the tunnel boring machine is 90°. Each group of dipole acoustic wave transducers includes one transmitting unit and two receiving units. The transmitting unit and the two receiving units are arranged in a triangular array, and the two receiving units are respectively distributed on two right-angled sides with the transmitting unit as the center.

[0016] Furthermore, in S4, during the detection process, the dipole acoustic wave detection module controls at least two sets of orthogonally distributed dipole acoustic wave transducers to focus and scan a fan-shaped area of ​​θ+15° according to the azimuth angle θ of the potential isolated rock area marked by the HSP detection. At the same time, an LFM pulse compression mechanism is used to improve the signal-to-noise ratio of directional detection.

[0017] Furthermore, in S5, the transmission-reception timing of the multiple sets of dipole acoustic transducer arrays is synchronized with the rotation cycle of the tunnel boring machine cutterhead, in order to reduce the strong noise interference caused by the contact between the cutterhead cutter and the rock mass.

[0018] Furthermore, in S6, the Bayesian probabilistic fusion algorithm is based on a feature library of isolated rock reflection acoustic waves established by introducing a historical geological data transfer learning mechanism. This feature library contains HSP response templates and dipole acoustic wave response templates corresponding to isolated rocks under different physical and mechanical parameters. During the calculation process of the Bayesian probabilistic fusion algorithm, the convolutional neural network in this feature library is called to perform similarity matching between the real-time detection data and the feature library, and outputs the probability distribution of the specified type of isolated rock.

[0019] Furthermore, the Bayesian probabilistic fusion algorithm in S6 includes the following implementation steps:

[0020] S6.1 Setting of prior probability: The prior probability p(isolated boulder) is set according to the historical geological database to determine the probability of the initial existence of isolated boulders. When the tunnel boring machine updates the stratum type in real time during the tunneling process, the prior probability p(isolated boulder) is automatically adjusted to follow the geological changes.

[0021] S6.2 Conditional Probability Modeling:

[0022] HSP conditional probability: Based on historical data statistics, a Gaussian distribution model p(v) is established for the wave velocity anomalies in the isolated rock area. 异常 |lone rock);

[0023] Conditional probability of dipole acoustic waves: Based on the amplitude attenuation coefficient threshold β and the transverse wave splitting angle deviation Δθ, the distribution of measured data is fitted using kernel density estimation, and a joint probability model p(β, Δθ|isolated rock) is established based on the data distribution.

[0024] S6.3, Posterior Probability Calculation: The posterior probability of the existence of the isolated rock is calculated using Bayes' theorem, expressed as follows:

[0025]

[0026] In the formula, the denominator p(v) 异常 , β, Δθ) are calculated using the full probability formula, covering all possible geological conditions.

[0027] Furthermore, during the calculation process, the Bayesian probability fusion algorithm obtains the HSP wave velocity anomaly region based on the HSP three-dimensional wave velocity analysis model, and the confidence weight of the HSP wave velocity anomaly region is higher than the dipole acoustic wave amplitude attenuation coefficient.

[0028] A computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the above-described tunnel boulder cooperative detection method.

[0029] The beneficial effects of this invention are reflected in:

[0030] 1. The detection method logic of this invention is further optimized. By first performing HSP detection to lock the area where the isolated rock exists in the near field, and then selectively starting dipole acoustic wave far detection, invalid full-range scanning is avoided, and detection efficiency is improved. The HSP results are used to guide the emission direction and parameters of the dipole acoustic wave, forming a closed-loop detection logic of near-field positioning and far-field verification.

[0031] 2. This invention can save the operation time of traditional HSP artificial excitation and improve detection efficiency.

[0032] 3. This invention enhances the effectiveness of artificial intelligence by introducing transfer learning and reinforcement learning, enabling the system to have stratum adaptability and reducing the need for human intervention. Attached Figure Description

[0033] The accompanying drawings, which are provided to further illustrate this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application.

[0034] Figure 1 This is a schematic flowchart of the detection method according to an embodiment of the present invention.

[0035] Figure 2 This is a schematic diagram of the signal acquisition module layout according to an embodiment of the present invention.

[0036] Figure 3 This is a schematic diagram of the arrangement of the dipole acoustic wave detection module according to an embodiment of the present invention.

[0037] Figure 4 This is a schematic diagram of the signal processing chain according to an embodiment of the present invention.

[0038] Figure 5 This is a structural block diagram of a computer device according to an embodiment of the present invention.

[0039] The components in the attached diagram are labeled as follows: 1. Triaxial accelerometer; 2. Flange; 3. Dipole acoustic transducer; 4. Shield shell of the tunnel boring machine; 5. Transmitting unit; 6. Receiving unit. Detailed Implementation

[0040] 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 them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] It should be noted that if directional indications (such as up, down, left, right, front, back, etc.) are involved in the embodiments of this invention, these directional indications are only used to explain the relative positional relationships and movement of the components in a specific posture (as shown in the attached figures). If the specific posture changes, the directional indications will also change accordingly. Furthermore, the meaning of "and / or" throughout the text includes three parallel solutions. Taking "A and / or B" as an example, it includes solution A, solution B, or a solution where A and B are satisfied simultaneously. Additionally, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0042] See Figure 1 and Figure 4 This invention provides a method for collaborative detection of isolated boulders in tunnels, comprising the following steps:

[0043] S1. The broadband vibration noise generated by the shield machine cutterhead cutting the rock and soil during tunnel excavation is used as the natural sound source for the HSP near-field detection stage. The reflected sound wave signal is collected by the signal acquisition module arranged behind the cutterhead.

[0044] S2. In the signal processing module, the acquired reflected sound wave signal is decomposed into wavelet packets to filter out the low-frequency noise of the cutter head drive motor and extract the characteristic reflected sound wave components of the boulder in the 5-15kHz frequency band.

[0045] S3. The extracted isolated rock feature reflected sound waves are processed by the travel time tomography algorithm to generate a three-dimensional HSP wave velocity distribution model in front of the tunnel boring machine and mark the potential isolated rock areas.

[0046] S4. Based on the orientation of the potential boulder area marked by the HSP detection, control the array of dipole acoustic wave detection modules arranged in the shield ring of the tunnel boring machine to position and transmit coded acoustic wave signals in the specified direction.

[0047] S5. The dipole acoustic wave detection module array receives the reflected coded acoustic wave signal and calculates the transverse wave splitting angle deviation and amplitude attenuation coefficient.

[0048] S6. Combining the HSP three-dimensional wave velocity analysis model and the dipole acoustic wave amplitude attenuation coefficient, a probability distribution map of the isolated rock space in front of the tunnel boring machine is generated based on the Bayesian probability fusion algorithm.

[0049] See Figure 2In this embodiment, in step S1, the signal acquisition module is a triaxial accelerometer 1 installed behind the cutterhead of the tunnel boring machine. The triaxial accelerometer 1 has 6 sets, which are evenly distributed around the circumference of the flange 2 of the cutterhead of the tunnel boring machine. The angle between two adjacent triaxial accelerometers 1 and the center of the flange 2 is 60° to ensure coverage of vibration signals across the entire cross section.

[0050] In this application, the triaxial accelerometer 1, i.e., the signal acquisition module, is fixedly assembled using a magnetic-mechanical lock composite method to prevent sensor displacement due to vibration during tunneling. The acceleration signal is transmitted to the signal processing unit located at the tail of the tunnel boring machine via shielded twisted-pair cable to reduce electromagnetic interference and signal attenuation, ensuring signal integrity and stability during transmission.

[0051] In this embodiment, in step S2, the frequency band dynamics of wavelet packet decomposition should be adapted to the tunneling parameters of the tunnel boring machine during the HSP near-field detection phase.

[0052] In this application, wavelet packet decomposition can also be referred to as wavelet packet, subband tree, or optimal subband tree structuring. The concept is to represent wavelet packets using an analysis tree, that is, to analyze the detailed parts of the input signal using multiple iterative wavelet transforms. It has the following characteristics:

[0053] Adaptive frequency band division: Dynamically select the frequency band range of the decomposition node according to the signal characteristics. For example, the frequency band can be divided into a 64Hz width under the three-level decomposition at a sampling frequency of 1024Hz.

[0054] Redundancy-free decomposition: The high-frequency part is further subdivided to avoid missing detailed information, which is suitable for analyzing complex signals such as edges and textures.

[0055] Wavelet packet decomposition can effectively characterize non-stationary signals (such as earthquake and biomedical signals) and solve the problem of insufficient high-frequency details in ordinary wavelet transform; it supports energy feature extraction (such as energy entropy and energy ratio), which facilitates signal classification and analysis.

[0056] See Figure 3In this embodiment, in step S4, the dipole acoustic wave detection module is a dipole acoustic wave transducer 3 pre-embedded inside the shield shell 4 of the tunnel boring machine. There are 4 groups of dipole acoustic wave transducers 3, which are evenly distributed around the circumference of the shield shell 4 of the tunnel boring machine. The angle between two adjacent dipole acoustic wave transducers 3 and the center of the shield shell 4 of the tunnel boring machine is 90°. Each group of dipole acoustic wave transducers 3 includes a transmitting unit 5 and two receiving units 6. The transmitting unit 5 and the two receiving units 6 are arranged in a triangular array, and the two receiving units 6 are respectively distributed on two right-angled sides with the transmitting unit 5 as the center.

[0057] In the actual assembly process of the detection device of this application, the interval between the transmitting unit 5 and the receiving unit 6 is set to 0.5m.

[0058] In this embodiment, during step S4, the dipole acoustic wave detection module controls at least two orthogonally distributed dipole acoustic wave transducers 3 to focus and scan a fan-shaped area of ​​θ+15° according to the azimuth angle θ of the potential boulder region marked by the HSP detection marker. The transducer emits coded acoustic wave signals with a frequency of 50-500Hz to the potential boulder region for azimuth positioning. At the same time, the LFM pulse compression mechanism is used to improve the signal-to-noise ratio of directional detection.

[0059] See Figure 3 In this embodiment, in step S5, the transmission-reception timing of the multiple sets of dipole acoustic transducer arrays 3 is synchronized with the rotation cycle of the tunnel boring machine cutterhead, in order to reduce the strong noise interference caused by the contact between the cutterhead cutter and the rock mass.

[0060] In this application, when the cutterhead of the tunnel boring machine rotates to a specific angle, the dipole acoustic transducer 3 is triggered to emit a pulse.

[0061] In this embodiment, in step S6, the Bayesian probabilistic fusion algorithm is implemented based on a feature library of isolated rock reflection acoustic waves established by introducing a historical geological data transfer learning mechanism. This feature library contains HSP response templates and dipole acoustic wave response templates corresponding to isolated rocks under different physical and mechanical parameters. During the calculation process of the Bayesian probabilistic fusion algorithm, the convolutional neural network in the feature library is called to perform similarity matching between the real-time detection data and the feature library, and outputs the probability distribution of the specified type of isolated rock.

[0062] In this embodiment, the Bayesian probabilistic fusion algorithm in S6 includes the following implementation steps:

[0063] S6.1 Setting of prior probability: The prior probability p(isolated boulder) is set according to the historical geological database to determine the probability of the initial existence of isolated boulders. When the tunnel boring machine updates the stratum type in real time during the tunneling process, the prior probability p(isolated boulder) is automatically adjusted to follow the geological changes.

[0064] S6.2 Conditional Probability Modeling:

[0065] HSP conditional probability: Based on historical data statistics, a Gaussian distribution model p(v) is established for the wave velocity anomalies in the isolated rock area. 异常 |Lone stone)~N (μ=20%, σ=5%);

[0066] Conditional probability of dipole acoustic waves: Based on the amplitude attenuation coefficient threshold β>2.5dB / m and the transverse wave splitting angle deviation Δθ>10°, the distribution of measured data is fitted using kernel density estimation, and a joint probability model p(β, Δθ|isolated rock) is established based on the data distribution.

[0067] S6.3, Posterior Probability Calculation: The posterior probability of the existence of the isolated rock is calculated using Bayes' theorem, expressed as follows:

[0068]

[0069] In the formula, the denominator p(v) 异常 , β, Δθ) are calculated using the full probability formula, covering all possible geological conditions.

[0070] The confidence level is graded based on the posterior probability:

[0071] Level 1 risk (posterior probability ≥ 90%): Red alert, shutdown and start advanced drilling.

[0072] Level 2 risk (50%–90%): Yellow alert, increase the frequency of HSP retesting.

[0073] Level 3 risk (<50%): Blue mark, indicating a geological anomaly.

[0074] In this embodiment, the Bayesian probabilistic fusion algorithm obtains the HSP wave velocity anomaly region based on the HSP three-dimensional wave velocity analysis model during the calculation process. The confidence weight of the HSP wave velocity anomaly region is higher than the dipole acoustic wave amplitude attenuation coefficient.

[0075] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the tunnel boulder cooperative detection method described above.

[0076] See Figure 5 The present invention also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the above-described tunnel boulder collaborative detection method.

[0077] This invention also provides a computer program product containing instructions that, when run on a computer, cause the computer to perform the steps of the tunnel boulder cooperative detection method described above.

[0078] It is understood that the systems, devices and storage media provided in the embodiments of the present invention correspond to the methods provided in the embodiments of the present invention, and the explanations, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above-mentioned tunnel boulder collaborative detection method.

[0079] It should be noted that those skilled in the art will understand that all or part of the steps implemented in the embodiments of the present invention can be implemented entirely or partially by software, hardware, firmware, or any combination thereof. When implemented in hardware, it can be implemented entirely or partially by purchasing standard parts or modifications. When implemented in software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid state disks (SSDs)).

[0080] In summary, this invention addresses the problems of low efficiency, high cost, poor timeliness, and susceptibility to working environment in traditional boulder detection methods by proposing a collaborative boulder detection method for tunnels. This method is simple to implement and easy to operate. By combining the HSP method with dipole acoustic long-range detection technology, and through multi-scale, multi-physics field data complementarity, it significantly improves the reliability and resolution of boulder detection, providing key technical support for safe tunnel construction. Furthermore, future integration with intelligent algorithms and real-time monitoring systems can achieve dynamic early warning and risk management, demonstrating promising market application prospects.

[0081] It should be understood that the examples and embodiments described herein are for illustrative purposes only and are not intended to limit the invention. Those skilled in the art can make various modifications or changes based on them. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the protection scope of the invention.

Claims

1. A method for collaborative detection of isolated boulders in tunnels, characterized in that, Includes the following steps: S1. The broadband vibration noise generated by the shield machine cutterhead cutting the rock and soil during tunnel excavation is used as the natural sound source for the HSP near-field detection stage. The reflected sound wave signal is collected by the signal acquisition module arranged behind the cutterhead. S2. In the signal processing module, wavelet packet decomposition is performed on the acquired reflected sound wave signal to extract the characteristic reflected sound wave components of the boulder. S3. The extracted isolated rock feature reflected sound waves are processed by the travel time tomography algorithm to generate a three-dimensional HSP wave velocity distribution model in front of the tunnel boring machine and mark the potential isolated rock areas. S4. Based on the orientation of the potential boulder area marked by the HSP detection, control the array of dipole acoustic wave detection modules arranged in the shield ring of the tunnel boring machine to position and transmit coded acoustic wave signals in the specified direction. S5. The dipole acoustic wave detection module array receives the reflected coded acoustic wave signal and calculates the transverse wave splitting angle deviation and amplitude attenuation coefficient. S6. Combining the HSP three-dimensional wave velocity analysis model and the dipole acoustic wave amplitude attenuation coefficient, a probability distribution map of the isolated rock space in front of the tunnel boring machine is generated based on the Bayesian probability fusion algorithm.

2. The tunnel boulder detection method as described in claim 1, characterized in that, In S1, the signal acquisition module is a triaxial accelerometer (1) installed behind the cutterhead of the tunnel boring machine. The triaxial accelerometer (1) has 6 sets, which are evenly distributed around the circumference of the cutterhead flange (2) of the tunnel boring machine. The angle between two adjacent triaxial accelerometers (1) and the center of the flange (2) is 60° to ensure that the vibration signal of the entire cross section is covered.

3. The tunnel boulder detection method as described in claim 1, characterized in that, In S2, the frequency band dynamics of wavelet packet decomposition should be adapted to the tunneling parameters of the tunnel boring machine during the HSP near-field detection phase.

4. The tunnel boulder detection method as described in claim 1, characterized in that, In S4, the dipole acoustic wave detection module is a dipole acoustic wave transducer (3) pre-embedded inside the shield shell (4) of the tunnel boring machine. The dipole acoustic wave transducer (3) has 4 groups, which are evenly distributed along the circumference of the shield shell (4) of the tunnel boring machine. The angle between two adjacent dipole acoustic wave transducers (3) and the center of the shield shell (4) of the tunnel boring machine is 90°. Each group of dipole acoustic wave transducers (3) includes a transmitting unit (5) and two receiving units (6). The transmitting unit (5) and the two receiving units (6) are arranged in a triangular array, and the two receiving units (6) are respectively distributed on two right-angled sides with the transmitting unit (5) as the center.

5. The tunnel boulder detection method as described in claim 4, characterized in that, In S4, during the detection process, the dipole acoustic wave detection module controls at least two sets of orthogonally distributed dipole acoustic wave transducers (3) to perform focused scanning in a fan-shaped area of ​​θ+15° according to the azimuth angle θ of the potential isolated rock area marked by the HSP detection. At the same time, the LFM pulse compression mechanism is used to improve the directional detection signal-to-noise ratio.

6. The tunnel boulder detection method as described in claim 4, characterized in that, In S5, the transmission-reception timing of the multiple sets of dipole acoustic transducers (3) arrays is synchronized with the rotation cycle of the shield machine cutterhead to reduce the strong noise interference from the contact between the cutterhead cutter and the rock mass.

7. The tunnel boulder detection method as described in claim 1, characterized in that, In S6, the Bayesian probabilistic fusion algorithm is based on a feature library of isolated rock reflection acoustic waves established by introducing a historical geological data transfer learning mechanism. This feature library contains HSP response templates and dipole acoustic wave response templates corresponding to isolated rocks under different physical and mechanical parameters. During the calculation process of the Bayesian probabilistic fusion algorithm, the convolutional neural network in this feature library is called to perform similarity matching between the real-time detection data and the feature library, and outputs the probability distribution of the specified type of isolated rock.

8. The tunnel boulder detection method as described in claim 7, characterized in that, The Bayesian probabilistic fusion algorithm in S6 includes the following implementation steps: S6.1 Setting of prior probability: The prior probability p(isolated boulder) is set according to the historical geological database to determine the probability of the initial existence of isolated boulders. When the tunnel boring machine updates the stratum type in real time during the tunneling process, the prior probability p(isolated boulder) is automatically adjusted to follow the geological changes. S6.2 Conditional Probability Modeling: HSP conditional probability: Based on historical data statistics, a Gaussian distribution model p(v) is established for the wave velocity anomalies in the isolated rock area. 异常 |lone rock); Conditional probability of dipole acoustic waves: Based on the amplitude attenuation coefficient threshold β and the transverse wave splitting angle deviation Δθ, the distribution of measured data is fitted using kernel density estimation, and a joint probability model p(β, Δθ|isolated rock) is established based on the data distribution. S6.3, Posterior Probability Calculation: The posterior probability of the existence of the isolated rock is calculated using Bayes' theorem, expressed as follows: In the formula, the denominator p(v) 异常 , β, Δθ) are calculated using the full probability formula, covering all possible geological conditions.

9. The tunnel boulder detection method as described in claim 7, characterized in that, During the calculation process, the Bayesian probabilistic fusion algorithm obtains the HSP wave velocity anomaly region based on the HSP three-dimensional wave velocity analysis model. The confidence weight of the HSP wave velocity anomaly region is higher than the dipole acoustic wave amplitude attenuation coefficient.

10. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the tunnel boulder cooperative detection method as described in any one of claims 1-9.

Images