Shield tunneling intelligent control method and system based on ground-tunnel-machine-information-man adaptation level

Abstract

The invention discloses a shield tunneling intelligent control method and system based on a ground-tunnel-machine-information-human adaptation level, and relates to the crossing field of underground engineering intelligent construction and information technology. Setting a first-level index and a second-level index; dividing the shield construction state into three levels of intelligent adaptation levels by combining the total score of the construction state with a dynamic threshold value; setting four targets of attitude intelligent control, settlement intelligent control, tunneling efficiency optimization and construction abnormity diagnosis; a machine learning prediction model is constructed, and the construction state parameters corresponding to the targets are predicted in real time; according to the shield intelligent adaptation level, the man-machine cooperation modes are divided into three classes of A level, B level and C level; and comparing actual data with a prediction result of the machine learning model, and dynamically correcting the control target parameters. Through real-time sensing of geological, mechanical and environmental multi-dimensional construction states and in combination with a dynamic weight distribution strategy, precise evaluation of the shield construction state is achieved, and meanwhile self-adaptive switching of the man-machine control right is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent construction of underground engineering and information technology, and in particular to a shield tunneling intelligent control method and system based on a ground-tunnel-machine-signal-person adaptive level. BACKGROUND

[0002] With the acceleration of urbanization, the scale of underground space development continues to expand, and the shield tunnel construction method has become a core technical means in the fields of urban rail transit, water conservancy engineering, etc. However, in the process of shield construction, it faces complex and changeable geological conditions, dynamic fluctuations of mechanical properties, and coupling effects of environmental risks. The traditional construction method has significant technical defects, which are as follows:

[0003] 1. The existing hierarchical control logic relies on a single index (such as thrust, torque), which cannot fully depict the multi-field coupling effect of geology-mechanical-environment, resulting in a large deviation between the construction state evaluation result and the actual working condition, making it difficult to support accurate decision-making.

[0004] 2. The fixed threshold is used to determine the construction risk, which cannot adapt to the nonlinear working condition changes caused by geological mutations (such as fault, karst cave exposure) or mechanical wear (such as tool wear, seal aging), and is prone to risk misjudgment or response lag problems.

[0005] 3. The traditional man-machine cooperation mechanism highly depends on the experience of operators, lacks an algorithm-driven dynamic cooperation mode, and is difficult to adjust the control right allocation according to real-time construction parameters, resulting in limited precision control and dynamic optimization capability in the construction process.

[0006] In view of the above problems, domestic and foreign scholars have carried out related research and proposed real-time monitoring technology based on sensor network and prediction model based on machine learning, but the existing technical solutions still have three key deficiencies: first, the multi-source data fusion capability is limited, which cannot realize the dynamic coupling analysis of geological parameters (such as stratum lithology, permeability coefficient), mechanical parameters (such as thrust efficiency, torque margin), and environmental parameters (such as ground settlement, segment stress); second, the control mode switching relies on preset fixed threshold, lacking self-adaptive adjustment capability for nonlinear changes in construction state; third, the man-machine cooperation mechanism is rigid, which cannot dynamically allocate man-machine control right according to real-time risk level, making it difficult to balance construction efficiency and safety risk.

[0007] In summary, there is an urgent need to build an intelligent adaptive level system based on multi-index dynamic coupling to realize real-time prediction and self-adaptive control of shield construction state, and to improve the efficiency and safety of shield tunnel construction under complex working conditions. SUMMARY

[0008] The purpose of the present application is to propose a shield tunneling intelligent control method and system based on ground-tunnel-machine-signal-person adaptation level, through real-time perception of multi-dimensional construction states of geology, machinery and environment, combined with dynamic weight distribution strategy, to realize accurate evaluation of shield construction state, and to achieve adaptive switching of human-machine control right.

[0009] According to a first aspect of the embodiments of the present disclosure, a shield tunneling intelligent control method based on ground-tunnel-machine-signal-person adaptation level is provided, comprising the following steps:

[0010] Taking geological adaptation degree G, mechanical adaptation degree M and environmental risk degree E as primary indicators, corresponding secondary indicators are set; through comprehensive scoring, the total score of the construction state is obtained, and the shield construction state is divided into three intelligent adaptation levels combined with dynamic threshold value;

[0011] Based on the above classification system, four targets of attitude intelligent control, settlement intelligent control, tunneling efficiency optimization and construction abnormality diagnosis are set; a machine learning prediction model is constructed to real-time predict the construction state parameters corresponding to each target;

[0012] According to the intelligent adaptation level of the shield, the human-machine collaborative mode is divided into three categories of A, B and C, and the human-machine permission allocation and operation process under different levels are clarified; wherein A is machine-assisted, B is human-machine collaboration, and C is human-assisted machine;

[0013] By real-time collection of geological, mechanical and environmental data, the actual data and the prediction results of the machine learning model are compared and analyzed, and the control target parameters are dynamically corrected.

[0014] According to a second aspect of the embodiments of the present disclosure, a shield tunneling intelligent control system based on ground-tunnel-machine-signal-person adaptation level is provided, comprising:

[0015] The intelligent adaptation level division module takes geological adaptation degree G, mechanical adaptation degree M and environmental risk degree E as primary indicators, and sets corresponding secondary indicators; through comprehensive scoring, the total score of the construction state is obtained, and the shield construction state is divided into three intelligent adaptation levels combined with dynamic threshold value;

[0016] The control target and prediction model module, based on the above classification system, sets four targets of attitude intelligent control, settlement intelligent control, tunneling efficiency optimization and construction abnormality diagnosis; a machine learning prediction model is constructed to real-time predict the construction state parameters corresponding to each target;

[0017] The human-machine collaborative mode matching module, according to the intelligent adaptation level of the shield, divides the human-machine collaborative mode into three categories of A, B and C, and clarifies the human-machine permission allocation and operation process under different levels; wherein A is machine-assisted, B is human-machine collaboration, and C is human-assisted machine;

[0018] The dynamic feedback correction module compares and analyzes the actual data and the prediction result of the machine learning model by collecting geological, mechanical and environmental data in real time, and dynamically corrects the control target parameters.

[0019] According to a third aspect of the embodiments of the present disclosure, an electronic device is provided, which comprises a memory, a processor and a computer program stored in the memory and run on the memory, and the processor implements the intelligent control method for shield tunneling based on the ground-tunnel-machine-signal-person adaptation level when executing the program.

[0020] According to a fourth aspect of the embodiments of the present disclosure, a computer readable storage medium is provided, which stores a computer program, and the program is executed by a processor to implement the intelligent control method for shield tunneling based on the ground-tunnel-machine-signal-person adaptation level.

[0021] Compared with the prior art, the above technical scheme of the present application has the following advantages: 1. The intelligent adaptation level grading system constructed by the present application takes geological adaptation degree (G), mechanical adaptation degree (M) and environmental risk degree (E) as the core first-level indicators, and is matched with secondary quantitative indicators such as stratum variation coefficient (G1), permeability mutation rate (G2), thrust efficiency coefficient (M1), cutterhead torque margin (M2), settlement gradient (E1) and segment stress exceeding rate (E2), which can accurately depict the coupling effect of geological-mechanical-environmental multi-physical field and solve the problem of one-sided evaluation caused by the traditional control relying on a single indicator (such as thrust and torque). At the same time, the system can dynamically adjust the indicator weight and the determination threshold according to the stratum type (such as soft soil and rock stratum) and the construction stage (such as initial tunneling, normal tunneling and receiving segment construction), establish an adaptive baseline through historical data training, update the risk level combined with real-time data deviation rate, replace the traditional fixed threshold mode, and significantly improve the adaptability to complex working conditions, thereby breaking through the limitation of poor universality of the traditional grading method. 2. The three-level collaborative mode of A (machine person auxiliary), B (man-machine collaboration) and C (man machine auxiliary) proposed by the present application can realize seamless switching from normal construction to emergency disposal through dynamic weight distribution and veto mechanism. In the I-level adaptation state of stable geology and efficient machinery, the A-level mode completes the tunneling parameter optimization and posture adjustment autonomously, reducing the lag of manual intervention; in the II-level critical state of local parameter anomaly, the B-level mode realizes rapid risk response through man-machine interaction correction; in the III-level risk state of stratum severe variation and sudden danger, the C-level mode triggers forced manual intervention to ensure construction safety. The mode breaks through the defects of traditional experience-dependent decision-making and low response efficiency in the mode dominated by manual intervention, significantly shortens the risk disposal time, and balances the construction efficiency and safety control.

[0022] 3.The application maps the construction state in real time through multi-source data (geological radar data, hydraulic sensor data, optical fiber monitoring data, etc.), sets an automatic trigger model retraining process every 10 rings of completed excavation, can timely fuse the latest working condition data to update the machine learning model parameters, ensures the continuous adaptability of the algorithm to complex working conditions such as geological mutations and mechanical wear, and avoids the prediction accuracy decay problem caused by the parameter solidification of the traditional model. At the same time, the multi-agent collaboration system is integrated, and the multi-objective optimization of attitude control, settlement control, efficiency optimization, and abnormal diagnosis is realized, solving the problem of disconnection between single target optimization and overall construction demand in traditional control, and improving the intelligent control level and comprehensive construction benefit of shield tunneling. BRIEF DESCRIPTION OF DRAWINGS

[0023] The accompanying drawings, which form a part of this specification, are included to provide a further understanding of the application, and are incorporated herein by reference. The illustrations are shown for the purpose of enabling those in the art to sufficiently understand the present application, and are not intended to limit the present application.

[0024] Figure 1 To adapt to the level classification flowchart;

[0025] Figure 2 To the CNN-LSTM combined model diagram;

[0026] Figure 3 To the construction data flow circulation closed loop diagram. DETAILED DESCRIPTION

[0027] The present disclosure will be further described with reference to the drawings and examples.

[0028] It should be noted that the following detailed description is merely exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0029] It should be noted that the terms used herein are only for the purpose of describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and furthermore, it should be understood that when the terms "comprise" and / or "include" are used in the specification, there is a feature, step, operation, device, component and / or combination thereof.

[0030] It should be noted that the flowcharts and block diagrams in the drawings illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of the present disclosure. It is noted that each block in the flowcharts or block diagrams can represent a module, a program segment, or a portion of code, which can include one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks noted in succession can in fact be executed substantially concurrently or can sometimes be executed in reverse order, depending on the functionality involved. It should also be noted that each block in the flowcharts and / or block diagrams and combinations of blocks in the flowcharts and / or block diagrams can be implemented by a dedicated hardware-based system that performs specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0031] Embodiment one:

[0032] The embodiment provides a shield tunneling intelligent control method based on a ground-tunnel-machine-signal-person adaptation level. The implementation process of the application is described in detail below in combination with a Jinan Yellow River large-diameter shield tunnel project (diameter 16.8 m). The intelligent control of the construction process is realized through multi-dimensional data fusion, dynamic scoring and closed-loop feedback of the geological mechanical environment. The specific steps are as follows:

[0033] Step one, taking the geological adaptation degree G, the mechanical adaptation degree M and the environmental risk degree E as the first-level indexes, corresponding second-level indexes are set; the total score of the construction state is obtained through comprehensive scoring, and the shield construction state is divided into three intelligent adaptation levels in combination with the dynamic threshold value;

[0034] Specifically, the geological fitness degree (G) is quantitatively characterized by its secondary indicators, the stratum variation coefficient (G1) and the permeability coefficient mutation rate (G2). The stratum variation coefficient (G1) is based on the energy entropy of the stratum parameters in front of the cutter head extracted by wavelet packet decomposition, and the deviation rate of the current entropy value from the historical baseline is calculated to reflect the geological uniformity. The permeability coefficient mutation rate (G2) is obtained by monitoring the time series data of the soil tank permeability coefficient, and the average of the mutation frequency and intensity per unit time is calculated to identify abnormal fluctuations in the stratum permeability. The mechanical fitness degree (M) is jointly evaluated by its secondary indicators, the thrust efficiency coefficient (M1) and the cutter head torque margin (M2). The thrust efficiency coefficient (M1) couples the actual thrust and the tunneling speed to measure the mechanical energy efficiency conversion level. The cutter head torque margin (M2) introduces a time decay factor to dynamically compensate for the impact of tool wear on the torque threshold, ensuring the timeliness of the mechanical health degree evaluation. The environmental risk degree (E) is comprehensively determined by its secondary indicators, the settlement gradient (E1) and the pipe stress exceeding rate (E2). The settlement gradient (E1) is integrated along the tunnel axis to capture the trend of continuous spatial deformation. The pipe stress exceeding rate (E2) uses a weighted statistical method to quantify the spatial distribution and exceeding degree of the stress points, and to strengthen the sensitivity of risk warning.

[0035] The comprehensive score adopts a normalized weighting method. After mapping the geological, mechanical, and environmental indicators to a unified dimension, the sum is calculated according to the preset weights (e.g., 40% for geology, 30% for machinery, and 30% for environment), and the construction level is divided in combination with a dynamic threshold. When the geological parameters mutate or the mechanical margin is below the safety red line, a one-vote veto mechanism is triggered, directly determining the high-risk level.

[0036] As shown in Figure 1 The following multi-source data is collected in real time:

[0037] 1) Geological data: stratum elastic modulus, permeability coefficient, and pore water pressure detected by geological radar;

[0038] 2) Mechanical data: cutter head thrust, torque, and propulsion cylinder pressure obtained by hydraulic sensors;

[0039] 3) Environmental data: surface settlement value monitored by static leveling instruments and pipe stress measured by fiber Bragg grating sensors.

[0040] After data cleaning to remove outliers, the geological, mechanical, and environmental data are matched according to the tunneling ring number (1.5 m per ring) and the timestamp. Then, the geological complexity (G1), fault stability (G2), thrust efficiency coefficient (M1), cutter head torque margin (M2), settlement gradient (E1), and groundwater depth (E2) are mapped to the interval [0, 100] according to the following formula:

[0041]

[0042] The secondary index is obtained, and specifically as follows:

[0043] The geological complexity (G1) is as follows:

[0044]

[0045] wherein S i is a sub-item score (rock type, hardness, karst rate, and development degree of adverse geological action, etc.); ω i is a weight coefficient (adjusted according to engineering practice, such as karst rate weight > 30%).

[0046] The fault stability (G2) is as follows:

[0047]

[0048] wherein D is the fault width (m); γ is the groundwater permeability coefficient (m / s); and α is the safety factor (K), which is determined according to the compressive strength of the surrounding rock and the support design (generally taken as 1.15-1.30).

[0049] The thrust efficiency coefficient (M1) is as follows:

[0050]

[0051] The thrust is decomposed into a rock-breaking force F5 and friction forces F1-F4, which are shield shell friction, earth pressure balance, and other non-effective thrusts.

[0052] The cutter head torque margin (M2) is as follows:

[0053]

[0054] wherein r0 is the installation radius of the roller cutter; T is the cutter head torque; the specific torque index is combined with the margin; h is the penetration depth; and T0 is the rated torque.

[0055] The settlement gradient (E1) is as follows:

[0056]

[0057] The groundwater depth (E2) is as follows:

[0058] E2 = h (m)

[0059] wherein h is the vertical distance from the ground surface to the top surface of the phreatic zone.

[0060] Specifically, the shield construction state is divided into three intelligent adaptation levels by dynamic threshold and dynamic baseline. The dynamic baseline is an adaptive reference standard based on real-time construction parameters, geological evolution characteristics and historical data rules. The historical data mean is calculated using the sliding window mechanism and variance σ 2 , to get the dynamic baseline S:

[0061]

[0062] The comprehensive score, i.e. relative degree deviation, is evaluated. The Mahalanobis Distance is used to quantify the comprehensive deviation of parameters. The Mahalanobis Distance of the sample distribution with a mean of vector X (thrust, torque, settlement gradient, grouting amount, etc.) and covariance S is as follows:

[0063]

[0064] The shield construction state is divided into three intelligent adaptation levels by comprehensive score and dynamic threshold, as follows:

[0065] Level I (intelligent adaptation): the comprehensive score deviates significantly positively from the baseline, and the single indicators are within the safety threshold.

[0066] Level II (critical adaptation): the comprehensive score is close to the baseline or the single indicators exceed the limit (reach the warning threshold).

[0067] Level III (risk adaptation): the comprehensive score deviates significantly negatively from the acceptable range or triggers a veto item (reaches the danger threshold).

[0068] The indicator threshold is referenced in the following table:

[0069] Indicator Safety threshold Warning threshold Danger threshold G1 ≤ 2% (no large cavities) 2% - 5% (locally broken) ≥ 5% (large faults / cavities) G2 ≤10 10-30 ≥30 M1 ≥0.60 0.40-0.60 ≤0.40 M2 ≥0.30 0.10-0.30 ≤0.10 E1 ≤0.5 0.5-1.0 ≥1.0 E2 ≥5.0 2.0-5.0 ≤2.0

[0070] Note: The table is for reference only. Specifically, the indicator weights are dynamically adjusted according to the stratum type and construction stage to ensure the scientificity and adaptability of the grading system.

[0071] Step two, based on the above grading system, set four goals of attitude intelligent control, settlement intelligent control, tunneling efficiency optimization and construction anomaly diagnosis; build a machine learning prediction model to predict the construction state parameters corresponding to each goal in real time;

[0072] The attitude intelligent control takes the axis deviation Δθ ≤ ± 15 mm as the constraint target, and uses a machine learning model to predict the future 5-ring attitude of the shield machine. The model inputs geological parameters and mechanical parameters, and trains the time series dependence relationship combined with historical attitude data; after completing 1 ring of tunneling, the predicted attitude is compared with the actual deviation, and the push cylinder pressure difference is dynamically adjusted to stabilize the attitude deviation Δθ within ± 5 mm / m.

[0073] The settlement intelligent control takes the ground surface settlement amount S≤10mm as the core target, and constructs a machine learning prediction model. The model inputs geological parameters (porosity, water content) and tunneling parameters (thrust, grouting pressure, tunneling speed), and outputs the future 5-ring settlement distribution; when the predicted value approaches the threshold value, the grouting pressure is automatically increased and the tunneling speed is reduced to suppress ground deformation.

[0074] The tunneling efficiency optimization uses a reinforcement learning algorithm to dynamically adjust the tunneling parameters. The agent takes the geological risk level, mechanical efficiency, and environmental indicators as the state space, and the thrust, cutter head speed, and grouting pressure as the action space. Every 5 rings, the strategy is updated according to real-time energy consumption and ground loss data, achieving maximum efficiency of tunneling speed.

[0075] The construction anomaly diagnosis fuses geological, mechanical, and environmental multi-source data to monitor the construction state in real time. The model analyzes the characteristic patterns of anomalies such as cutter head jamming and seal leakage through association rule mining. When an abnormal signal is identified, a hierarchical warning is triggered and the root cause is located. Further, according to the type of anomaly, a targeted adjustment scheme is pushed to ensure that the anomaly recognition rate is ≥90% and the response time is shortened to within 30 seconds.

[0076] Taking tunneling attitude control as an example, a CNN-LSTM combined model is used to predict the tunneling attitude. The data set consists of shield machine thrust, propulsion speed, cutter torque, earth pressure parameters, shield head horizontal deviation, shield head vertical deviation, shield tail horizontal deviation, and shield tail vertical deviation. The CNN-LSTM combined model is shown in Figure 2 The tunneling data of t-D ring (shield machine thrust, propulsion speed, cutter torque, and earth pressure parameters) are input, and the attitude parameters of t+1 ring (shield head horizontal deviation, shield head vertical deviation, shield tail horizontal deviation, and shield tail vertical deviation) are output. When the tunneling parameters are input into the model, CNN will automatically extract adaptive features from the data, thereby avoiding errors caused by manual feature extraction. Then, LSTM will further extract the time series correlation in the original data, further improving the prediction accuracy. Finally, the output of LSTM enters the fully connected layer and is activated by the Sigmoid function to obtain the predicted shield attitude data.

[0077] Step three, according to the shield intelligent adaptation level, the man-machine cooperation mode is divided into three categories: A, B, and C, and the man-machine permission allocation and operation process under different levels are clarified; among them, A is machine-assisted, B is man-machine cooperation, and C is man-machine assistance;

[0078] The machine master (A-level) mode is applicable to a first-level intelligent adaptive state with stable geology, high mechanical efficiency and low environmental risk. In this mode, the shield machine autonomously performs excavation parameter optimization and posture adjustment, and the human role focuses on data monitoring and model iteration; the system feeds back sensor data to the digital twin platform in real time, the operator regularly reviews key parameters, and optimizes the machine learning prediction model through historical data training to improve the algorithm generalization ability.

[0079] The man-machine cooperation mode (B-level) is for a second-level critical adaptive state with local parameter abnormalities. When the system detects index fluctuations, it triggers an audible and light warning and pushes adjustment suggestions, and the human needs to confirm the risk before making a joint decision; the operator can modify the control parameters based on the machine's suggestions, the system dynamically executes the correction instructions and continuously monitors the feedback until the parameters return to the safety threshold.

[0080] The man-machine auxiliary mode (C-level) is used for a third-level risk adaptive state with severe stratum variation and continuously over-standard segment stress or sudden major risk events. The system immediately alarms and locks the automatic control permission, and the human takes over to analyze the risk source by calling multi-source data; the operator develops a risk elimination plan, and the system assists in executing fine operations until the risk is eliminated.

[0081] The collaborative mode is dynamically adjusted according to real-time scoring and hard constraints. When all indicators return to the first level and stable operation for more than a certain period of time (such as 5 minutes), it automatically switches to the machine master mode; if any indicator enters the second level or the model prediction is inaccurate, it is downgraded to the man-machine cooperation mode; when there is a geological mutation or a sudden event such as water inrush and collapse, it is forcibly upgraded to the man-machine auxiliary mode. Through the above hierarchical response mechanism, seamless connection from normal optimization to emergency intervention is realized, ensuring construction safety and continuity under complex conditions.

[0082] For example, when it is determined that the shield machine is in a first-level intelligent adaptive state, the model is retrained based on the latest data after every 10 rings of excavation to ensure prediction accuracy and ensure that the prediction error SSIM is greater than or equal to 0.90.

[0083] The man-machine collaborative mode adopts an A-level machine-man auxiliary mode, the shield machine autonomously controls the tunneling parameters, and the digital twin platform displays data such as cutter torque, grouting pressure, and settlement gradient in real time. The operator reviews the data every 30 minutes, and the model is iteratively updated every 2 hours. When the shield machine is in a Ⅱ-level critical adaptive state, the man-machine collaborative mode adopts a B-level man-machine collaborative mode, for example, the system detects a sudden change in the permeability coefficient of the silt layer and pushes a “raise the grouting amount” suggestion. After manual confirmation, the grouting pressure is increased, and the tunneling speed is decreased. The system continuously monitors the parameters within 2 rings until it falls within the safety threshold. When the shield machine is in a Ⅲ-level risk adaptive state, the man-machine collaborative mode adopts a C-level machine-man auxiliary mode, at which time the system locks the automatic control authority and triggers an audible and light alarm. The manual retrieves the geological radar scanning data, confirms the water gushing position, and executes high-pressure grouting and cutter speed reduction to ensure that the permeability coefficient is restored within 30 minutes.

[0084] Step four, by real-time collection of geological, mechanical and environmental data, the actual data is compared and analyzed with the prediction results of the machine learning model, and the control target parameters are dynamically corrected.

[0085] Specifically, data collection is achieved by integrating geological radar, hydraulic sensors and optical fiber monitoring equipment to collect geological data, mechanical data and environmental data in real time. After standardization processing, the data is synchronously mapped to the three-dimensional model, realizing the global visualization of the construction state and the positioning of abnormal hotspots.

[0086] After each 1 -ring tunneling is completed, the predicted value is compared with the actual construction data to obtain the mean square error (MSE) and the structural similarity (SSIM) to evaluate the model accuracy. When the prediction error is > 15% or the environmental risk level jumps, the model retraining process is automatically started, the latest working condition data is fused to update the network weight parameters, and the adaptability is ensured. Through the above closed-loop feedback mechanism, the whole-link intelligentization from data perception, prediction and warning to control optimization is realized, and the stability and risk resistance of complex stratum construction are significantly improved.

[0087] Figure 3 To circulate the construction data, the data transmission direction of the above prediction process is summarized. Specifically, taking shield tunneling posture prediction as an example, the data required for tunneling posture prediction is monitored in real time, and then the above data is preprocessed; at the same time, the CNN-LSTM model is trained for prediction, and the model is continuously optimized and updated; according to the posture prediction result, the posture correction is carried out and the construction suggestion is given; then the command is issued to the hydraulic system, the pressure of the push cylinder is adjusted, and the cutter speed is synchronously fine-tuned; the shield machine continues construction according to the command and feeds back the actual posture parameters, energy consumption and equipment state to the digital twin platform in real time, completing the feedback loop.

[0088] Embodiment two:

[0089] The embodiment provides a shield tunneling intelligent control system based on a ground-tunnel-machine-signal-person adaptation level, which comprises the following steps:

[0090] An intelligent adaptation level division module, taking a geological adaptation degree G, a mechanical adaptation degree M and an environmental risk degree E as first indexes, and setting corresponding second indexes in a matched manner; obtaining a total score of a construction state through comprehensive scoring, and dividing the shield construction state into three intelligent adaptation levels in combination with a dynamic threshold value;

[0091] A control target and prediction model module, which sets four targets of posture intelligent control, settlement intelligent control, tunneling efficiency optimization and construction abnormality diagnosis based on the above classification system, and constructs a machine learning prediction model to perform real-time prediction on construction state parameters corresponding to each target;

[0092] A man-machine collaborative mode matching module, which divides man-machine collaborative modes into three categories of A, B and C according to the shield intelligent adaptation level, and clearly defines man-machine permission allocation and operation processes under different levels; wherein the A level is machine-assisted, the B level is man-machine collaboration, and the C level is man-machine-assisted;

[0093] A dynamic feedback correction module, which compares and analyzes actual data and prediction results of the machine learning model by collecting geological, mechanical and environmental data in real time, and dynamically corrects control target parameters.

[0094] Embodiment three

[0095] An electronic device comprising a memory, a processor and a computer program stored on the memory, wherein the processor implements the above shield tunneling intelligent control method based on a ground-tunnel-machine-signal-person adaptation level when executing the program, which comprises the following steps:

[0096] Taking a geological adaptation degree G, a mechanical adaptation degree M and an environmental risk degree E as first indexes, and setting corresponding second indexes in a matched manner; obtaining a total score of a construction state through comprehensive scoring, and dividing the shield construction state into three intelligent adaptation levels in combination with a dynamic threshold value;

[0097] Setting four targets of posture intelligent control, settlement intelligent control, tunneling efficiency optimization and construction abnormality diagnosis based on the above classification system, and constructing a machine learning prediction model to perform real-time prediction on construction state parameters corresponding to each target;

[0098] Dividing man-machine collaborative modes into three categories of A, B and C according to the shield intelligent adaptation level, and clearly defining man-machine permission allocation and operation processes under different levels; wherein the A level is machine-assisted, the B level is man-machine collaboration, and the C level is man-machine-assisted;

[0099] Comparing and analyzing actual data and prediction results of the machine learning model by collecting geological, mechanical and environmental data in real time, and dynamically correcting control target parameters.

[0100] Embodiment four:

[0101] A computer readable storage medium, having stored thereon a computer program, which, when executed by a processor, implements the above-mentioned shield tunneling intelligent control method based on the ground-tunnel-machine-signal-person adaptation level, comprising:

[0102] Taking the geological adaptation degree G, the mechanical adaptation degree M and the environmental risk degree E as the first-level indexes, corresponding second-level indexes are set; the total score of the construction state is obtained through comprehensive scoring, and the shield construction state is divided into three intelligent adaptation levels in combination with the dynamic threshold value;

[0103] Based on the above classification system, four targets of posture intelligent control, settlement intelligent control, tunneling efficiency optimization and construction anomaly diagnosis are set; a machine learning prediction model is constructed to predict the construction state parameters corresponding to each target in real time;

[0104] According to the intelligent adaptation level of the shield, the man-machine collaborative mode is divided into three categories of A, B and C, and the man-machine permission allocation and operation process under different levels are clarified; the A level is machine-assisted, the B level is man-machine collaboration, and the C level is man-machine assistance;

[0105] By collecting geological, mechanical and environmental data in real time, the actual data and the prediction results of the machine learning model are compared and analyzed, and the control target parameters are dynamically corrected.

[0106] Those skilled in the art should understand that each module or each step of the above-mentioned disclosure can be realized by a general computer device, and alternatively, they can be realized by program codes executable by a computing device, so that they can be stored in a storage device and executed by a computing device, or they can be respectively manufactured into each integrated circuit module, or a plurality of modules or steps among them can be manufactured into a single integrated circuit module to realize. The present disclosure is not limited to any specific combination of hardware and software.

[0107] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0108] The above describes the specific embodiments of the present disclosure in combination with the drawings, but is not a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present disclosure without creative labor are still within the protection scope of the present disclosure.

Claims

1. A shield tunneling intelligent control method based on ground-tunnel-machine-information-human adaptation level, characterized in that, Includes the following steps: Geological compatibility (G), mechanical compatibility (M), and environmental risk level (E) are used as primary indicators, and corresponding secondary indicators are set up. The total score of the construction status is obtained through comprehensive scoring, and the shield tunneling construction status is divided into three levels of intelligent compatibility based on dynamic thresholds. Based on the above hierarchical system, four objectives are set: intelligent attitude control, intelligent settlement control, tunneling efficiency optimization, and construction anomaly diagnosis; a machine learning prediction model is constructed to predict the construction state parameters corresponding to each objective in real time. Based on the shield tunneling machine's intelligent adaptation level, the human-machine collaboration mode is divided into three categories: Level A, Level B, and Level C, and the allocation of human-machine permissions and operation procedures under different levels are clearly defined; among them, Level A is machine master-assistant, Level B is human-machine collaboration, and Level C is human master-assistant. By collecting geological, mechanical, and environmental data in real time, the actual data is compared and analyzed with the prediction results of machine learning models to dynamically adjust the control target parameters.

2. The intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in claim 1, characterized in that, The secondary indicators of geological compatibility include the formation variation coefficient and the permeability coefficient mutation rate; the secondary indicators of mechanical compatibility include the thrust efficiency coefficient and the cutterhead torque margin; and the secondary indicators of environmental risk include the settlement gradient and the segment stress exceedance rate.

3. The intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in claim 2, characterized in that, The formation variation coefficient G1 is based on wavelet packet decomposition to extract the energy entropy of the formation parameters in front of the cutterhead, obtaining the deviation rate between the current entropy value and the historical baseline, reflecting the geological homogeneity; the permeability coefficient mutation rate G2 is obtained by monitoring the time series data of the soil chamber permeability coefficient, statistically analyzing the number of mutations and the average intensity per unit time, and identifying abnormal fluctuations in formation permeability; the thrust efficiency coefficient M1 is coupled with the actual thrust and tunneling speed to measure the level of mechanical energy efficiency conversion; the cutterhead torque margin M2 introduces a time decay factor to dynamically compensate for the influence of cutter wear on the torque threshold, ensuring the timeliness of the mechanical health assessment; the settlement gradient E1 is integrated along the tunnel axis to capture the continuous spatial deformation trend; The stress exceedance rate E2 of the tunnel segments is obtained by using a weighted statistical method to quantify the spatial distribution and degree of exceedance of stress points, thereby enhancing the sensitivity of risk warning.

4. The intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in claim 1, characterized in that, The comprehensive scoring adopts a normalized weighted method, which maps geological, mechanical, and environmental indicators to a unified dimension, sums them according to preset weights, and classifies the construction level in combination with dynamic thresholds; when geological parameters change abruptly or the mechanical margin is lower than the safety red line, a veto mechanism is triggered, and the level is directly judged as high risk.

5. The intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in claim 1, characterized in that, The intelligent attitude control uses the axis deviation Δθ≤±15mm as the constraint target and dynamically adjusts the pressure difference of the propulsion cylinder. The intelligent settlement control system takes a surface settlement of S≤10mm as its core objective and is used to adjust grouting pressure and tunneling speed to suppress stratum deformation. The tunneling efficiency optimization adopts a reinforcement learning algorithm to dynamically adjust tunneling parameters. It uses geological risk level, mechanical efficiency, and environmental indicators as the state space, and thrust, cutterhead speed, and grouting pressure as the action space. It updates the strategy based on real-time energy consumption and formation loss data to maximize tunneling speed and efficiency. The construction anomaly diagnosis integrates real-time monitoring of construction status using multi-source data from geology, machinery, and environment.

6. The intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in claim 1, characterized in that, Level A mode is suitable for Level I intelligent adaptation state, Level B mode is for Level II critical adaptation state, and Level C mode is used for Level III risk adaptation state or sudden major risk events. In the Class A mode, the tunnel boring machine autonomously optimizes tunneling parameters and adjusts its attitude, while manual monitoring focuses on data monitoring and model iteration. In the Class B mode, when fluctuations in indicators are detected, an audible and visual warning is triggered and adjustment suggestions are pushed out. After manual confirmation of the risk, a joint decision is made. In the Class C mode, an alarm is immediately triggered to stop the machine and lock the automatic control authority. After manual takeover, the root cause of the danger is analyzed and a risk mitigation plan is formulated.

7. The intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in claim 1, characterized in that, Geological, mechanical, and environmental data are standardized and synchronously mapped onto a 3D model to achieve global visualization of the construction status and location of abnormal hotspots. When all indicators return to Level I and operate stably for a period of time, the system automatically switches to Level A mode. If any indicator enters Level II or the model prediction fails, the system is downgraded to Level B mode. In the event of geological changes or sudden events such as water inrush or landslides, the system is forcibly upgraded to Level C mode.

8. A shield tunneling intelligent control system based on ground-tunnel-machine-information-human adaptation level, characterized in that, include: The intelligent adaptation level classification module uses geological adaptation degree (G), mechanical adaptation degree (M), and environmental risk degree (E) as primary indicators, and sets corresponding secondary indicators. The total score of the construction status is obtained through comprehensive scoring, and the shield tunneling construction status is classified into three levels of intelligent adaptation level by combining dynamic thresholds. The control target and prediction model module, based on the above hierarchical system, sets four targets: intelligent attitude control, intelligent settlement control, tunneling efficiency optimization, and construction anomaly diagnosis; and constructs a machine learning prediction model to predict the construction state parameters corresponding to each target in real time. The human-machine collaboration mode matching module divides the human-machine collaboration mode into three categories: Level A, Level B, and Level C, based on the shield tunneling intelligent adaptation level, and clarifies the human-machine permission allocation and operation process under different levels; among them, Level A is master-assistant, Level B is human-machine collaboration, and Level C is master-assistant. The dynamic feedback correction module collects geological, mechanical, and environmental data in real time, compares and analyzes the actual data with the prediction results of the machine learning model, and dynamically corrects the control target parameters.

9. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and running thereon, characterized in that, When the processor executes the program, it implements the intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the intelligent control method for shield tunneling based on the adaptation level of ground-tunnel-machine-information-human as described in any one of claims 1-7.

Images