Shield tunneling machine integrated full-section geological intelligent sensing system and construction method

Abstract

The invention discloses a shield tunneling machine integrated full-section geological intelligent sensing system and a construction method. The system comprises a composite sensing cutterhead structure, a multi-parameter fusion analysis module and a digital twin driving platform. According to the composite sensing cutterhead structure, detection modules are arranged between blades, and multi-source data such as vibration, water pressure and images are collected; the multi-parameter fusion analysis module performs feature extraction, blade life prediction, dynamic parameter optimization and federated learning coprocessing on the acquired data to generate a stratum parameter analysis result and a construction parameter optimization suggestion; and the digital twin driving platform constructs a three-dimensional geologic model based on an analysis result, so that dynamic adjustment and visualization of the tunneling process of the shield tunneling machine are realized, and the construction efficiency and safety are remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of tunnel construction equipment, and in particular relates to a shield machine integrated full-section geological intelligent perception system and a construction method. Background Art

[0002] Currently, the main stratum exploration technologies commonly used in shield tunneling construction in China include drilling and seismic wave detection. The drilling method involves drilling holes ahead of the tunnel face to collect core samples for geological analysis. However, this method has significant limitations: First, the drilling process is time-consuming and requires frequent replacement of drill bits and equipment, resulting in low detection efficiency. Second, the drilling process may disturb the surrounding strata, affecting the accuracy of detection results. Furthermore, drilling equipment is large and complex to operate, requiring a high construction space requirement, which is particularly disadvantageous in confined tunnel environments. Seismic wave detection involves generating seismic waves within the tunnel or on the surface, and then inferring stratum structure by leveraging the propagation characteristics of these waves in different media. While this method can acquire geological information over a wide area, its accuracy is limited by various factors. For example, the propagation path of seismic waves in complex geological conditions is difficult to accurately predict, and signal attenuation and interference can lead to significant errors in detection results. Furthermore, the installation and commissioning of seismic wave detection equipment is complex and requires specialized technicians, increasing construction costs and time. Therefore, developing a detection technology that is highly accurate, easy to operate, and can provide real-time feedback on stratum information is of great significance for improving the safety and efficiency of shield construction. Summary of the Invention

[0003] The present invention proposes a shield machine integrated full-section geological intelligent perception system and a construction method to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above objectives, the present invention provides a shield machine integrated full-section geological intelligent perception system, comprising:

[0005] Composite sensing cutterhead, used to integrate multimodal sensors and collect full-section geological data in real time;

[0006] A multi-parameter fusion analysis module analyzes the full-section geological data based on a distributed data processing architecture of federated learning to obtain formation parameter analysis results;

[0007] The digital twin driving platform is used to combine the analysis results with the real-time status of the shield machine through virtual mapping and visualization methods, and dynamically adjust the grouting pressure and advancement speed.

[0008] Optionally, the composite sensing cutterhead is provided with a detection module between the blades and a dual-channel piping system is pre-buried inside the cutterhead;

[0009] The detection module is embedded between adjacent blade bases and evenly distributed along the axial direction of the cutter head;

[0010] A dual-channel piping system includes a detection channel and a sampling channel. The detection channel is used for laying optical fibers and cables, and the sampling channel is used for pneumatic rock cuttings recovery.

[0011] Optionally, the detection module includes:

[0012] The micro impact generator is set in the impact cavity at the front of the module and is used to generate a controllable seismic source to excite stress waves in a direction;

[0013] A three-component accelerometer, arranged in a regular tetrahedron configuration, is used to detect vibration signals;

[0014] A miniature high-definition camera unit with a lens group flush with the outer surface of the cutterhead is used for optical imaging of the tunnel face. It is equipped with a cleaning nozzle and LED lighting, and the lighting LED array is arranged in a ring.

[0015] Pore ​​water pressure sensor, used to monitor water pressure changes, with temperature compensation;

[0016] Temperature control board, used to maintain the stable temperature of the pore water pressure sensor.

[0017] Optionally, the multi-parameter fusion analysis module includes:

[0018] The data fusion unit is used to receive the vibration signal collected by the cutterhead sensor, the oil pressure value, torque fluctuation rate and rock parameters collected by the propulsion system, and construct the temperature-pressure coupling parameters;

[0019] The blade life prediction unit is used to analyze vibration characteristics, temperature-pressure coupling parameters and torque gradient based on the online LSTM network to predict the tool wear status in real time;

[0020] Dynamic parameter optimization unit, used to calculate the optimal combination of advancement speed and grouting pressure that meets construction constraints using the improved NSGA-II algorithm;

[0021] The federated learning collaborative unit is used to interact model parameters between the edge computing layer and the cloud analysis layer within the multi-parameter fusion analysis module, and to update the global model through a dynamic weighted average algorithm.

[0022] Optionally, the digital twin driving platform includes:

[0023] Feedback control unit, used to dynamically adjust the shield machine's propulsion speed and grouting pressure based on real-time geological parameters and mechanical status data through a feedforward-feedback composite control algorithm;

[0024] The visualization and anomaly processing unit is used to construct a three-dimensional geological model through the marching cube algorithm and render the abnormal area in real time. When a water pressure mutation rate greater than 0.8 MPa / s or vibration anomaly is detected, a graded warning signal is generated.

[0025] The present invention also provides a construction method of the system, comprising the following steps:

[0026] Start system hardware self-test and digital twin platform loading;

[0027] After self-testing and loading are completed, vibration, water pressure, optical imaging, and rock cuttings data are collected to construct a basic data set;

[0028] Process basic data sets in real time through multi-parameter fusion analysis algorithms to generate early warnings and optimize decisions;

[0029] Virtual mapping and dynamic regulation of the construction process through a digital twin drive platform;

[0030] The propulsion speed and grouting pressure are dynamically adjusted according to the analysis results of the multi-parameter fusion analysis algorithm.

[0031] Optionally, real-time processing of basic data sets using multi-parameter fusion analysis algorithms includes:

[0032] Perform wavelet packet decomposition on the vibration signal to extract the frequency band energy characteristics;

[0033] Construct temperature-pressure coupling parameters;

[0034] Use online LSTM network to predict blade life;

[0035] The improved NSGA-II algorithm is used to optimize the advancing speed and grouting pressure.

[0036] Optional virtual mapping and dynamic control of the construction process through the digital twin driving platform include:

[0037] According to real-time geological parameters and mechanical status data, the shield machine propulsion speed and grouting pressure are dynamically adjusted through a feedforward-feedback composite control algorithm;

[0038] A three-dimensional geological model was constructed using the marching cube algorithm. When a water pressure mutation rate > 0.8 MPa / s was detected, a red warning area was marked in the model.

[0039] Compared with the prior art, the present invention has the following advantages and technical effects:

[0040] This invention integrates multimodal sensors on the shield machine cutterhead to collect real-time data from multiple sources, including vibration, water pressure, and imaging. Combined with a multi-parameter fusion analysis algorithm, it achieves high-precision geological anomaly identification and dynamic construction parameter optimization. The system accurately senses geological anomalies such as faults and fracture zones, providing early warning of potential risks. It also dynamically adjusts the propulsion speed and grouting pressure based on real-time formation parameters, ensuring efficient and stable operation of the shield machine under complex geological conditions. Furthermore, an online LSTM network is used to intelligently predict blade life, enabling proactive maintenance planning and reducing the risk of construction interruptions. Leveraging a federated learning collaborative mechanism, the system achieves efficient collaboration between edge computing and cloud-based analysis, shortening warning response time and enhancing adaptive capabilities. The digital twin drive platform provides intuitive decision support for operators through the real-time construction and visualization of three-dimensional geological models, enabling human-machine collaborative optimization of construction parameters and further improving construction efficiency and safety. In summary, this invention significantly enhances the intelligent level of shield machine construction and provides a strong technical foundation for tunnel construction under complex geological conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0042] Figure 1 This is a diagram showing the distribution of the cutterheads of the detection modules according to an embodiment of the present invention;

[0043] Figure 2 A side view of a detection module according to an embodiment of the present invention;

[0044] Figure 3 A front view of a detection module according to an embodiment of the present invention;

[0045] Figure 4 This is a schematic diagram of the multi-parameter fusion analysis algorithm architecture according to an embodiment of the present invention;

[0046] Figure 5 This is a schematic diagram of the digital twin drive platform architecture of an embodiment of the present invention. DETAILED DESCRIPTION

[0047] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0049] Example 1

[0050] like Figure 1-3 As shown, this embodiment provides a shield machine integrated full-section geological intelligent perception system, including:

[0051] Composite sensing cutterhead, used to integrate multimodal sensors and collect full-section geological data in real time;

[0052] A multi-parameter fusion analysis module analyzes the full-section geological data based on a distributed data processing architecture of federated learning to obtain formation parameter analysis results;

[0053] The digital twin driving platform is used to combine the analysis results with the real-time status of the shield machine through virtual mapping and visualization methods, and dynamically adjust the grouting pressure and advancement speed.

[0054] Furthermore, the composite sensing cutterhead is provided with a detection module between the blades and a dual-channel piping system is pre-buried inside the cutterhead;

[0055] The detection module is embedded between adjacent blade bases and evenly distributed along the axial direction of the cutter head;

[0056] A dual-channel piping system includes a detection channel and a sampling channel. The detection channel is used for laying optical fibers and cables, and the sampling channel is used for pneumatic rock cuttings recovery.

[0057] Furthermore, the detection module includes:

[0058] The micro impact generator is set in the impact cavity at the front of the module and is used to generate a controllable seismic source to excite stress waves in a direction;

[0059] A three-component accelerometer, arranged in a regular tetrahedron configuration, is used to detect vibration signals;

[0060] A miniature high-definition camera unit with a lens group flush with the outer surface of the cutterhead is used for optical imaging of the tunnel face. It is equipped with a cleaning nozzle and LED lighting, and the lighting LED array is arranged in a ring.

[0061] Pore ​​water pressure sensor, used to monitor water pressure changes, with temperature compensation;

[0062] Temperature control board, used to maintain the stable temperature of the pore water pressure sensor.

[0063] The components are set up as follows:

[0064] The composite sensing cutterhead structure arranges 5 sets of detection modules between the blades and a dual-channel piping system is pre-buried inside the cutterhead.

[0065] The detection modules are embedded between adjacent blade bases and evenly distributed along the axial direction of the cutter disc.

[0066] The central axis of the detection module is inclined at an angle to the radial direction of the cutterhead to ensure that the detection direction points to the area 1.5D (D is the cutterhead diameter) in front of the tunnel face.

[0067] The detection module includes a micro impact generator, a three-component accelerometer, a micro high-definition camera unit, a pore water pressure sensor (0-5MPa range), and a temperature control board.

[0068] The micro impact generator can generate a 10-500Hz controllable vibration source and is installed in the impact cavity at the front of the module. The center point is 25mm away from the outer edge of the cutter disc. The axis of the impact rod forms an angle with the tangent direction of the cutter disc and maintains a dynamic gap of 10mm with the motion trajectory of the adjacent blade.

[0069] The three-component accelerometer has a frequency response of 0.1-2kHz and features three accelerometer chips arranged in a regular tetrahedron configuration. The main sensor is located at the module's geometric center, 8mm from the rear end of the impact rod. Auxiliary sensor A is offset by +5mm radially along the cutterhead, and auxiliary sensor B is offset by -3mm axially. To achieve effective interference resistance, silicone vibration isolation pads are used to decouple the cutterhead from the main body, and the electromagnetic shielding layer achieves a 95% coverage ratio. The electromagnetic shielding layer is constructed of μ-metal alloy.

[0070] The micro high-definition camera unit has a lens group: the front lens is flush with the outer surface of the blade disc, the lighting LED array is arranged in a ring around the lens, and the cleaning nozzle has four micro jet holes distributed in a cross shape.

[0071] The pore water pressure sensor has a range of 0-5 MPa. The pressure tap opening is located 8 mm behind the outer edge of the cutterhead, and the temperature compensation module is attached to the back of the sensor (with a spacing of 0.5 mm). The generated pressure transmission path is a carbide pressure pipe, a silicone oil-filled cavity, and a piezoresistive chip.

[0072] The temperature control board is installed on the back of the module so that the temperature difference between the inside and outside does not exceed 15°.

[0073] In the dual-pipeline system, the detection channel is a 50mm diameter stainless steel pipe used for optical fiber / cable laying, and the sampling channel is an 80mm diameter ceramic liner pipe with a built-in pneumatic rock chip recovery system.

[0074] The detection pipeline is arranged along the back side of the cutter head.

[0075] The sampling channel is located 15 mm below the detection module.

[0076] Furthermore, the multi-parameter fusion analysis module includes:

[0077] The data fusion unit is used to receive the vibration signal collected by the cutterhead sensor, the oil pressure value, torque fluctuation rate and rock parameters collected by the propulsion system, and construct the temperature-pressure coupling parameters;

[0078] Real-time fusion of multi-source data includes:

[0079] Data input layer:

[0080] Cutter head sensor: vibration signal (1kHz sampling, real-time transmission of 200ms window data packets)

[0081] Propulsion system: oil pressure value (accuracy ±0.5% FS), torque fluctuation rate (calculation cycle 50ms)

[0082] Geological database: Mohr-Coulomb parameters of rocks from previous exploration;

[0083] Feature extraction:

[0084] The vibration signal is decomposed into 6 layers of wavelet packets to extract the energy proportion E_high in the 3.2-3.8kHz frequency band

[0085] Construct the temperature-pressure coupling parameter ξ=ΔT×ΔP / (T0×P0), where T0 and P0 are initial values

[0086] The blade life prediction unit is used to analyze vibration characteristics, temperature-pressure coupling parameters and torque gradient based on the online LSTM network to predict the tool wear status in real time;

[0087] The blade life prediction model is constructed as follows:

[0088] Prediction algorithm:

[0089] An online LSTM network is used with an input dimension of 12 (including E_high, ξ, torque gradient, etc.);

[0090] Network structure: 64 hidden layers, dropout rate 0.2, sliding window length 120s;

[0091] Training mechanism:

[0092] When the cutterhead speed change is detected to be greater than 15%, the model is triggered to update online;

[0093] The loss function includes the attention mechanism: L = α·MAE + (1-α)·cos_sim (gradient distribution, historical fault characteristics);

[0094] Dynamic parameter optimization unit, used to calculate the optimal combination of advancement speed and grouting pressure that meets construction constraints using the improved NSGA-II algorithm;

[0095] The dynamic parameter optimization decisions are as follows:

[0096] Optimization model:

[0097] Establish a multi-objective function: min[ω1·(W_pred-W_threshold)+ω2·(Q_actual / Q_design-1)2];

[0098] Constraints:

[0099] The propulsion speed v satisfies: 0.8v0≤v≤1.2v0 (v0 is the design value);

[0100] Grouting pressure P satisfies: |ΔP|≤0.3MPa / 10min;

[0101] Solving algorithm:

[0102] The improved NSGA-II algorithm is used, with a population size of 50 and a crossover probability of 0.7;

[0103] A tabu search strategy is introduced to prevent local optimality, and the length of the tabu table is 15.

[0104] Improvements to the NSGA-II algorithm include:

[0105] 1. Dynamic reference point generation mechanism

[0106] Step 1: In the initialization phase, a reference point hyperplane is established and a uniformly distributed reference point set is generated according to formula (1):

[0107]

[0108] Where k is the number of adaptive reference points, which is dynamically adjusted according to formula (2):

[0109]

[0110] t is the current iteration number.

[0111] Step 2: Update the reference point coordinates in each generation and establish the target space mapping through formula (3)

[0112]

[0113] Where α = 0.8 is the empirical coefficient, is the target mean of the current population dimension.

[0114] 2. The entropy-crowding joint evaluation strategy establishes an evaluation system covering three dimensions:

[0115] a) Dominance hierarchy (non-dominated sorting results);

[0116] b) Entropy weight fitness (calculate information entropy weight according to formula 4);

[0117]

[0118] c) Improved crowding distance (Equation 5);

[0119]

[0120] 3. Stage Adaptive Crossover Operator

[0121] Design the bimodal crossover probability function (Equation 6)

[0122]

[0123] Where T is the maximum number of iterations. The initial stage (t≤0.3T) focuses on exploration, while the later stage focuses on development.

[0124] The federated learning collaborative unit is used to interact model parameters between the edge computing layer and the cloud analysis layer within the multi-parameter fusion analysis module, and to update the global model through a dynamic weighted average algorithm.

[0125] The collaborative mechanism of federated learning is as follows:

[0126] Edge-Cloud Interaction:

[0127] The cutter head executes L1 warning (delay < 50ms):

[0128] When E_high lasts for 0.5s>0.25, an emergency speed reduction of 10% is triggered;

[0129] The cloud aggregates the model parameters of each shield machine every 5 minutes:

[0130] Use dynamic weighted averaging: w_i = 1 / (1+σ_i2), where σ_i is the local model loss variance;

[0131] Data Security:

[0132] Gradient information is transmitted using homomorphic encryption, and the key update cycle is 24 hours.

[0133] Furthermore, the multi-parameter comprehensive analysis algorithm establishes a mapping model between shield propulsion parameters and formation parameters as follows:

[0134]

[0135] The multi-parameter comprehensive analysis algorithm is developed based on a distributed data processing architecture of federated learning to achieve collaborative optimization of edge computing (knife head end) and cloud analysis.

[0136] Furthermore, the digital twin driving platform includes:

[0137] Feedback control unit, used to dynamically adjust the shield machine's propulsion speed and grouting pressure based on real-time geological parameters and mechanical status data through a feedforward-feedback composite control algorithm;

[0138] Feedback control is performed as follows:

[0139] Actuator control:

[0140] The propulsion speed is adjusted using feedforward-feedback compound control:

[0141] The feedforward term is based on the geological hardness prediction value H: v_ff = K·H^(-0.7);

[0142] The feedback term is adjusted according to the cutterhead temperature gradient ΔT: v_fb = β·tanh(ΔT / 15);

[0143] Grouting pressure PID parameter self-tuning:

[0144] When the angularity of rock fragments is greater than 40%, the differential gain is increased by 30%.

[0145] The visualization and anomaly processing unit is used to construct a three-dimensional geological model through the marching cube algorithm and render the abnormal area in real time. When a water pressure mutation rate greater than 0.8 MPa / s or vibration anomaly is detected, a graded warning signal is generated.

[0146] Visualization and exception handling are as follows:

[0147] 3D geological model construction:

[0148] Using marching cube algorithm, grid resolution is dynamically adjusted (0.5D-1.2D);

[0149] Abnormal geological body marking: When the water pressure mutation rate is greater than 0.8MPa / s, it will be rendered as a red warning area;

[0150] Human-machine collaborative decision-making:

[0151] The system provides three sets of recommended parameter solutions, which the operator can select or modify within 15 seconds;

[0152] If there is no response within 30 seconds, the solution with the highest confidence level will be automatically executed and a log will be recorded.

[0153] This embodiment also discloses a construction method of the system, comprising the following steps:

[0154] Start system hardware self-test and digital twin platform loading;

[0155] After self-testing and loading are completed, vibration, water pressure, optical imaging, and rock cuttings data are collected to construct a basic data set;

[0156] Process basic data sets in real time through multi-parameter fusion analysis algorithms to generate early warnings and optimize decisions;

[0157] Virtual mapping and dynamic regulation of the construction process through a digital twin drive platform;

[0158] The propulsion speed and grouting pressure are dynamically adjusted according to the analysis results of the multi-parameter fusion analysis algorithm.

[0159] Furthermore, the basic data sets are processed in real time through multi-parameter fusion analysis algorithms including

[0160] Perform wavelet packet decomposition on the vibration signal to extract the frequency band energy characteristics;

[0161] Construct temperature-pressure coupling parameters;

[0162] Use online LSTM network to predict blade life;

[0163] The improved NSGA-II algorithm is used to optimize the advancing speed and grouting pressure.

[0164] Furthermore, the virtual mapping and dynamic control of the construction process through the digital twin driving platform include:

[0165] According to real-time geological parameters and mechanical status data, the shield machine propulsion speed and grouting pressure are dynamically adjusted through a feedforward-feedback composite control algorithm;

[0166] A three-dimensional geological model was constructed using the marching cube algorithm. When a water pressure mutation rate > 0.8 MPa / s was detected, a red warning area was marked in the model.

[0167] Example 2

[0168] like Figure 1-Figure 3 As shown, the present invention provides an integrated full-section geological intelligent perception system and construction method mounted on the cutter head of a shield machine.

[0169] During use, as construction begins, the system begins hardware self-test and loading of the digital twin platform. The cutterhead detection module is powered on for self-test to confirm that the sensor has zero bias, there is no communication delay, and the energy storage is in good condition. The cuttings recovery system is pressure tested; the 5G communication module establishes a secure connection; geological survey baseline data is imported; the mechanical simulation model is loaded; and the federated learning node is initialized.

[0170] As the shield machine starts excavation, multimodal data collection begins.

[0171] Specifically, vibration and water pressure data were collected simultaneously. A micro-impact generator was activated for a frequency sweep of 10 to 500 Hz per 5 seconds. The three-component accelerometer was continuously sampled at 1 kHz with a 16-bit resolution. The pore water pressure sensor was temperature compensated using a PT1000 calibration with an accuracy of ±0.1°C.

[0172] Specifically, the optical imaging trigger mechanism continuously monitors the variance of the vibration signal and triggers the camera when any of the following conditions occur:

[0173] Vibration energy mutation >30% threshold; water pressure gradient change >0.2MPa / m; lithofacies identification confidence <85%.

[0174] Specifically, the cuttings were analyzed online, the negative pressure valve was opened at 0.5 MPa for 2 seconds to allow the cuttings to enter the ceramic liner, and pulse blowing was performed at 0.8 MPa for 0.05 seconds to prevent pipe blockage.

[0175] Specifically, edge computing is used for processing. If a water pressure mutation rate greater than 0.5 MPa / s or a vibration frequency deviation greater than 15% is detected, an L1 warning is triggered and recorded in the construction log. If the SiO2 content in the rock cuttings suddenly changes by more than 20%, an L2 warning is triggered, and parameters are automatically fine-tuned until normal operation is restored.

[0176] Specifically, if Figures 4 and 5 , run the digital twin drive platform and multi-parameter fusion analysis algorithm, detect the blade life in real time, generate a 3D geological model, optimize the grouting pressure, dynamically analyze the early warning threshold, and provide real-time feedback of data and analysis results.

[0177] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A shield machine integrated full-section geological intelligent perception system, characterized by: include: Composite sensing cutterhead, used to integrate multimodal sensors and collect full-section geological data in real time; A multi-parameter fusion analysis module analyzes the full-section geological data based on a distributed data processing architecture of federated learning to obtain formation parameter analysis results; The digital twin driving platform is used to combine the analysis results with the real-time status of the shield machine through virtual mapping and visualization methods, and dynamically adjust the grouting pressure and advancement speed.

2. The system according to claim 1, wherein: The composite sensing cutterhead is provided with a detection module between the blades and a dual-channel piping system is pre-buried inside the cutterhead; The detection module is embedded between adjacent blade bases and evenly distributed along the axial direction of the cutter head; A dual-channel piping system includes a detection channel and a sampling channel. The detection channel is used for laying optical fibers and cables, and the sampling channel is used for pneumatic rock cuttings recovery.

3. The system according to claim 2, characterized in that The detection module includes: The micro impact generator is set in the impact cavity at the front of the module and is used to generate a controllable seismic source to excite stress waves in a direction; A three-component accelerometer, arranged in a regular tetrahedron configuration, is used to detect vibration signals; A miniature high-definition camera unit with a lens group flush with the outer surface of the cutterhead is used for optical imaging of the tunnel face. It is equipped with a cleaning nozzle and LED lighting, and the lighting LED array is arranged in a ring. Pore ​​water pressure sensor, used to monitor water pressure changes, with temperature compensation; Temperature control board, used to maintain the stable temperature of the pore water pressure sensor.

4. The system according to claim 1, wherein: The multi-parameter fusion analysis module includes: The data fusion unit is used to receive the vibration signal collected by the cutterhead sensor, the oil pressure value, torque fluctuation rate and rock parameters collected by the propulsion system, and construct the temperature-pressure coupling parameters; The blade life prediction unit is used to analyze vibration characteristics, temperature-pressure coupling parameters and torque gradient based on the online LSTM network to predict the tool wear status in real time; Dynamic parameter optimization unit, used to calculate the optimal combination of advancement speed and grouting pressure that meets construction constraints using the improved NSGA-II algorithm; The federated learning collaborative unit is used to interact model parameters between the edge computing layer and the cloud analysis layer within the multi-parameter fusion analysis module, and to update the global model through a dynamic weighted average algorithm.

5. The system according to claim 1, wherein: The digital twin drive platform includes: Feedback control unit, used to dynamically adjust the shield machine's propulsion speed and grouting pressure based on real-time geological parameters and mechanical status data through a feedforward-feedback composite control algorithm; The visualization and anomaly processing unit is used to construct a three-dimensional geological model through the marching cube algorithm and render the abnormal area in real time. When a water pressure mutation rate greater than 0.8 MPa / s or vibration anomaly is detected, a graded warning signal is generated.

6. A construction method of the system according to claim 1, characterized in that: The following steps are involved: Start system hardware self-test and digital twin platform loading; After self-testing and loading are completed, vibration, water pressure, optical imaging, and rock cuttings data are collected to construct a basic data set; Process basic data sets in real time through multi-parameter fusion analysis algorithms to generate early warnings and optimize decisions; Virtual mapping and dynamic regulation of the construction process through a digital twin drive platform; The propulsion speed and grouting pressure are dynamically adjusted according to the analysis results of the multi-parameter fusion analysis algorithm.

7. The method according to claim 6, characterized in that Real-time processing of basic data sets through multi-parameter fusion analysis algorithms includes: Perform wavelet packet decomposition on the vibration signal to extract the frequency band energy characteristics; Construct temperature-pressure coupling parameters; Use online LSTM network to predict blade life; The improved NSGA-II algorithm is used to optimize the advancing speed and grouting pressure.

8. The method according to claim 6, characterized in that Virtual mapping and dynamic control of the construction process through the digital twin drive platform include: According to real-time geological parameters and mechanical status data, the shield machine propulsion speed and grouting pressure are dynamically adjusted through a feedforward-feedback composite control algorithm; A three-dimensional geological model was constructed using the marching cube algorithm. When a water pressure mutation rate > 0.8 MPa / s was detected, a red warning area was marked in the model.

Images