TBM shield shell multi-point stress real-time monitoring and anti-jamming self-adaptive adjustment system

Abstract

This invention relates to the field of tunneling technology, specifically disclosing a real-time monitoring system for multi-point stress on a TBM shield and an adaptive adjustment system to prevent jamming. The system includes a multi-point stress sensing module, a data fusion and state reconstruction module, a jamming risk dynamic assessment module, a multi-objective optimization decision-making module, and an adaptive adjustment execution module. It collects shield stress data in real time using a high-density sensor array, reconstructs a stress distribution cloud map, dynamically assesses jamming risk, and generates a multi-objective optimization adjustment strategy. Finally, the various execution subsystems work together to achieve closed-loop adaptive control, thereby improving the continuity, safety, and efficiency of TBM tunneling under adverse geological conditions.

Description

Technical Field

[0001] This invention belongs to the field of tunnel technology, specifically relating to a real-time monitoring system for multi-point stress on TBM shield shell and an adaptive adjustment system for preventing jamming. Background Technology

[0002] In the field of large-scale underground engineering and tunnel excavation, the application of full-face tunnel boring machines, especially TBMs, is the core equipment for achieving efficient and safe construction. As a key component that directly contacts the surrounding rock and bears complex geological loads, the stress state of its shield structure directly affects the overall machine's propulsion efficiency and operational safety.

[0003] Stress monitoring of the TBM shield and adaptive adjustment of the machine based on monitoring data are key technologies for ensuring the continuity and stability of the tunneling process. This technology aims to dynamically adjust tunneling parameters by sensing the load distribution around the shield in real time to prevent the shield from being stuck by the surrounding rock, thereby maintaining tunneling efficiency and reducing equipment risks.

[0004] Existing technologies typically rely on static stress data from a limited number of measuring points or the operator's experience to assess the stress condition of the shield, making it difficult to comprehensively and accurately reflect the spatial non-uniformity and time-varying nature of the interaction between the shield and the surrounding rock. Traditional monitoring schemes suffer from sparse measuring point layout and delayed data updates, resulting in insufficient ability to identify early signs of localized stress concentration or machine jamming risks.

[0005] Meanwhile, existing regulation systems mostly employ fixed threshold alarms and manual intervention, lacking an adaptive decision-making mechanism based on multi-source stress information. This makes it impossible to generate and execute optimal anti-jamming regulation strategies in real time under complex and variable geological conditions. This disconnect between monitoring and control results in a higher risk of TBM jamming and structural damage when traversing weak, uneven, or jointed strata. Therefore, there is an urgent need to develop an adaptive regulation system capable of real-time and accurate sensing of stress at multiple measurement points on the shield and deeply integrated with anti-jamming control. Summary of the Invention

[0006] The technical problem this invention aims to solve is to overcome the shortcomings of existing TBM shield stress monitoring technologies, such as sparse measuring points, data lag, reliance on manual experience for anti-jamming adjustments, and a lack of adaptive decision-making capabilities. This invention provides a real-time multi-point stress monitoring system for TBM shields and an adaptive anti-jamming adjustment system. This system aims to achieve high-density, real-time sensing of the interaction forces between the shield and the surrounding rock, and based on this, to construct a closed-loop control system capable of autonomously analyzing jamming risks and dynamically generating and executing optimal adjustment strategies.

[0007] The technical solution of the present invention is as follows: The system consists of a multi-point stress sensing module, a data fusion and state reconstruction module, a card machine risk dynamic assessment module, a multi-objective optimization decision module, and an adaptive adjustment execution module.

[0008] The multi-point stress sensing module is responsible for collecting real-time stress data of the shield structure. This module contains an array of 256 fiber optic stress sensors densely distributed at specific circumferential and axial positions on the outer surface of the shield. Each sensor synchronously measures the contact pressure value at its location at a sampling frequency of 1000 Hz. The sensor array is connected to the data acquisition box via a built-in corrosion-resistant armored optical cable. The data acquisition box integrates a signal demodulation unit and a temperature compensation unit to ensure the accuracy and stability of the stress measurement data.

[0009] The data fusion and state reconstruction module receives the raw stress data stream from the multi-point stress sensing module. This module first preprocesses the input data by denoising and removing outliers. Further, it uses the Kriging space interpolation algorithm to reconstruct a continuous stress distribution cloud map covering the entire outer surface of the shield based on stress measurements from 256 discrete points. This reconstruction process is updated every 0.1 seconds, thereby generating a spatiotemporal evolution sequence of the shield's stress state.

[0010] The system jamming risk dynamic assessment module quantifies risk based on the stress distribution cloud map sequence output by the data fusion and state reconstruction module. This module incorporates a jamming risk quantification model, which defines the risk index as a weighted composite function of local stress concentration, stress gradient change rate, and the area growth rate of high-stress zones. Specifically, the module calculates the average stress within a 2-degree circumferential fan-shaped region on the shield surface in real time, identifying stress concentration areas exceeding 85% of the preset material yield strength. Simultaneously, it calculates the changes in the stress distribution cloud map within adjacent update cycles, extracting the maximum value and location of the stress gradient. Furthermore, it statistically analyzes the expansion rate of the high-stress area over the past 3 seconds. These three parameters are input into the risk quantification model, outputting a dimensionless jamming risk index within the range of 0 to 1. When the index exceeds 0.7, the system is determined to have entered a high jamming risk state.

[0011] The multi-objective optimization decision module responds to the risk index output by the TBM jamming risk dynamic assessment module. The core of this module is a multi-objective optimization solver, whose objective function simultaneously minimizes the jamming risk index, maximizes the tunneling speed, and minimizes the total energy consumption of the propulsion system. The decision variables for this module are the four key tunneling parameters of the TBM: The optimization process is constrained by a series of limitations, including the maximum output limits of each actuator, the ultimate strength of the shield structure, and the allowable deviation of the tunnel design axis. This module employs a non-dominated sorting genetic algorithm with an elitist strategy to solve the aforementioned multi-objective optimization problem, ultimately outputting a set of optimal tunneling parameter adjustment schemes under the current geological and machine conditions. The total thrust, cutterhead torque, shield tail clearance adjustment, and articulated system deflection angle are all considered.

[0012] The adaptive adjustment execution module receives tunneling parameter adjustment schemes from the multi-objective optimization decision-making module. This module includes a propulsion hydraulic subsystem controller, a cutterhead drive subsystem controller, a tail shield sealing subsystem controller, and an articulated hydraulic subsystem controller. Each controller precisely adjusts the output of its corresponding actuator according to the received adjustment commands. The propulsion hydraulic subsystem controller sets the pressure and flow rate of the main propulsion cylinder according to the adjustment scheme. The cutterhead drive subsystem controller adjusts the output torque and speed of the variable frequency motor. The tail shield sealing subsystem controller adjusts the tail shield grease injection pressure to change the gap between the shield and the tunnel segments. The articulated hydraulic subsystem controller controls the extension and retraction of the articulated cylinders to achieve fine-tuning of the shield's attitude. The execution status and feedback signals of all control commands are collected in real time and sent back to the data fusion and state reconstruction module, forming a closed-loop control circuit.

[0013] In a preferred embodiment of the present invention, the fiber Bragg grating sensor deployment strategy in the multi-point stress sensing module is optimized based on the mechanical properties of the shield structure. The sensors are non-uniformly distributed across the shield's cut ring, support ring, and tail ring. Specifically, in the soil-facing region of the cut ring, where stress concentration is expected to be high, the sensor density is twice that of other regions. Each sensor node is encapsulated within a special alloy protective shell and reliably coupled to the shield material through a dual fixing method of magnetic attraction and welding.

[0014] Furthermore, the Kriging interpolation algorithm used in the data fusion and state reconstruction module has a variogram model customized based on prior knowledge of the shield's geometry and typical surrounding rock load distribution. This module also integrates a self-learning mechanism, which dynamically optimizes the interpolation algorithm parameters by comparing historical reconstruction data with subsequent actual stress verification results, thereby improving the accuracy of stress field reconstruction.

[0015] Furthermore, the weighting coefficients of the risk quantification model in the dynamic assessment module for machine jam risk are not fixed. The system is pre-built with a training database based on a massive amount of historical TBM tunneling cases. Through an online learning mechanism, this module dynamically adjusts the weighting of the three input parameters in the risk quantification model according to the current tunneling strata type and the machine jam warning records, making the risk assessment more closely aligned with the current engineering reality.

[0016] Furthermore, a real-time feedback correction mechanism is introduced into the solution process of the multi-objective optimization decision module. This module continuously receives actual parameters from the adaptive adjustment execution module and risk indices updated by the machine risk dynamic assessment module. If, within two control cycles after the adjustment plan is implemented, the risk index does not decrease by 80% of the expected target, the optimization solver will immediately restart the optimization calculation with the current machine state as the initial condition, generating a corrected adjustment plan until the risk is effectively controlled.

[0017] Furthermore, the entire system operates within a hierarchical, synchronous scheduling framework. Data acquisition and stress reconstruction run at high speed with a cycle of 100 milliseconds. Card jam risk assessment and multi-objective optimization decision-making are performed with a cycle of 500 milliseconds. Adaptive adjustment execution, based on the mechanical response characteristics of each subsystem, completes the issuance and execution of control commands within a cycle ranging from 10 to 100 milliseconds. The system ensures data synchronization and task coordination among modules through a central timing controller.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: By combining a high-density fiber optic grating sensor array with a spatial interpolation algorithm, the stress state of the TBM shield can be accurately reconstructed from discrete points to a continuous field, significantly improving the early identification capability of local stress concentration phenomena.

[0019] By constructing a dynamic risk assessment model that integrates multi-dimensional stress characteristics, the risk of machine jamming is transformed from a qualitative, empirical judgment into a quantifiable index, providing a scientific basis for accurate early warning. Through the establishment of a multi-objective optimization decision-making mechanism centered on reducing machine jamming risk, ensuring tunneling efficiency, and controlling energy consumption, a leap from passive response to proactive optimization in anti-jamming regulation has been achieved, enabling the automatic generation of systematic optimal regulation strategies under complex geological conditions.

[0020] Through the rapid and precise linkage and closed-loop feedback control of each execution subsystem, an adaptive intelligent system integrating perception, decision-making, and execution is formed, which fundamentally improves the continuity, safety, and overall efficiency of TBM tunneling under adverse geological conditions. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall technical solution architecture of the TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of multi-point stress sensing and data fusion reconstruction in this invention; Figure 3 This is a logical flowchart of the dynamic risk assessment and multi-objective optimization decision-making of the card machine in this invention; Figure 4 This is a schematic diagram of the multi-level interaction relationship and data flow between adaptive adjustment execution and closed-loop feedback control in this invention. Detailed Implementation

[0022] Please refer to the attached document. Figures 1 to 4This embodiment details the technical implementation of the TBM shield shell multi-point real-time force monitoring and anti-jamming adaptive adjustment system. The system aims to construct a complete perception-decision-execution closed loop through high-density sensor networks, real-time data processing, intelligent risk assessment, and multi-objective optimization control to address the risk of shield shell jamming in tunnel boring machines under complex geological conditions.

[0023] The multi-point stress sensing module is the foundation for realizing the system's functions. The core of this module is a dense array of 256 high-precision fiber Bragg grating stress sensors. These sensors are non-uniformly distributed according to the mechanical characteristics of the shield structure. Specifically, the strategy involves using differentiated density configurations in three key areas of the shield: the notch ring, the support ring, and the tail ring.

[0024] In the cut-off face region, where stress concentration is expected to increase significantly, the sensor deployment density is twice that of other areas, ensuring precise capture of the stress state in this high-risk area. Each fiber optic stress sensor is encapsulated in a protective housing made of a special alloy, designed to withstand the impact of high-pressure mud and water and abrasion from rock cuttings generated during tunneling. The sensors are fixed to the shield material using a dual mechanism combining magnetic attraction and welding. The magnetic attraction provides initial positioning and pre-tightening, while the welding provides permanent, high-strength mechanical coupling, ensuring effective transmission of stress waves.

[0025] The sensor array is interconnected and connected to the data acquisition box via built-in corrosion-resistant armored optical cables. The armor layer effectively prevents physical damage to the optical cables from sharp objects in the construction environment. The data acquisition box integrates a signal demodulation unit and a temperature compensation unit. The signal demodulation unit is responsible for converting the offset of the fiber optic grating's reflected wavelength into the corresponding micro-strain value, with a conversion accuracy better than 0.1%.

[0026] The temperature compensation unit incorporates a highly stable temperature sensor that monitors the ambient temperature at the sensor mounting point in real time. It then corrects the original strain data based on the temperature sensitivity coefficient of the fiber optic grating, eliminating measurement errors introduced by temperature variations and ensuring that the output contact pressure value accurately reflects mechanical stress. All 256 sensors synchronously acquire data at a sampling frequency of 1000 Hz. This high-frequency sampling capability provides a data foundation for capturing the transient dynamic characteristics of the interaction between the shield and the surrounding rock.

[0027] The data fusion and state reconstruction module receives the raw stress data stream from the multi-point stress sensing module. This module first performs a data preprocessing procedure, including noise reduction and outlier removal. Noise reduction employs a wavelet transform-based threshold denoising algorithm to effectively separate high-frequency noise from actual stress fluctuations in the signal. Outlier removal is achieved through statistical process control, calculating the mean and standard deviation of each sensor data stream within a time window, and identifying data points deviating from the mean by more than three times the standard deviation as outliers, which are then replaced or marked. The preprocessed data then enters the core reconstruction stage. This module uses a Kriging space interpolation algorithm to reconstruct a continuous stress distribution cloud map covering the entire outer surface of the shield shell based on stress measurements from 256 discrete measuring points.

[0028] Please refer to the attached document. Figure 2 The key to this reconstruction process lies in the establishment of the variogram model. The variogram model used in this system is not a general model, but rather a customized design based on prior knowledge of the complex geometry of the shield and typical surrounding rock load distribution. Its model parameters are obtained by fitting historical tunneling data, thus more accurately describing the spatial correlation of the stress field on the shield surface. Interpolation calculations are updated at a period of 0.1 seconds, generating a series of temporally continuous stress distribution cloud maps, constituting a spatiotemporal evolution sequence of the shield's stress state. To further improve reconstruction accuracy, this module also integrates a self-learning mechanism. This mechanism continuously records historical reconstruction data and compares it with the actual stress trends indirectly verified by other means. Based on the comparison results, it dynamically adjusts key parameters in the Kriging interpolation algorithm, such as the range and sill value, enabling the stress field reconstruction model to adapt to changes in geological conditions.

[0029] The core task of the shield jamming risk dynamic assessment module is to dynamically quantify the shield jamming risk based on the stress distribution cloud map sequence output by the data fusion and state reconstruction module. This module incorporates a jamming risk quantification model, which defines the risk index as a weighted composite function of three key stress characteristic parameters.

[0030] First, the module calculates the average stress value in every 2-degree circumferential fan-shaped area on the shield surface in real time, and scans the entire shield surface to identify all stress concentration areas where the average stress exceeds 85% of the preset material yield strength.

[0031] Secondly, the module calculates the difference between the stress distribution cloud maps of two adjacent update cycles, i.e., 0.1 seconds apart, and extracts the maximum value of the stress gradient on the entire shield surface and its spatial coordinates.

[0032] Finally, this module calculates the expansion rate of the total area of ​​the identified high-stress regions over the past 3-second time window. These three parameters—local stress concentration, stress gradient change rate, and high-stress area growth rate—are input into the risk quantification model in real time. The model then calculates the final risk index using a weighted summation function.

[0033] ; Where R represents the calculated card risk index, whose value range is normalized to between 0 and 1. Represents the normalized local stress concentration This represents the normalized rate of change of the stress gradient. This represents the normalized growth rate of the high-stress zone area. , , These are the weight coefficients of these three feature parameters, and they satisfy... + + =1. The weighting coefficients of this risk quantification model are not fixed.

[0034] The system is pre-built with a training database based on a massive amount of historical TBM tunneling cases. This module uses an online learning mechanism to dynamically adjust the weighting coefficients based on the current tunneling strata type and the historical TBM jam warnings and handling effects recorded by the system. , , The allocation ratio. For example, in weak and heterogeneous formations, more attention may be paid to abrupt changes in stress gradients, therefore The weight of [something] will increase accordingly; while in hard, intact rock formations, more attention may be paid to the absolute stress level. The weight of the risk index will then take precedence. This allows the risk assessment to be more closely aligned with the actual situation of the current project. When the calculated risk index R exceeds the threshold of 0.7, the module determines that the system has entered a high-risk state and triggers subsequent decision-making processes.

[0035] The multi-objective optimization decision-making module is activated upon receiving a high-risk status signal from the TBM risk dynamic assessment module. The core of this module is the multi-objective optimization solver, whose task is to find a set of optimal TBM tunneling parameter adjustment schemes. Please refer to the appendix. Figure 3 The optimization problem is defined as follows: The objective function is to simultaneously minimize the jamming risk index, maximize the tunneling speed, and minimize the total energy consumption of the propulsion system. This is a typical multi-objective optimization problem, where there are often conflicts between the objectives. The decision variables are the four key tunneling parameters of the TBM: total propulsion force, cutterhead torque, tail shield clearance adjustment, and articulated system deflection angle. The optimization process is subject to a series of hard and soft constraints.

[0036] Hard constraints include the maximum output limits of each actuator, such as the maximum thrust of the main propulsion cylinder, the maximum torque of the cutterhead drive motor, and the maximum stroke of the articulated cylinder; the strength limit of the shield structure material to ensure that the adjusted stress state will not damage the shield; and the allowable deviation range of the tunnel design axis to guarantee the tunnel forming quality. Soft constraints involve the stability of the tunneling process, such as the fluctuation range of thrust and torque. This module uses a non-dominated sorting genetic algorithm with an elitist strategy to solve the above multi-objective optimization problem.

[0037] The algorithm initializes a population of 100 individuals, each representing a candidate solution for a set of tunneling parameters. It iterative evolution is achieved through genetic operations such as selection, crossover, and mutation. Non-dominated sorting and crowding calculations are used to maintain the diversity and convergence of the solution set, while an elitist strategy ensures that superior individuals are not lost. After a preset 500 generations of evolution or when the convergence condition is met, the algorithm outputs a set of tunneling parameter adjustment schemes that are optimal under the current geological and machine conditions. This scheme represents a compromise within the Pareto-optimal solution set. This module also incorporates a real-time feedback correction mechanism.

[0038] The module continuously receives feedback from the adaptive adjustment execution module regarding the actual execution values ​​of each parameter, as well as the risk index updated by the machine risk dynamic assessment module. If, within two control cycles (i.e., one second) after the adjustment plan issued by the optimization decision module is executed, the monitored risk index decrease does not reach the expected decrease target of 80%, the optimization solver will immediately restart the optimization calculation with the current actual machine state as the initial condition, generate a corrected adjustment plan, and continue until the risk index is effectively controlled.

[0039] The adaptive adjustment execution module is responsible for translating the theoretical adjustment schemes generated by the multi-objective optimization decision-making module into the actual actions of each actuator. This module is a distributed control system containing four main subsystem controllers: the propulsion hydraulic subsystem controller, the cutterhead drive subsystem controller, the shield tail sealing subsystem controller, and the articulated hydraulic subsystem controller.

[0040] Please refer to the attached document. Figure 4 Each controller independently and precisely adjusts the output of its corresponding actuator based on the specific adjustment instructions received. The propulsion hydraulic subsystem controller, according to the total thrust target value set in the adjustment plan, precisely adjusts the inlet pressure and flow rate of the main propulsion cylinder assembly via proportional servo valves, thereby controlling the magnitude of the total thrust. The cutterhead drive subsystem controller, typically based on variable frequency speed control technology, adjusts the output current frequency and voltage of the drive motor according to the cutterhead torque target value set in the adjustment plan, thereby controlling the rotational torque and speed of the cutterhead.

[0041] The tail seal subsystem controller, based on the tail clearance adjustment amount set in the adjustment plan, controls the amount and pressure of grease injected into the tail seal cavity by adjusting the outlet pressure of the tail grease injection pump, thereby actively changing the gap between the shield shell and the installed segments, affecting the frictional resistance at the rear of the shield shell. The articulation hydraulic subsystem controller, based on the articulation system deflection angle target set in the adjustment plan, controls the extension and retraction and speed of each articulation cylinder, achieving fine-tuning of the relative attitude of the front and rear parts of the shield body to optimize the contact state between the shield body and the surrounding rock.

[0042] All controllers integrate high-precision sensors to detect the actual state of the actuators, such as cylinder pressure sensors, motor speed encoders, grease pressure sensors, and cylinder displacement sensors. These feedback signals are acquired in real time and sent back to the data fusion and state reconstruction module via the system bus. This constitutes a crucial closed-loop control circuit, enabling the system to continuously evaluate the effectiveness of control actions and provide data support for subsequent decision optimization.

[0043] The entire system relies on a sophisticated hierarchical synchronous scheduling framework. Data acquisition and stress field reconstruction tasks are set to the highest priority, running at high speed with a period of 100 milliseconds to ensure real-time sensing. Risk assessment and multi-objective optimization decision-making tasks are executed with a period of 500 milliseconds, balancing computational complexity with the timeliness of decision response. Adaptive adjustment tasks are allocated within control periods ranging from 10 to 100 milliseconds based on the inherent response characteristics of each mechanical subsystem; for example, the propulsion system, with its slower response, has a slightly longer period, while the articulated system, with its faster response, can have a shorter period.

[0044] The system uses a central timing controller to uniformly manage the task scheduling and clock synchronization of each module, ensuring consistent timestamps for data transmission between different modules and preventing decision-making errors due to asynchronous data. All data interaction between modules is achieved via high-speed industrial Ethernet. The communication protocol adopts a customized application layer protocol based on TCP / IP, which defines a strict data frame format, including frame header, data length, module identifier, timestamp, data payload, and cyclic redundancy check (CRC) code, ensuring the reliability and integrity of data transmission. The system also has a robust exception handling mechanism. When any module detects a communication timeout, data verification failure, actuator malfunction, or critical parameter exceeding limits, it immediately reports an exception code to the central timing controller. The system can then degrade operation or safely shut down according to preset strategies, ensuring the safety of equipment and personnel.

[0045] This embodiment provides an alternative implementation scheme for a real-time monitoring system of multi-point stress on TBM shield and an adaptive adjustment system for preventing jamming. Its core feature is that the multi-point stress sensing module and the jamming risk dynamic assessment module have been specifically optimized to adapt to the working conditions of tunneling in extremely hard and abrasive rock formations.

[0046] Regarding the multi-point stress sensing module, considering the severe vibrations and significant impact loads encountered during hard rock tunneling, this embodiment features an enhanced design for the packaging and installation of the fiber optic stress sensor. The sensor's protective housing is made of special high-hardness wear-resistant alloy steel, and its surface undergoes nitriding treatment to further improve surface hardness and impact resistance.

[0047] The sensor and shield are secured using a triple-fixation mechanism, in addition to magnetic attraction and welding, by adding a high-toughness epoxy resin structural adhesive bonding process. This effectively suppresses fretting wear and connection loosening that may occur in the sensor under high-frequency vibration environments, ensuring long-term measurement stability. Furthermore, the signal demodulation unit inside the data acquisition box has been upgraded to a model with a wider dynamic range and higher shock resistance, capable of accurately capturing instantaneous extremely high stress pulses that may occur during hard rock tunneling.

[0048] In the dynamic assessment module for machine jamming risk, this embodiment expands the characteristic parameters of the risk quantification model, recognizing that machine jamming risks in hard rock formations are often related to mechanisms such as cutterhead jamming and localized shield lock-up. In addition to the existing local stress concentration, stress gradient change rate, and high-stress zone area growth rate, a new characteristic parameter—stress distribution asymmetry—is added. This parameter is quantified by calculating the stress difference at symmetrical locations along the circumference of the shield. In hard rock tunneling, if the surrounding rock on one side is abnormally hard, it can lead to severe uneven stress distribution on both sides of the shield, generating a huge eccentric moment and significantly increasing the risk of machine jamming. Therefore, stress distribution asymmetry becomes an important risk indicator. Accordingly, the risk quantification model is adjusted to a four-parameter weighted model.

[0049] ; in, This represents the asymmetry of the normalized stress distribution. Let these be the corresponding weight coefficients. At this point, the weight coefficients satisfy... The online learning mechanism also applies to this extended model, dynamically adjusting the four weighting coefficients based on the unique data patterns of hard rock tunneling, making risk assessment more effective at capturing early signs of jamming in hard rock. The risk judgment threshold can be adaptively adjusted based on the uniaxial compressive strength of the specific rock. For example, in extremely high-strength rock formations, the threshold can be appropriately increased from 0.7 to 0.75 to avoid excessively frequent false alarms.

[0050] The objective function and constraints of the multi-objective optimization decision module are also adjusted accordingly. In the context of hard rock tunneling, the weight of maximizing tunneling speed may be relatively reduced, while ensuring tunneling stability and reducing the vibration of the cutterhead and shield become more important implicit objectives. Therefore, during the optimization process, stricter constraints are imposed on decision variables such as the rate of change of cutterhead torque to prevent sudden torque changes from impacting the transmission system. Simultaneously, the adjustment of the articulated system deflection angle is given greater importance to counteract the off-center loading trend caused by asymmetrical loads through active attitude adjustment.

[0051] In the adaptive adjustment execution module, for hard rock conditions, the control algorithm of the propulsion hydraulic subsystem controller enhances its ability to suppress pressure shocks. Employing a feedforward compensation algorithm, it preemptively fine-tunes the servo valve opening to buffer pressure peaks when a sharp increase in thrust is detected. The cutterhead drive subsystem controller strengthens overload protection logic and power smoothing control to ensure a smooth transition when encountering extremely hard rock masses, avoiding stalling. The overall system control cycle, especially the control loops related to cutterhead torque and articulation adjustment, could be appropriately shortened to respond more quickly to rapidly changing load conditions that may occur in hard rock formations.

Claims

1. A real-time monitoring system for multi-point force on TBM shield shell and an adaptive adjustment system for preventing jamming, characterized in that, include: The multi-point stress sensing module is used to collect real-time stress data of the shield structure. The module contains an array of 256 fiber optic stress sensors arranged on the outer surface of the shield in the circumferential and axial directions. Each sensor synchronously measures the contact pressure value at a sampling frequency of 1000 Hz. The sensor array is connected to the data acquisition box through a corrosion-resistant armored optical cable. The data acquisition box integrates a signal demodulation unit and a temperature compensation unit. The data fusion and state reconstruction module is used to receive the raw stress data stream from the multi-point stress sensing module; The card machine risk dynamic assessment module is used to quantify risk based on the stress distribution cloud map sequence output by the data fusion and state reconstruction module; A multi-objective optimization decision-making module is used to respond to the risk index output by the card machine risk dynamic assessment module; The adaptive adjustment execution module is used to receive the tunneling parameter adjustment scheme issued by the multi-objective optimization decision module. This module includes a propulsion hydraulic subsystem controller, a cutterhead drive subsystem controller, a shield tail sealing subsystem controller, and an articulated hydraulic subsystem controller. Each controller adjusts the output of its corresponding actuator according to the adjustment command.

2. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 1, characterized in that, The multi-objective optimization decision module is a multi-objective optimization solver. Its optimization objective function simultaneously minimizes the jamming risk index, maximizes the tunneling speed, and minimizes the total energy consumption of the propulsion system. The decision variables of this module are four key tunneling parameters of the TBM: total propulsion force, cutterhead torque, shield tail gap adjustment, and articulated system deflection angle. The optimization process is constrained by the maximum output limit of each actuator, the strength limit of the shield structure, and the allowable range of tunnel design axis deviation. This module uses a non-dominated sorting genetic algorithm with an elite strategy to solve the multi-objective optimization problem.

3. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 1, characterized in that, The fiber optic grating sensor deployment strategy in the multi-point stress sensing module is optimized based on the mechanical characteristics of the shield structure. The sensors are deployed non-uniformly in the cut ring, support ring and tail ring of the shield. In the soil-facing area of ​​the cut ring where stress concentration is expected to be high, the sensor deployment density is twice that of other areas.

4. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 3, characterized in that, Each node of the fiber optic grating sensor is encapsulated in a special alloy protective housing and is reliably coupled to the shield material through a dual fixing method of magnetic attraction and welding.

5. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 1, characterized in that, The Kriging interpolation algorithm used in the data fusion and state reconstruction module has a variogram model that is customized based on the prior knowledge of the shield's geometry and typical surrounding rock load distribution.

6. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 5, characterized in that, The data fusion and state reconstruction module also integrates a self-learning mechanism, which can dynamically optimize the parameters of the interpolation algorithm based on the comparison between historical reconstruction data and subsequent actual stress verification results.

7. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 1, characterized in that, The risk quantification model of the jamming risk dynamic assessment module dynamically adjusts its weight coefficients through an online learning mechanism. The system is pre-built with a training database based on a large number of historical TBM tunneling cases. The module dynamically adjusts the weight distribution of the three input parameters in the risk quantification model according to the current tunneling stratum type and the jamming warning records.

8. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 2, characterized in that, The solution process of the multi-objective optimization decision module introduces a real-time feedback correction mechanism. This module continuously receives the actual parameters fed back by the adaptive adjustment execution module and the risk index updated by the card risk dynamic assessment module. If the risk index does not decrease by 80% of the expected target within two control cycles after the adjustment plan is implemented, the optimization solver immediately restarts the optimization calculation with the current machine state as the initial condition.

9. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 1, characterized in that, In the adaptive adjustment execution module, the propulsion hydraulic subsystem controller sets the pressure and flow rate of the main propulsion cylinder according to the adjustment scheme, the cutterhead drive subsystem controller adjusts the output torque and speed of the variable frequency motor, the shield tail sealing subsystem controller adjusts the shield tail grease injection pressure to change the gap between the shield shell and the segments, and the articulation hydraulic subsystem controller controls the extension and retraction of the articulation cylinder to achieve fine adjustment of the shield body attitude.

10. The TBM shield shell multi-point force real-time monitoring and anti-jamming adaptive adjustment system according to claim 1, characterized in that, The execution status and feedback signals of all control commands are collected in real time and sent back to the data fusion and status reconstruction module to form a closed-loop control circuit.

Images