A method and system for intelligent identification of strata in shield tunnels

Abstract

This invention discloses an intelligent geological identification method for shield tunnels, comprising: collecting acoustic emission signals and tunneling control parameters during shield tunneling, and performing noise reduction processing to extract acoustic emission time parameters related to rock mass fracture; synchronizing and aligning the noise-reduced acoustic emission events with the tunneling control parameters in time; extracting acoustic emission features from the aligned acoustic emission events and operational features from the tunneling control parameters within a preset analysis window; standardizing and fusing the acoustic emission features and operational features to construct a fused feature vector; and inputting the fused feature vector into a trained geological identification model to obtain the current geological identification result. This invention also discloses a system for the above method. This invention provides advanced and accurate geological perception capabilities for shield tunneling construction, which is of great significance for ensuring construction safety and improving tunneling efficiency.

Description

Technical Field

[0001] This invention relates to the fields of tunnel engineering construction and geological prediction technology, and in particular to a method and system for intelligent identification of strata in shield tunnels based on the fusion of acoustic emission and tunneling parameters. Background Technology

[0002] During shield tunnel construction, accurately identifying the lithology and geological conditions of the strata ahead (such as rock strength and rock quality density, RQD) is crucial for safe and efficient tunneling. Currently, stratum identification mainly relies on preliminary geological surveys or advanced geological forecasts during construction.

[0003] Currently used technologies include borehole drilling, ground-penetrating radar (GPR), and tunnel seismic prediction (TSP). All three have significant limitations in practical engineering applications: While borehole drilling is intuitive, it provides only a single, uninterrupted view, making continuous detection difficult, and it is costly and time-consuming, failing to meet the real-time prediction requirements of continuous tunnel boring machine (TBM) excavation. GPR signals are susceptible to strong interference from underground metal structures and water bodies, resulting in multiple interpretations and a high dependence on personnel experience, with reliability significantly reduced under complex geological conditions. Tunnel seismic prediction (TSP) has low resolution, requires work stoppages, severely restricting construction efficiency and continuity, and cannot predict the most dangerous risk of water inrush. Furthermore, all three methods require a process from data acquisition and processing to report generation, making true real-time, continuous prediction impossible and exhibiting poor real-time performance. Summary of the Invention

[0004] The technical problem to be solved by the embodiments of the present invention is to provide a method and system for intelligent identification of strata in shield tunnels based on the fusion of acoustic emission and tunneling parameters. This method can solve the problem of information technology in the prior art that makes it difficult to predict and identify the actual strata of shield tunnels in real time, autonomously and accurately. In particular, in shield tunnel construction, there are often engineering risks caused by the uncertainty of changes in strata types.

[0005] To address the aforementioned technical problems, this invention provides a method for intelligent identification of strata in shield tunnels based on the fusion of acoustic emission and tunneling parameters, comprising the following steps:

[0006] S1: Collect acoustic emission signals and tunneling control parameters during the shield tunneling process;

[0007] S2: Perform noise reduction processing on the acoustic emission signal and extract acoustic emission event parameters related to rock mass fracture;

[0008] S3: The acoustic emission event parameters after noise reduction are synchronized and aligned with the tunneling control parameters using a unified time base for ground entry time.

[0009] S4: Within the preset analysis window, extract acoustic emission features from the aligned acoustic emission events and extract operational features from the tunneling control parameters;

[0010] S5: The acoustic emission features and the operational features are both standardized to generate basic parameters. Based on the basic parameters, derived parameters are constructed. After the derived parameters are standardized a second time, they are finally concatenated with the basic parameters to generate the final fused feature vector.

[0011] S6: Input the final fused feature vector into the trained stratigraphic identification model to obtain the current stratigraphic identification result.

[0012] Furthermore, the noise reduction process includes preliminary filtering using a bandpass filter; secondary noise reduction using wavelet transform; and third noise reduction using a blind source separation algorithm.

[0013] Furthermore, the tunneling control parameters include average thrust F, cutting torque T, and average tunneling speed. .

[0014] Furthermore, the acoustic emission event parameters include average energy. AE count Amplitude A.

[0015] Further, S4 includes the step of: calculating acoustic emission characteristic parameters from the aligned acoustic emission events within a preset analysis window: AE count. Average energy The amplitude A is calculated, and the tunneling characteristic parameters are calculated from the tunneling control parameters: average thrust F, cutterhead torque T, and average tunneling speed. The specific steps include:

[0016] S41: Extract acoustic emission features using the analyzed acoustic emission events:

[0017] Average Energy Given the energy of a single acoustic emission event within the window, calculate the average energy of all acoustic emission events. ;

[0018] AE count : Frequency of all acoustic emission events within the statistics window: ;

[0019] Amplitude A: Retrieves the peak signal intensity of the acoustic emission event within the calculated window, defined as: ;

[0020] S42: Calculate the tunneling parameter characteristics sequentially using the tunneling data after the unified clock exported from the PLC system:

[0021] Average thrust F: The average level of the tunnel boring machine's thrust within the window is calculated based on the thrust of a single tunnel boring machine obtained from the PLC. P represents the number of instantaneous thrust values ​​collected within the window. For the first The instantaneous value of the thrust;

[0022] Cutting torque T: The average level of cutterhead torque within the window is obtained based on the torque of a single tunnel boring machine obtained from the PLC. Q represents the number of instantaneous torque values ​​collected within the window. For the first The instantaneous value of the thrust;

[0023] Average tunneling speed Based on the start and end times of the window obtained from the PLC, and the corresponding positions of the tunnel boring machine at those times, the average tunneling speed of the tunnel boring machine within the window is calculated using the formula. T end T start L represents the start and end times of the window. end L start The start and end times of the window correspond to the positions of the tunneling machine.

[0024] Furthermore, the standardization step in S5 includes: performing a first standardization using the Z-score method.

[0025]

[0026] in, : Original feature value, μ: Mean of all samples for this feature, σ: Standard deviation of all samples for this feature.

[0027] Furthermore, the step of constructing the derived parameters in S5 includes:

[0028] S51: Determine the standardized feature vector;

[0029] S52: Select highly correlated feature pairs and perform cross operations to construct the following three derived parameters:

[0030] Amplitude × Torque × Thrust product, AE count / average thrust, average energy / average tunneling speed;

[0031] S53: Perform a second standardization process on the three derived parameters, and concatenate the base parameters after the first standardization with the derived parameters after the second standardization to form the final fused feature vector:

[0032] .

[0033] Furthermore, step S6 includes:

[0034] S61: Data collection. During the excavation of a known geological section at the tunnel's starting point, collect a set number of fused feature vectors and label the geological formations.

[0035] S62: Data processing, checking and handling missing values, and standardizing the remaining values;

[0036] S63: Dataset partitioning, which is divided into training set, validation set and test set according to the different uses of the dataset;

[0037] S64: XGBoost model training, using the XGBClassifier model, trains the model on the training set with selected hyperparameters, evaluates the performance on the validation set after each training round, and stops early: if the performance on the validation set no longer improves after N rounds, training is stopped to prevent overfitting;

[0038] S65: Model validation and evaluation. Final evaluation is performed on the test set. The best model is input into the test set that has never been used for training and parameter tuning to obtain the prediction results.

[0039] S66: Model application, that is, inputting the fused feature vector into the trained model, and the model outputs the predicted probability and final category of the stratum category;

[0040] S67: Performance monitoring. Regularly verify whether the model accuracy has decreased using newly labeled data and iterate the model. When tunneling enters a new geological section or performance declines, collect new data, retrain and update the model.

[0041] Accordingly, embodiments of the present invention also provide a system for executing the above-described intelligent identification method for shield tunnel strata based on the fusion of acoustic emission and tunneling parameters, comprising:

[0042] An array of acoustic emission sensors mounted on the tunnel boring machine is used to collect acoustic emission signals;

[0043] The data acquisition unit is used to obtain tunneling control parameters from the tunnel boring machine's PLC system;

[0044] A synchronous clock source is used to provide a unified timestamp for all data for standardization.

[0045] The data processing unit is configured to perform noise reduction, feature extraction, feature standardization, derived parameter construction, feature fusion, and model recognition steps.

[0046] A human-computer interaction interface is used to display the stratigraphic identification results in real time.

[0047] The acoustic emission sensor array is arranged on the back of the tunnel boring machine cutterhead or inside the shield shell.

[0048] Implementing the embodiments of the present invention has the following beneficial effects: The present invention can qualitatively and quantitatively determine the karst cave type and rock fracture zone of the tunnel, which can ensure the accuracy of the stratum information, thereby determining the shield tunneling parameters suitable for the corresponding stratum, improving the efficiency and safety of shield construction, solving the problems of lagging, fuzzy and poor anti-interference ability of the existing technology, and providing advanced and accurate geological perception capabilities for shield construction, which is of great significance for ensuring construction safety and improving tunneling efficiency. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the overall architecture of the system of the present invention;

[0050] Figure 2 This is a flowchart of the data processing and machine learning recognition process of the method of this invention;

[0051] Figure 3 This is a flowchart and effect diagram of the multi-level acoustic emission signal noise reduction processing method used in this invention;

[0052] Figure 4 This is a comparison chart of the recognition accuracy of the fused feature and single feature models in the embodiment;

[0053] Figure 5 This is a schematic diagram of a preferred arrangement of acoustic emission sensors on a tunnel boring machine. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.

[0055] Combination Figure 1 , Figure 2 As shown, an embodiment of the present invention provides a method for intelligent identification of strata in shield tunnels based on the fusion of acoustic emission and tunneling parameters, which is implemented through the following steps.

[0056] S1: Collect acoustic emission signals and tunneling control parameters during the shield tunneling process.

[0057] At least eight, preferably twelve, protective sleeves are installed on the back of the tunnel boring machine cutterhead, each equipped with a built-in high-frequency acoustic emission sensor (frequency range 50-400 kHz). Data is transmitted to a data acquisition unit inside the machine cabin via slip rings. Tunneling parameters are acquired from the tunnel boring machine's PLC system via industrial Ethernet.

[0058] Pre-processing is performed through the following steps before implementation:

[0059] S11: Sound shielding devices need to be installed before construction: Multiple high-voltage, high-interference-resistant acoustic emission sensors (such as piezoelectric acoustic sensors or fiber optic acoustic emission sensors) should be placed at key locations such as the cutterhead panel, shield shell, or main drive of the tunnel boring machine. The sensor wires should use double-shielded twisted-pair cables (inner aluminum foil + outer copper mesh). The collection module housing should use a brass shielding box. All equipment should be grounded at a single point to prevent electromagnetic interference from entering the signal. The sensors should be "double-isolated": fixed with damping and shock-absorbing brackets (silicone + metal composite brackets), and then elastic gaskets (such as PTFE gaskets) should be installed between the sensor and the measured structure to further weaken the transmission of mechanical vibration, allowing for passive listening to collect the original acoustic emission waveform signals generated by rock fracturing during tunneling.

[0060] S12: Real-time acquisition of tunneling parameters from the tunnel boring machine PLC control system, including: total thrust, cutterhead torque, cutterhead speed, tunneling speed, and soil chamber pressure.

[0061] S2: The acoustic emission signal is denoised to extract acoustic emission events related to rock mass fracture.

[0062] The acoustic emission sampling rate is set to 1 MHz, and the tunneling parameter acquisition frequency is set to 1 Hz.

[0063] like Figure 3 As shown, after collecting the original mixed signal from S1, it is first subjected to a bandpass filter of 100kHz-400kHz for preliminary noise reduction; then, wavelet threshold denoising method (using db5 wavelet, decomposition level 5) is used to suppress non-stationary noise; finally, blind source separation algorithms such as independent component analysis (ICA) are used to separate and remove noise components related to hydraulic pump and motor vibration from the multi-sensor mixed signal, retaining the real rock mass fracture signal, that is, the required acoustic emission signal, which includes: average energy ( ), AE count ( ), Amplitude (A).

[0064] S3: The acoustic emission event parameters after noise reduction are synchronized with the tunneling control parameters using a unified time base for ground entry time synchronization.

[0065] A high-precision GPS / BITS clock source or IEEE 1588 Precision Time Protocol (PTP) is selected to provide a unified time base for all data acquisition units, ensuring that all data timestamps are synchronized with an error of less than 1 millisecond. The noise-reduced acoustic emission events are synchronized with the tunneling control parameters at fixed tunneling mileage intervals (e.g., every 1 meter) or fixed time windows.

[0066] S4: Within the preset analysis window, extract acoustic emission features from the aligned acoustic emission events and extract operational features from the tunneling control parameters. This is implemented through the following steps.

[0067] Within the preset analysis window, acoustic emission characteristic parameters are calculated from the aligned acoustic emission events: AE count. Average energy The amplitude A is calculated, and the tunneling characteristic parameters are calculated from the tunneling control parameters: average thrust F, cutterhead torque T, and average tunneling speed. The specific steps include:

[0068] S41: Extract acoustic emission features using the analyzed acoustic emission events:

[0069] Average Energy Given the energy of a single acoustic emission event within the window, calculate the average energy of all acoustic emission events. ;

[0070] AE count : Frequency of all acoustic emission events within the statistics window: ;

[0071] Amplitude A: Retrieves the peak signal intensity of the acoustic emission event within the calculated window, defined as: ;

[0072] S42: Calculate the tunneling parameter characteristics sequentially using the tunneling data after the unified clock exported from the PLC system:

[0073] Average thrust F: The average level of the tunnel boring machine's thrust within the window is calculated based on the thrust of a single tunnel boring machine obtained from the PLC. P represents the number of instantaneous thrust values ​​collected within the window. For the first The instantaneous value of the thrust;

[0074] Cutting torque T: The average level of cutterhead torque within the window is obtained based on the torque of a single tunnel boring machine obtained from the PLC. Q represents the number of instantaneous torque values ​​collected within the window. For the first The instantaneous value of the thrust;

[0075] Average tunneling speed Based on the start and end times of the window obtained from the PLC, and the corresponding positions of the tunnel boring machine at those times, the average tunneling speed of the tunnel boring machine within the window is calculated using the formula. T end T start L represents the start and end times of the window. end L start The start and end times of the window correspond to the positions of the tunneling machine.

[0076] S5: The acoustic emission features and the operational features are both standardized to generate basic parameters. Based on the basic parameters, derived parameters are constructed. After the derived parameters are standardized a second time, they are finally concatenated with the basic parameters to generate the final fused feature vector.

[0077] For the selected acoustic emission parameters: average energy ( ), AE count ( The basic parameters are generated by standardizing the following parameters: amplitude (A); tunneling characteristic parameters: average thrust (F), cutterhead torque (T), and average tunneling speed (V).

[0078] The standardization steps include: performing the first standardization using the Z-score method (the standardized data have a mean of 0 and a standard deviation of 1, eliminating the influence of dimensions):

[0079]

[0080] in, : Original characteristic values ​​(such as acoustic emission energy at a certain moment);

[0081] μ: The mean of this feature across all samples (e.g., the average acoustic emission energy across all time steps).

[0082] σ: The standard deviation of this feature across all samples (reflecting the degree of data dispersion).

[0083] The purpose of standardizing acoustic emission parameters and tunneling characteristic parameters is to eliminate dimensional differences and ensure fair weighting of the two types of data when constructing derived parameters. This method takes into account data such as acoustic emission energy and tunneling torque, which may have slight fluctuations.

[0084] The steps to construct the derived parameters include:

[0085] 1) Determine the standardized feature vectors:

[0086] Acoustic emission eigenvector: [average energy ( ), AE count ( Amplitude (A)

[0087] Eigenvector of tunneling parameters: [average thrust (F), cutterhead torque (T), average tunneling speed (F)] )]

[0088] 2) Select highly correlated feature pairs and perform cross operations to construct the following three derived parameters:

[0089] Amplitude × Torque × Thrust product ( The product of amplitude, torque, and thrust comprehensively reflects the combined load required to drive such a scale of fracturing. It is one of the indicators for measuring "fracture efficiency" or "system energy conversion intensity," and can be used to identify rock-breaking intervals: when the rock mass has good drillability, smaller torque and thrust values ​​(T and F) can produce a large value (AE). The value is within a moderate optimization range. A value that is too high may indicate that the rock mass is hard (requiring high T and F), but the fracture scale (A) is small, resulting in low efficiency; a value that is too low may indicate that the rock mass is weak or the tool is severely worn (low AE amplitude).

[0090] AE count / average thrust ( The ratio of the AE count to the average thrust reflects the number of microfractures induced per unit thrust. This index can be used to assess rock mass integrity: for jointed and fractured rock masses, a relatively small thrust can induce a large number of microfractures along the joints (high...). ),lead to The value is too high. For intact, homogeneous hard rock, a large thrust is required to produce even a small number of fractures. The value is relatively low. Therefore, this parameter can serve as a sensitive indicator of rock mass structure. Additionally, this index can be used to assess the rationality of thrust setting: in a constant rock stratum, if... A sudden drop in value may mean that the thrust setting is too high relative to the current rock conditions (high thrust but low effective fracturing), resulting in excessive energy consumption or excessive fragmentation.

[0091] Average energy / average tunneling speed ( The ratio of average energy to average tunneling velocity reflects the accumulated energy (AE) per unit distance (or unit time) advanced. It measures the relationship between tunneling efficiency and rock breaking energy consumption. This indicator can be used as a measure of rock mass strength or abrasion resistance; that is, under stable operating parameters, if the rock mass becomes harder or more abrasive, more and stronger microfractures are required to advance the same distance. (Increase), but speed It will decrease, leading to The value increases significantly. Therefore, this ratio is also a very efficient early warning indicator for changes in lithology or deterioration of working face conditions.

[0092] 3) It is necessary to analyze the characteristic parameters and tunneling parameters in the acoustic emission event, along with the three currently generated derived parameters: amplitude × torque × thrust product. ), AE count / average thrust ( Average energy / average tunneling speed The data is standardized to eliminate differences in dimensions and ensure fair weighting of the three types of data in subsequent fusion and modeling. This method takes into account that data such as acoustic emission energy and tunneling torque may have slight fluctuations, and uses Z-score standardization (the standardized data has a mean of 0 and a standard deviation of 1, eliminating the influence of dimensions).

[0093] 4) Concatenate the standardized base parameters with the derived parameters after secondary standardization to form the final fused feature vector:

[0094] .

[0095] S6: Input the final fused feature vector into the trained stratigraphic identification model to obtain the current stratigraphic identification result.

[0096] S61: Data collection. During the excavation of the known strata section at the tunnel's starting point, collect 5,000 sets of fused feature vectors and label them with strata tags, such as "intact granite", "fractured shale", and "sand and gravel layer".

[0097] S62: Data processing, checking and processing missing values ​​(such as filling with the mean of the front and rear windows), outliers (vectors under sensor failure or extreme conditions), and standardizing the remaining values;

[0098] S63: Dataset partitioning, that is, according to the different uses of the dataset, it is divided into training set (3500 sets, sampled by stratum), validation set (750 sets), and test set (750 sets, used for final evaluation of the model method's capabilities).

[0099] S64: XGBoost model training, using the XGBClassifier model, trains the model on the training set with selected hyperparameters, evaluates the performance on the validation set after each training round, and stops early: if the performance on the validation set no longer improves after N rounds, training is stopped to prevent overfitting;

[0100] S65: Model validation and evaluation. Final evaluation is performed on the test set. The best model is input into the test set that has never been used for training and parameter tuning to obtain the prediction results.

[0101] S66: Model application, that is, inputting the fused feature vector into the trained model, and the model outputs the predicted probability and final category of the stratum category;

[0102] S67: Performance monitoring. Regularly verify whether the model accuracy has decreased using newly labeled data and iterate the model. When tunneling enters a new geological section or performance declines, collect new data, retrain and update the model.

[0103] On the test set, as shown in the appendix Figure 4As shown, in the identification test of intact granite, the model using only tunneling parameters had an accuracy of 68%, the model using only acoustic emission features had an accuracy of 72%, while the model using the fusion feature vector of the present invention achieved an accuracy of 95%. In addition, for fractured shale and gravel layers, the model using the fusion feature vector of the present invention showed a significant improvement in accuracy compared to the models using tunneling parameters or acoustic emission features alone.

[0104] The above methods can qualitatively and quantitatively determine the karst cave type and rock fracture zone of the tunnel, ensuring the accuracy of the geological information, thereby determining the appropriate shield tunneling parameters for the corresponding strata and improving the efficiency and safety of shield construction.

[0105] This invention also provides a system for performing the above-described method, comprising:

[0106] An array of acoustic emission sensors mounted on the tunnel boring machine is used to collect acoustic emission signals;

[0107] The data acquisition unit is used to obtain tunneling control parameters from the tunnel boring machine's PLC system;

[0108] A synchronous clock source is used to provide a unified timestamp for all data for standardization.

[0109] The data processing unit is configured to perform noise reduction, feature extraction, feature standardization, derived parameter construction, feature fusion, and model recognition steps.

[0110] A human-computer interaction interface is used to display the stratigraphic identification results in real time.

[0111] Among them, the acoustic emission sensor array is arranged on the back of the tunnel boring machine cutterhead or on the inside of the shield shell, such as Figure 5 As shown.

[0112] The above description is merely a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for intelligent identification of strata in shield tunnels based on the fusion of acoustic emission and tunneling parameters, characterized in that, Includes the following steps: S1: Collect acoustic emission signals and tunneling control parameters during the shield tunneling process; S2: Perform noise reduction processing on the acoustic emission signal and extract acoustic emission event parameters related to rock mass fracture; S3: The acoustic emission event parameters after noise reduction are synchronized and aligned with the tunneling control parameters using a unified time base for ground entry time. S4: Within the preset analysis window, extract acoustic emission features from the aligned acoustic emission events and extract operational features from the tunneling control parameters; S5: Both the acoustic emission features and the operational features are standardized to generate basic parameters. Derived parameters are constructed based on these basic parameters. The derived parameters are then standardized a second time and finally concatenated with the basic parameters to generate the final fused feature vector. The step of constructing the derived parameters includes: S51: Determine the standardized feature vector; S52: Select highly correlated feature pairs and perform cross operations to construct the following three derived parameters: Amplitude × Torque × Thrust product, AE count / average thrust, average energy / average tunneling speed; S53: Perform a second standardization process on the three derived parameters, and concatenate the base parameters after the first standardization with the derived parameters after the second standardization to form the final fused feature vector: ； in, , , The values ​​represent the average energy, AE count, and amplitude A of the acoustic emission characteristics after the first normalization using the Z-score method, respectively. 、 、 The average thrust F, torque T, and average tunneling speed represent the operational characteristics, respectively. The result after the first standardization using the Z-score method; 、 、 These represent the results of amplitude × torque × average thrust, AE count / average thrust, and average energy / average tunneling speed after the second standardization, respectively. Indicates AE counting, Indicates average energy; S6: Input the final fused feature vector into the trained stratigraphic identification model to obtain the current stratigraphic identification result.

2. The intelligent identification method for shield tunnel strata based on the fusion of acoustic emission and tunneling parameters according to claim 1, characterized in that, The noise reduction process includes preliminary filtering using a bandpass filter; secondary noise reduction using wavelet transform; and a third noise reduction using a blind source separation algorithm.

3. The intelligent identification method for shield tunnel strata based on the fusion of acoustic emission and tunneling parameters according to claim 1, characterized in that, S4 includes the step of: calculating acoustic emission characteristics from aligned acoustic emission events within a preset analysis window: AE count. Average energy The amplitude A is calculated, and the operating characteristics are calculated from the tunneling control parameters: average thrust F, torque T, and average tunneling speed. The specific steps include: S41: Extract acoustic emission features using the analyzed acoustic emission events: Average Energy Given the energy of a single acoustic emission event within the window, calculate the average energy of all acoustic emission events. ; AE count : Frequency of all acoustic emission events within the statistics window: ; Amplitude A: Retrieves the peak signal intensity of the acoustic emission event within the calculated window, defined as: ; S42: Calculate the operational characteristics sequentially using the tunneling data after the unified clock exported from the PLC system: Average thrust F: The average level of the tunnel boring machine's thrust within the window is calculated based on the thrust of a single tunnel boring machine obtained from the PLC. P represents the number of instantaneous thrust values ​​collected within the window. For the first The instantaneous value of the thrust; Torque T: The average level of cutterhead torque within the window is obtained based on the torque of a single tunnel boring machine obtained from the PLC. Q represents the number of instantaneous torque values ​​collected within the window. For the first The instantaneous value of the thrust; Average tunneling speed Based on the start and end times of the window obtained from the PLC, and the corresponding positions of the tunnel boring machine at those times, the average tunneling speed of the tunnel boring machine within the window is calculated using the formula. T start T end These are the start and end times of the window, L. start L end These represent the start and end times of the window, and the corresponding positions of the tunneling machine.

4. The intelligent identification method for shield tunnel strata based on the fusion of acoustic emission and tunneling parameters according to claim 3, characterized in that, The standardization steps in S5 include: performing the first standardization using the Z-score method. in, : Original feature value, μ: Mean of all samples for this feature, σ: Standard deviation of all samples for this feature.

5. The intelligent identification method for shield tunnel strata based on the fusion of acoustic emission and tunneling parameters according to claim 4, characterized in that, S6 includes the following steps: S61: Data collection. During tunnel excavation in the known geological section at the tunnel's starting point, collect the required number of fused feature vectors and label the geological formations. S62: Data processing, checking and handling missing values, and standardizing the remaining values; S63: Dataset partitioning, based on the different uses of the dataset, into training set, validation set, and test set; S64: XGBoost model training, using the XGBClassifier model, trains the model on the training set with selected hyperparameters, evaluates the performance on the validation set after each training round, and stops early: if the performance on the validation set no longer improves after N rounds, training is stopped to prevent overfitting; S65: Model validation and evaluation. Final evaluation is performed on the test set. The best model is input into the test set that has never been used for training and parameter tuning to obtain the prediction results. S66: Model application, the final fused feature vector is input into the trained model, and the model outputs the predicted probability and final category of the stratum category; S67: Performance monitoring. Regularly verify whether the model accuracy has decreased using newly labeled data and iterate the model. When tunneling enters a new geological section or performance declines, collect new data, retrain and update the model.

6. A system for use in the method according to any one of claims 1-5, characterized in that, include: An array of acoustic emission sensors mounted on the tunnel boring machine is used to collect acoustic emission signals; The data acquisition unit is used to obtain tunneling control parameters from the tunnel boring machine's PLC system; A synchronous clock source is used to provide a unified timestamp for all data for standardization. The data processing unit is configured to perform noise reduction, feature extraction, feature standardization, derived parameter construction, feature fusion, and model recognition steps. A human-computer interaction interface is used to display the stratigraphic identification results in real time.

7. A system for use in the method of claim 6, characterized in that, The acoustic emission sensor array is arranged on the back of the tunnel boring machine cutterhead or inside the shield shell.

Images