Shield tunnel spoil improvement parameter prediction method and system thereof

Abstract

The application discloses a shield tunnel spoil improvement parameter prediction method and system, relates to the field of tunnel engineering construction technology, and comprises the following steps: multiple mode sensors are arranged at key positions of a shield machine, tunneling parameters, stratum parameters, multiple mode sensor signals, spoil physical properties and improvement agent injection parameters are collected; then, change characteristics are extracted, a mutation sensitive factor is calculated and working condition determination is performed; a normal state prediction sub-model library and a mutation rapid response model are constructed, a stratum model adaptation index is calculated and model adaptability determination is performed; a working condition coupling prediction model is established by using a graph neural network, nonlinear coupling characteristics are extracted, a working condition coupling index is calculated and prediction stability is determined; actual construction performance data are collected, model prediction results are compared, a prediction correction index is calculated and deviation analysis is performed, and corresponding correction strategies are triggered. The method can realize intelligent prediction and dynamic correction of shield tunnel spoil improvement parameters, and improve the safety and stability of the construction process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel engineering construction, in particular to a shield tunnel spoil improvement parameter prediction method and system. BACKGROUND

[0002] Shield tunnel construction technology has become one of the main construction methods for urban rail transit, underground pipe gallery and highway tunnel construction due to its small disturbance to the ground surface, fast construction speed and high safety. In the shield construction process, spoil improvement, as an important link to ensure construction efficiency, construction safety and reduce ground damage, is of great significance. The spoil improvement process involves multiple complex links such as improvement agent injection amount, tunneling parameter adjustment, ground property determination, etc., which directly affect the advancing efficiency, cutter wear, energy consumption and ground stability of shield construction.

[0003] In existing shield construction, the determination of spoil improvement parameters relies on experience rules or single sensor data analysis, which has the following main problems:

[0004] Data collection is not comprehensive. Traditional methods often rely on single type of sensor data, such as relying only on torque, thrust or water content for judgment, ignoring the coupling relationship between the multi-modal characteristics of the ground and the tunneling parameters, resulting in insufficient adaptability to complex working conditions.

[0005] Insufficient feature extraction. In existing technology, data such as spoil water content, torque, and soil pressure are mostly based on original values or simple statistical analysis, lacking in-depth analysis based on time series change characteristics and mutation detection, making it difficult to accurately judge ground mutations and construction risks.

[0006] Single prediction model with poor adaptability. Existing methods usually use a single prediction model, lack of differentiation and rapid response mechanism for normal and abnormal working conditions, and cannot dynamically adjust the prediction results, which can easily lead to increased prediction bias.

[0007] Lack of model adaptability evaluation. In current shield construction improvement parameter prediction, there is a lack of effective model adaptability evaluation method, which cannot judge the reliability of the prediction model under specific working conditions in real time, and it is difficult to realize model self-adaptive correction.

[0008] Coupling relationship is not fully utilized. In the process of shield tunneling, there is a complex nonlinear coupling relationship between tunneling parameters, ground parameters and spoil physical properties, and existing technology has not fully utilized this coupling feature for prediction optimization.

[0009] Therefore, the prior art urgently needs a shield tunnel spoil improvement parameter prediction method capable of comprehensively collecting tunneling parameters, stratum parameters, multi-modal signals and improvement agent injection parameters, and combining time sequence feature extraction, model adaptability evaluation, working condition coupling analysis and online adaptive correction, so as to improve prediction accuracy, enhance model adaptability, improve construction efficiency and reduce construction risk. SUMMARY

[0010] In view of the deficiencies of the prior art, the present application provides a shield tunnel spoil improvement parameter prediction method and system to solve the problems mentioned in the background art.

[0011] To achieve the above object, the present application is implemented by the following technical solution: a shield tunnel spoil improvement parameter prediction method, comprising the following steps:

[0012] Step one, by laying multiple sensors at key positions of the shield machine, real-time collection of cutterhead torque, thrust force, cabin earth pressure, spoil moisture content, stratum dry density, spectrum and resistivity signals, spoil particle size distribution, spoil liquid-plastic limit parameters and improvement agent injection parameters;

[0013] Step two, based on spoil moisture content, cutterhead torque and resistivity signals, the change rate, acceleration term and mutation quantity are extracted by time sequence difference, sliding window regression, filter differentiation and anomaly detection method, and the mutation sensitive factor is calculated and compared with the stratum mutation judgment threshold mth to determine whether the current working condition is normal or stratum mutation;

[0014] Step three, a convolutional neural network normal prediction sub-model library trained based on normal working condition data is used for prediction of different stratum categories; a mutation rapid response model constructed by combining convolutional neural network and recurrent neural network is used for emergency prediction of mutation working conditions, a stratum model adaptation index CMI is calculated through consistency of historical data fitting and stratum identification, and compared with the stratum model adaptation threshold Cth to determine whether the current model is adapted, and if not, the first strategy is given;

[0015] Step four, a working condition correlation atlas is constructed based on tunneling parameters, stratum parameters and improvement agent addition amount, a working condition coupling prediction model is established using a graph neural network, key nonlinear coupling features are extracted, a working condition coupling index GCI is calculated, and compared with the working condition coupling threshold Gth to determine whether the prediction is stable, and if not, the second strategy is given;

[0016] Step five, the actual performance data of construction is collected and compared with the predicted value of the working condition coupling prediction model, a prediction correction index AEI is calculated, and compared with the prediction correction threshold Ath to determine whether the prediction deviation of the working condition coupling prediction model is within a reasonable range, and if not, the third strategy is given.

[0017] Preferably, the step one comprises:

[0018] S11, collecting cutter torque data Tp by installing a torque sensor on the main drive system of the shield machine; collecting propulsion thrust data F by installing a thrust sensor on the propulsion cylinder of the shield machine; collecting soil pressure data P in the soil pressure chamber by arranging soil pressure sensors on the inner wall of the soil pressure chamber; and establishing a tunneling parameter data set;

[0019] S12, collecting muck water content data Wt by setting a near-infrared spectrum sensor at the muck sampling port; obtaining stratum dry density data Gd by calling a stratum database at a construction log collection unit; and establishing a stratum parameter data set;

[0020] S13, collecting stratum spectrum response signal data Snir by the near-infrared spectrum sensor; collecting stratum resistivity signal data Sres by the resistivity probe; and establishing a multi-modal sensor signal data set;

[0021] S14, collecting muck particle size distribution data Dp by arranging a micro-particle online analysis device at the outlet of the soil pressure chamber; collecting muck liquid-plastic limit parameter data Lp by arranging a micro-particle online analysis device at the outlet of the soil pressure chamber; and establishing a muck physical property data set;

[0022] S15, collecting real-time injection parameters of the modifier, including flow rate, pressure, concentration and injection duration, by arranging flow rate sensors, pressure sensors and concentration detection devices on the injection device, to form a modifier injection parameter set.

[0023] Preferably, the step two comprises:

[0024] S21, based on the muck water content data Wt, using time series difference and sliding window regression technology to calculate the first derivative of the water content change in the continuous time series, thereby obtaining the water content change rate ; based on the cutter torque data Tp, using time series filtering and numerical differentiation method to first smooth and denoise the torque sequence, and then calculate the second derivative to obtain the acceleration term of the torque ; based on the resistivity signal data Sres, using anomaly detection and time series mutation analysis method to compare the difference of adjacent time period resistivity signals, identify significant electrical property mutations, correct the effectiveness of the resistivity signal mutations combined with the reflectivity offset of the spectrum response signal data Snir, and obtain the final resistivity mutation .

[0025] Preferably, the step two further comprises:

[0026] S22, obtaining the water content change rate , acceleration term of torque , and resistivity mutation variable , after dimensionless processing, the mutation sensitive factor is calculated ;

[0027] S23, a preset formation mutation determination threshold mth is determined, and the mutation sensitive factor is compared and analyzed with the formation mutation determination threshold mth to obtain a first evaluation result, including:

[0028] When the mutation sensitive factor ≤ the formation mutation determination threshold mth, it is determined that the current working condition is a normal working condition;

[0029] When the mutation sensitive factor > the formation mutation determination threshold mth, it is determined that the current working condition is a formation mutation working condition.

[0030] Preferably, the step three includes:

[0031] S31, a convolutional neural network is used to construct a convolutional neural network initial model, and the cutter torque data Tp, the spoil moisture content data Wt, the resistivity signal data Sres, the spectral response signal data Snir, and the corresponding formation category and modifier injection parameters collected under the normal working condition are used as training features to train and test the convolutional neural network initial model; the trained convolutional neural network initial model is used as a normal prediction sub-model library, wherein each sub-model corresponds to a formation category; during operation, according to the output result of the formation identification module, the prediction sub-model corresponding to the current formation category is called, and a prediction result is output;

[0032] S32, for the mutation working condition sample data, a convolutional neural network combined with a recurrent neural network is used to construct a mutation rapid response initial model, and the change rate , acceleration term of torque , and resistivity mutation variable of the moisture content under the mutation working condition are used as training features to incrementally train and test the model; the trained model is used as a mutation rapid response prediction model, which automatically switches to call the mutation rapid response model to generate an emergency prediction result when the working condition is determined to be a mutation working condition;

[0033] S33, historical construction data is collected and acquired, and the goodness of fit of the normal prediction sub-model library and the mutation rapid response model is evaluated, and the residual sum of squares is compared with the actual collected data to obtain the goodness of fit index Rmodel of each model;

[0034] S34, based on the real-time formation identification result, extracting the probability distribution after feature fusion and clustering discrimination, obtaining the consistency probability Pmatch between the formation identification result and the called prediction model category.

[0035] Preferably, the step three further comprises:

[0036] S35, by obtaining the fitting goodness index Rmodel of each model and the consistency probability Pmatch between the formation identification result and the called prediction model category, after non-dimensional processing, the formation model adaptation index CMI is calculated and obtained;

[0037] S36, by presetting the formation model adaptation threshold Cth, and comparing and analyzing the formation model adaptation index CMI with the formation model adaptation threshold Cth, the second evaluation result is obtained, including:

[0038] When the formation model adaptation index CMI is greater than or equal to the formation model adaptation threshold Cth, it indicates that the current model is adapted, and the prediction result output by the current model is maintained;

[0039] When the formation model adaptation index CMI is less than the formation model adaptation threshold Cth, it indicates that the current model is not adapted, and the first warning instruction is triggered, and the first strategy is generated: adopting a multi-model weighting mechanism, calling a backup model pre-trained in different scenarios, outputting a temporary prediction result through a weighted average method, so that the result will not be completely invalid; calling the emergency improver injection reference parameters under the corresponding formation category from the database as the protective compensation input of the model output, ensuring that even if the model is not adapted, reasonable injection control basis can still be provided; dynamically correcting the prediction result, differentiating the comparison between the model prediction result and the emergency reference value, and correcting the prediction result to form a temporarily corrected prediction output; while correcting the prediction, mark the current working condition as "model misfit state", and write the key parameters in the correction process, including CMI value, corrected prediction value and reference value calling condition, into the log.

[0040] Preferably, the step four comprises:

[0041] S41, based on the cutter torque data Tp, the thrust force data F and the cabin earth pressure data P of the tunneling parameter group, the muck water content data Wt of the formation parameter data group, the muck particle size distribution data Dp and the muck liquid-plastic limit parameter data Lp of the muck physical property data group, and combined with the actual improver addition amount, a working condition correlation atlas is established; based on the graph neural network GNN, a working condition coupling initial model is constructed, and the tunneling parameters, formation parameters and improver addition amount are used as input nodes to train and test the working condition coupling initial model; the trained working condition coupling initial model is used as a working condition coupling prediction model for extracting nonlinear coupling features in runtime, including the correlation between the cutter torque data and the muck water content data correlation between the propulsive force data and the stratum pressure distribution data correlation between the in-cabin earth pressure data and the muck water permeability data .

[0042] Preferably, the step four further comprises:

[0043] S42, correlation between the cutterhead torque data and the muck water content data extracted by the working condition coupling prediction model correlation between the propulsive force data and the stratum pressure distribution data correlation between the in-cabin earth pressure data and the muck water permeability data After non-dimensional processing, the working condition coupling index GCI is calculated and obtained;

[0044] S43, by comparing the working condition coupling index GCI with the preset working condition coupling threshold Gth, a third evaluation result is obtained, including:

[0045] When the working condition coupling index GCI is greater than or equal to the working condition coupling threshold Gth, it is determined that the prediction is stable, the working condition coupling prediction model output result is maintained, and the muck improvement parameter prediction value is directly generated;

[0046] When the working condition coupling index GCI is less than the working condition coupling threshold Gth, it is determined that the prediction is unstable, a second warning instruction is triggered, and a second strategy is generated: performing adaptive weighted correction on the input parameters, including increasing the weight compensation of the torque, pressure and water content signals, and combining the optimal correction factor of the historical construction database to recalculate the prediction result; At the same time, the GCI value in the correction process, the prediction value before and after correction, and the parameter weighting situation are recorded into the log for subsequent model retraining and optimization.

[0047] Preferably, the step five comprises:

[0048] S51, collecting actual performance data in the construction process, including muck flowability Lf, cutter wear rate Mc and energy consumption Ec construction result indicators; At the same time, the corresponding parameter values predicted by the working condition coupling prediction model are collected, including muck flowability prediction value Lpred, cutter wear rate prediction value Mpred and energy consumption prediction value Epred;

[0049] S52, comparing the collected actual construction performance data with the predicted parameter values to calculate and obtain a prediction correction index AEI;

[0050] S53, by comparing the prediction correction index AEI with the preset prediction correction threshold Ath, a fourth evaluation result is obtained, including:

[0051] When the prediction correction index AEI < the prediction correction threshold Ath, it is determined that the prediction deviation of the working condition coupling prediction model is in a reasonable range, no correction is made, and continuous monitoring is performed;

[0052] When the prediction correction index AEI ≥ the prediction correction threshold Ath, it is determined that the prediction deviation of the working condition coupling prediction model is not in a reasonable range, the prediction deviation is large, a third early warning instruction is triggered, and a third strategy is generated: starting an online correction and migration learning mechanism, including incrementally updating the parameters of the working condition coupling prediction model, adjusting the model weight and bias, and retraining the model in batches according to the AEI value recorded in the construction database and the model correction record.

[0053] Preferably, a shield tunnel spoil improvement parameter prediction system comprises:

[0054] A multi-modal data acquisition module is used to arrange multiple sensors at key positions of the shield machine, and to acquire in real time the cutter torque, the thrust force, the soil pressure in the cabin, the spoil moisture content, the formation dry density, the spectrum and resistivity signals, the spoil particle size distribution, the spoil liquid-plastic limit parameters and the improvement agent injection parameters.

[0055] A mutation detection and feature extraction module is used to extract the change rate, acceleration term and mutation variable based on the spoil moisture content, cutter torque and resistivity signals through time series difference, sliding window regression, filtering differentiation and anomaly detection methods, and to calculate a mutation sensitive factor , and compare and analyze the mutation sensitive factor with a formation mutation determination threshold mth to determine whether the current working condition is a normal working condition or a formation mutation working condition.

[0056] A model prediction and adaptation evaluation module is used to train a convolutional neural network normal prediction sub-model library based on normal working condition data, for prediction of different formation categories; a mutation rapid response model constructed by combining a convolutional neural network and a recurrent neural network, for emergency prediction of a mutation working condition; a formation model adaptation index CMI calculated through consistency of historical data fitting and formation identification; and comparison and analysis of the formation model adaptation index CMI with a formation model adaptation threshold Cth to determine whether the current model is adapted, and if not, the first strategy is given.

[0057] A working condition coupling analysis and prediction module is used to construct a working condition correlation graph based on the tunneling parameters, formation parameters and improvement agent addition amount, to establish a working condition coupling prediction model by using a graph neural network, to extract key nonlinear coupling features, to calculate a working condition coupling index GCI, and to compare and analyze the working condition coupling index GCI with a working condition coupling threshold Gth to determine whether the prediction is stable, and if not, the second strategy is given.

[0058] The construction performance evaluation and correction module is used for comparing the actual construction performance data with the predicted value of the working condition coupling prediction model, calculating a prediction correction index AEI, and comparing the AEI with a prediction correction threshold Ath to determine whether the prediction deviation of the working condition coupling prediction model is within a reasonable range, and if not, a third strategy is provided.

[0059] The application provides a shield tunnel spoil improvement parameter prediction method and system.

[0060] (1) The shield tunnel spoil improvement parameter prediction method and system can accurately extract the change rate, acceleration term and mutation variable by real-time collection of tunneling parameters, stratum parameters, spoil physical properties and improvement agent injection parameters through multi-modal sensors, and combination of time series difference, sliding window regression, filtering differentiation and anomaly detection methods, and can realize high-precision real-time prediction of shield tunnel spoil improvement parameters, and significantly improve the reliability and response speed of prediction.

[0061] (2) The shield tunnel spoil improvement parameter prediction method and system introduces a double-model architecture combining a normal prediction sub-model library and a mutation rapid response model, and calculates a mutation sensitive factor determines the working condition, can call different models for normal and mutation working conditions respectively, improves the adaptability and prediction stability of the system under complex working conditions, and reduces the construction risk caused by stratum mutation.

[0062] (3) The shield tunnel spoil improvement parameter prediction method and system can effectively extract the nonlinear coupling features between the tunneling parameters, stratum parameters and improvement agent injection amount by constructing a working condition correlation graph and introducing a graph neural network to establish a working condition coupling prediction model, calculate a working condition coupling index GCI, realize stability determination and strategy adjustment for complex working conditions, and thus ensure the efficiency, safety and economy of the construction process.

[0063] (4) The shield tunnel spoil improvement parameter prediction method and system introduces a prediction correction index AEI and an online correction and migration learning mechanism, realizes dynamic comparison, deviation correction and model updating of the construction process performance and the prediction result, ensures the long-term adaptability and accuracy of the prediction model, and provides continuous optimization technical support through strategy generation and log recording. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 It is a shield tunnel spoil improvement parameter prediction method step schematic diagram of the application;

[0065] Figure 2 It is a shield tunnel spoil improvement parameter prediction system block diagram flow chart of the application. DETAILED DESCRIPTION

[0066] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0067] Embodiment 1

[0068] Please refer to Figure 1 The present application provides a method for predicting parameters of improved spoil in shield tunneling, comprising the following steps:

[0069] Step one, by laying multiple sensors at key positions of the shield machine, real-time acquisition of cutter torque, thrust force, soil pressure in the cabin, spoil moisture content, formation dry density, spectrum and resistivity signal, spoil particle size distribution, spoil liquid-plastic limit parameters and modifier injection parameters;

[0070] Step two, based on the spoil moisture content, cutter torque and resistivity signal, the change rate, acceleration term and mutation variable are extracted by the methods of time series difference, sliding window regression, filter differentiation and anomaly detection, and the mutation sensitive factor is calculated , and compared with the formation mutation judgment threshold mth to determine whether the current working condition is normal or formation mutation;

[0071] Step three, the convolutional neural network normal prediction sub-model library trained based on normal working condition data is used for prediction of different formation categories; the mutation rapid response model constructed by combining convolutional neural network and recurrent neural network is used for emergency prediction of mutation working condition, the formation model adaptation index CMI is calculated through consistency of historical data fitting and formation identification, and compared with the formation model adaptation threshold Cth to determine whether the current model is adapted, and if not, the first strategy is given;

[0072] Step four, based on the excavation parameters, formation parameters and modifier addition amount, a working condition correlation atlas is constructed, a working condition coupling prediction model is established by using a graph neural network, key nonlinear coupling features are extracted, a working condition coupling index GCI is calculated, and compared with the working condition coupling threshold Gth to determine whether the prediction is stable, and if not, the second strategy is given;

[0073] Step five, the actual performance data of construction is collected and compared with the predicted value of the working condition coupling prediction model, the prediction correction index AEI is calculated, and compared with the prediction correction threshold Ath to determine whether the prediction deviation of the working condition coupling prediction model is within a reasonable range, and if not, the third strategy is given.

[0074] In this embodiment, by constructing a shield tunnel spoil improvement parameter prediction method including five system modules of multi-modal data acquisition, mutation detection, model prediction and adaptive evaluation, working condition coupling analysis and construction performance evaluation, real-time, high-precision prediction and dynamic self-adaptive adjustment of spoil improvement parameters under different stratum working conditions are realized, which significantly improves the safety, stability and construction efficiency of the shield construction process, while reducing the construction risk and operation cost.

[0075] Embodiment 2

[0076] This embodiment is an explanation and description in embodiment 1. Specifically, the step one comprises:

[0077] S11, by installing a torque sensor on the main drive system of the shield machine, collecting cutter torque data Tp; by installing a thrust sensor on the thrust cylinder of the shield machine, collecting thrust data F; by arranging soil pressure sensors on the inner wall of the soil pressure chamber, collecting chamber soil pressure data P; establishing a tunneling parameter data set;

[0078] S12, by setting a near-infrared spectrum sensor at the spoil sampling port, collecting spoil moisture content data Wt; by calling the stratum database in the construction log acquisition unit, obtaining stratum dry density data Gd; establishing a stratum parameter data set;

[0079] S13, by the near-infrared spectrum sensor, collecting stratum spectrum response signal data Snir; by the resistivity probe, collecting stratum resistivity signal data Sres; establishing a multi-modal sensor signal data set;

[0080] S14, by arranging a micro-particle online analysis device at the outlet of the soil pressure chamber, collecting spoil particle size distribution data Dp; by arranging a micro-particle online analysis device at the outlet of the soil pressure chamber, collecting spoil liquid-plastic limit parameter data Lp; establishing a spoil physical property data set;

[0081] S15, by arranging flow sensors, pressure sensors and concentration detection devices on the injection device, real-time collection of improvement agent injection parameters, including flow, pressure, concentration and injection time, forming an improvement agent injection parameter set.

[0082] In this embodiment, by arranging various sensors at key positions of the shield machine, cutter torque, thrust, chamber soil pressure, spoil moisture content, stratum dry density, spectrum and resistivity signal, spoil particle size distribution, liquid-plastic limit parameter and improvement agent injection parameter are collected in real time, realizing multi-modal, full-time, high-precision data collection of tunneling parameters, stratum characteristics and spoil physical properties, providing a comprehensive and reliable data basis for subsequent mutation detection, model prediction and optimization.

[0083] Embodiment 3

[0084] This embodiment is an explanation in embodiment 2, specifically, the step two includes:

[0085] S21, based on the slag water content data Wt, using time series difference and sliding window regression technology, the change of water content in continuous time series is calculated by first derivative, so as to obtain the change rate of water content ; based on the cutter torque data Tp, using time series filtering and numerical differentiation method, first, the torque sequence is smoothed and denoised, and then the second derivative is calculated to obtain the acceleration term of torque ; based on the resistivity signal data Sres; using anomaly detection and time series mutation analysis method, the difference of adjacent time period resistivity signal is compared, the significant electrical property mutation is identified, the reflectivity offset of spectral response signal data Snir is combined, the validity of resistivity signal mutation is corrected, and the final resistivity mutation variable is obtained .

[0086] In this embodiment, by using time series difference, sliding window regression, filtering differentiation and anomaly detection method, the change rate of slag water content, the acceleration term of cutter torque and the resistivity mutation variable are accurately extracted, the dynamic change and mutation characteristics of key parameters in shield tunneling process are realized, the reliable feature input is provided for stratum mutation judgment and subsequent model prediction, and the timeliness and accuracy of prediction are improved.

[0087] Embodiment 4

[0088] This embodiment is an explanation in embodiment 3, specifically, the step two further includes:

[0089] S22, by obtaining the change rate of water content , the acceleration term of torque And the resistivity mutation variable , after dimensionless treatment, the mutation sensitive factor is calculated and obtained, the formula is as follows:

[0090]

[0091] In the formula, w1, w2 and w3 represent weight coefficients;

[0092] : representing the influence of cutter torque second derivative on stratum mutation, occupying higher weight, is a key index, directly reflecting the mutation characteristics of torque change;

[0093] : representing the influence of slag water content gradient change on stratum mutation, occupying medium weight, reflecting the stratum water content fluctuation and mutation sensitivity;

[0094] : Characterize the influence of the resistivity signal difference on the formation mutation, which is a minor weight, and reflect the contribution of the formation conductivity difference to the mutation working condition;

[0095] S23, a preset formation mutation determination threshold mth is obtained, and the mutation sensitive factor is compared and analyzed with the formation mutation determination threshold mth to obtain a first evaluation result, including:

[0096] When the mutation sensitive factor ≤ the formation mutation determination threshold mth, it is determined that the current working condition is a normal working condition;

[0097] When the mutation sensitive factor > the formation mutation determination threshold mth, it is determined that the current working condition is a formation mutation working condition.

[0098] The acquisition method of the formation mutation determination threshold mth: through statistical analysis on the time sequence variation characteristics of a large amount of data such as spoil water content, cutter torque and resistivity signal in the process of shield construction, the distribution range of the mutation sensitive factor under normal working condition and mutation working condition is extracted, combined with construction experience and geological survey results, a reasonable determination threshold interval is determined. Referring to the shield construction industry standard, the geological engineering specification and the related construction cases, the specific numerical value is determined to effectively distinguish the normal working condition and the formation mutation working condition, and to ensure the safety and adaptability of the tunneling process.

[0099] In this embodiment, the mutation sensitive factor is constructed, and combined with the preset formation mutation determination threshold mth, the rapid and accurate determination of the normal working condition and the formation mutation working condition in the process of shield tunneling is realized, which helps to improve the real-time and reliability of working condition identification, and further provides a solid basis for subsequent model switching and emergency response.

[0100] Embodiment 5

[0101] This embodiment is an explanation and description in embodiment 4. Specifically, the step three includes:

[0102] S31, a convolutional neural network is used to construct a convolutional neural network initial model, and the cutter torque data Tp, the spoil water content data Wt, the resistivity signal data Sres, the spectral response signal data Snir and the corresponding formation category and modifier injection parameters under the normal working condition are used as training features to train and test the convolutional neural network initial model; the trained convolutional neural network initial model is used as a normal prediction sub-model library, and each sub-model corresponds to a formation category; when running, according to the output result of the formation recognition module, the prediction sub-model corresponding to the current formation category is called, and the prediction result is output;

[0103] S32, for the mutation working condition sample data, a convolutional neural network is used to combine a recurrent neural network to construct a mutation fast response initial model, and the change rate of the water content under the mutation working condition is used as a training feature , the acceleration term of the torque , and the resistivity mutation variable are trained and tested; the trained model is used as a mutation fast response prediction model, and when it is determined that the working condition is a mutation working condition, the mutation fast response model is automatically switched and called to generate an emergency prediction result;

[0104] S33, historical construction data is collected and acquired, and the goodness of fit of the normal prediction sub-model library and the mutation fast response model is evaluated; the residual sum of squares is compared with the actual collected data to obtain the goodness of fit index Rmodel of each model;

[0105] S34, based on the real-time formation identification result, the probability distribution after feature fusion and clustering discrimination is extracted to obtain the consistency probability Pmatch between the formation identification result and the called prediction model category.

[0106] In this embodiment, by constructing the normal prediction sub-model library and the mutation fast response prediction model, and combining the formation identification result for dynamic model switching, precise prediction and rapid response of the slag soil improvement parameters under different working conditions are realized, and the adaptability and prediction accuracy of the shield tunneling process are greatly improved.

[0107] Embodiment 6

[0108] This embodiment is an explanation and description in embodiment 5, specifically, the step three further comprises:

[0109] S35, by obtaining the goodness of fit index Rmodel of each model and the consistency probability Pmatch between the formation identification result and the called prediction model category, after non-dimensional processing, the formation model adaptation index CMI is calculated and obtained, and the formula is as follows:

[0110]

[0111] In the formula, a1 and a2 represent weight coefficients;

[0112] : represents the influence of the model fitting correlation Rmodel on the adaptability, occupies the main weight, and reflects the explanation ability of the prediction model to the current formation data;

[0113] : represents the influence of the formation matching probability Pmatch on the adaptability, occupies the secondary weight, and embodies the coincidence degree of the formation identification result and the actual working condition;

[0114] S36, by presetting the formation model fitting threshold Cth, and comparing the formation model fitting index CMI with the formation model fitting threshold Cth, a second evaluation result is obtained, including:

[0115] When the formation model fitting index CMI is greater than or equal to the formation model fitting threshold Cth, it indicates that the current model is fitted, and the prediction result output by the current model is maintained;

[0116] When the formation model fitting index CMI is less than the formation model fitting threshold Cth, it indicates that the current model is not fitted, a first warning instruction is triggered, and a first strategy is generated: a multi-model weighting mechanism is used to call a backup model pre-trained under different scenes, a temporary prediction result is output by a weighted average method, so that the result will not be completely invalid; the reference parameters of the emergency improver injection under the corresponding formation category are called from the database as the protective compensation input of the model output, so that even if the model is not fitted, reasonable injection control basis can still be provided; the prediction result is dynamically corrected, the model prediction result is compared with the emergency reference value, the prediction result is corrected, and a temporarily corrected prediction output is formed; while correcting the prediction, the current working condition is marked as "model not fitted state", and the key parameters in the correction process including the CMI value, the corrected prediction value and the reference value calling condition are written into the log.

[0117] The formation model fitting threshold Cth is obtained by: fitting analysis of convolutional neural network and recurrent neural network prediction on historical construction data of different formation categories, statistical analysis of goodness of fit indicators and model consistency probability distribution of each formation model, determination of a reasonable threshold interval combined with professional technical personnel experience and formation recognition accuracy requirements. The threshold is formulated according to the industry specifications and construction experience of formation prediction model to ensure that the selected model is highly fitted to the actual working condition, improve the prediction accuracy and reduce the construction risk.

[0118] In this embodiment, by introducing the formation model fitting index CMI and the fitting threshold Cth, dynamic evaluation and adaptive correction of the fitting of the prediction model are realized, which ensures that the backup model can be switched and protective compensation can be performed when the model is not fitted, effectively improving the stability and reliability of the prediction.

[0119] Embodiment 7

[0120] This embodiment is an explanation and description in embodiment 6, specifically, the step four includes:

[0121] S41, based on the cutterhead torque data Tp, the advancing force data F, and the cabin earth pressure data P of the tunneling parameter group, the muck water content data Wt of the stratum parameter group, the muck particle size distribution data Dp and the muck liquid-plastic limit parameter data Lp of the muck physical property data group, and combined with the actual modifier addition amount, a working condition correlation atlas is established; based on a graph neural network GNN, a working condition coupling initial model is constructed, and the tunneling parameters, stratum parameters, and modifier addition amount are taken as input nodes to train and test the working condition coupling initial model; the trained working condition coupling initial model is taken as a working condition coupling prediction model for extracting nonlinear coupling features in runtime, including the correlation between the cutterhead torque data and the muck water content data , the correlation between the advancing force data and the stratum pressure distribution data , and the correlation between the cabin earth pressure data and the muck water permeability data .

[0122] In this embodiment, by constructing a working condition correlation atlas and introducing a graph neural network (GNN) for nonlinear coupling feature extraction, the complex correlation between the tunneling parameters, stratum characteristics, and modifier addition amount is effectively revealed, and the accuracy and adaptability of the muck improvement parameter prediction are significantly improved.

[0123] Embodiment 8

[0124] This embodiment is an explanation and description in embodiment 7. Specifically, the step four further includes:

[0125] S42, the nonlinear coupling features extracted by the working condition coupling prediction model, the correlation between the cutterhead torque data and the muck water content data , the correlation between the advancing force data and the stratum pressure distribution data , and the correlation between the cabin earth pressure data and the muck water permeability data After dimensionless processing, the working condition coupling index GCI is calculated and obtained, and the formula is as follows:

[0126]

[0127] In the formula, s1, s2, and s3 represent weight coefficients;

[0128] : represents the correlation between the advancing force and the cutterhead torque , which has a medium weight on the working condition coupling and reflects the coupling strength of the tunneling dynamics and the load;

[0129] : represents the correlation between the earth pressure and the muck flowability , which has a higher weight on the working condition coupling and is a key index, reflecting the matching degree of the stratum pressure and the muck characteristics;

[0130] : Characterizing the correlation between the liquid-plastic limit of the slag soil and the injection amount of the modifier The influence of the working condition coupling accounts for a minor weight, and embodies the coupling relationship between the soil improvement measure and the rheological property;

[0131] S43, by presetting a working condition coupling threshold Gth, and comparing and analyzing the working condition coupling index GCI with the working condition coupling threshold Gth, a third evaluation result is obtained, including:

[0132] When the working condition coupling index GCI is greater than or equal to the working condition coupling threshold Gth, it is determined that the prediction is stable, the working condition coupling prediction model output result is maintained, and the slag soil improvement parameter prediction value is directly generated;

[0133] When the working condition coupling index GCI is less than the working condition coupling threshold Gth, it is determined that the prediction is unstable, a second warning instruction is triggered, and a second strategy is generated: performing adaptive weighted correction on the input parameters, including increasing the weight compensation of the torque, pressure and water content signal, and combining the optimal correction factor of the historical construction database to recalculate the prediction result; at the same time, the GCI value in the correction process, the prediction value before and after the correction, and the parameter weighting situation are recorded into the log for subsequent model retraining and optimization.

[0134] The working condition coupling threshold Gth is obtained by: performing graph neural network coupling feature analysis on the historical data of the tunneling parameters, stratum parameters and modifier injection amount in the shield construction process, extracting the coupling feature distribution interval under different working conditions, combining the expert experience and the construction stability requirement, and determining a reasonable threshold. According to the shield construction working condition analysis standard and the engineering safety specification, the threshold is set to effectively judge the stability of the working condition coupling prediction, and ensure the reliability of the prediction result in the construction process.

[0135] In this embodiment, the working condition coupling index GCI is introduced for dynamic stability evaluation, and the adaptive weighted correction mechanism is combined, so that the input parameters and the model output can be adjusted in real time when the prediction is unstable, thereby effectively improving the reliability and robustness of the slag soil improvement parameter prediction, and ensuring the safety and construction quality of the shield construction process.

[0136] Embodiment 9

[0137] This embodiment is an explanation and description in embodiment 8, specifically, the step five includes:

[0138] S51, collecting actual performance data in the construction process, including the slag soil fluidity Lf, the cutter wear rate Mc and the energy consumption Ec construction result index; at the same time, collecting the corresponding parameter values predicted by the working condition coupling prediction model, including the slag soil fluidity prediction value Lpred, the cutter wear rate prediction value Mpred and the energy consumption prediction value Epred;

[0139] S52, compare the collected actual construction performance data with the predicted parameter value, calculate the predicted correction index AEI, the formula is as follows:

[0140]

[0141] In the formula, d1, d2 and d3 represent weight coefficients;

[0142] : represents the influence of the deviation of the flowability of the muck Lf on the prediction correction, accounts for the main weight, is a key parameter, and directly reflects the difference in the excavability of the soil in the construction process;

[0143] : represents the influence of the deviation of the tool wear rate Mc on the prediction correction, accounts for the medium weight, and reflects the difference between the tool performance and the construction energy consumption;

[0144] : represents the influence of the deviation of the energy consumption Ec on the prediction correction, accounts for the secondary weight, and embodies the contribution of the difference between the overall energy efficiency and the prediction model;

[0145] S53, by presetting the prediction correction threshold Ath, and comparing the prediction correction index AEI with the prediction correction threshold Ath, a fourth evaluation result is obtained, including:

[0146] When the prediction correction index AEI is less than the prediction correction threshold Ath, it is determined that the prediction deviation of the working condition coupling prediction model is in a reasonable range, no correction is made, and continuous monitoring is performed;

[0147] When the prediction correction index AEI is greater than or equal to the prediction correction threshold Ath, it is determined that the prediction deviation of the working condition coupling prediction model is not in a reasonable range, the prediction deviation is large, a third early warning instruction is triggered, and a third strategy is generated: starting the online correction and migration learning mechanism, including incrementally updating the parameters of the working condition coupling prediction model, adjusting the model weight and deviation; and retraining the model in batches according to the AEI value recorded in the construction database and the model correction record.

[0148] The acquisition method of the prediction correction threshold Ath: by statistically analyzing the deviation of a large amount of construction actual performance data (such as the flowability of muck, the tool wear rate, and the energy consumption) and the predicted value, the distribution range of the prediction correction index is obtained, the reasonable threshold interval is determined by combining the historical construction error experience and the model stability requirement. According to the shield construction quality control specification and the prediction model self-adaptive correction standard, the threshold is set to effectively identify whether the prediction deviation is in a reasonable range, so as to ensure the accuracy and safety of the construction prediction.

[0149] In this embodiment, the prediction correction index AEI is introduced to evaluate the real-time deviation of the output of the working condition coupling prediction model, and combined with the online correction and migration learning mechanism, when the deviation exceeds the reasonable range, the model parameters and weights can be automatically adjusted to realize the adaptive optimization of the prediction model, and the accuracy and reliability of the prediction of the shield construction process are significantly improved.

[0150] Embodiment 10

[0151] A shield tunnel spoil improvement parameter prediction system, please refer to Figure 2 , comprising:

[0152] A multi-modal data acquisition module is used to arrange multiple sensors at key positions of the shield machine, and real-time acquisition of cutter torque, thrust force, in-cabin earth pressure, spoil water content, formation dry density, spectrum and resistivity signals, spoil particle size distribution, spoil liquid-plastic limit parameters and improvement agent injection parameters is performed.

[0153] A mutation detection and feature extraction module is used to extract the change rate, acceleration term and mutation variable based on the spoil water content, cutter torque and resistivity signals through time series difference, sliding window regression, filtering differentiation and anomaly detection methods, and calculate the mutation sensitive factor , and compare and analyze it with the formation mutation judgment threshold mth to determine whether the current working condition is normal or formation mutation;

[0154] A model prediction and adaptation evaluation module is used to train a convolutional neural network normal prediction sub-model library based on normal working condition data for prediction of different formation categories; a mutation rapid response model constructed by combining convolutional neural network and recurrent neural network is used for emergency prediction of mutation working condition, a formation model adaptation index CMI is calculated through consistency of historical data fitting and formation identification, and compared and analyzed with the formation model adaptation threshold Cth to determine whether the current model is adapted, and if not, the first strategy is given.

[0155] A working condition coupling analysis and prediction module is used to construct a working condition correlation atlas based on the tunneling parameters, formation parameters and improvement agent addition amount, use a graph neural network to establish a working condition coupling prediction model, extract key nonlinear coupling features, calculate a working condition coupling index GCI, and compare and analyze it with the working condition coupling threshold Gth to determine whether the prediction is stable, and if not, the second strategy is given.

[0156] A construction performance evaluation and correction module is used to compare the actual performance data of the construction with the predicted value of the working condition coupling prediction model, calculate a prediction correction index AEI, and compare and analyze it with the prediction correction threshold Ath to determine whether the prediction deviation of the working condition coupling prediction model is within a reasonable range, and if not, the third strategy is given.

[0157] In the embodiment, through the modular design of multi-modal data acquisition, mutation detection, model prediction adaptation, working condition coupling analysis and construction performance correction, the fusion and intelligent processing of multi-source information in the shield construction process are realized, the stratum mutation can be identified in real time, the model adaptability can be dynamically evaluated, the construction working condition coupling stability can be predicted, and the prediction deviation can be automatically corrected when the prediction deviation is out of limit, so that the construction safety, the prediction accuracy and the decision reliability are effectively improved.

[0158] The size of the threshold is set for comparison, and the size of the threshold depends on how much sample data and the number of base set by the person skilled in the art for each group of sample data; as long as the proportional relationship between the parameters and the quantized values is not affected.

[0159] The above formulas are obtained by collecting a large amount of data for software simulation and selecting a formula close to the true value, and the coefficients in the formula are set by the person skilled in the art according to the actual situation. The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can make equivalent replacement or change according to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for predicting parameters of soil improvement in shield tunnels, characterized in that, Includes the following steps: Step 1: By deploying various sensors at key parts of the tunnel boring machine, real-time data are collected on the cutterhead torque, thrust, earth pressure inside the chamber, moisture content of the excavated soil, dry density of the stratum, spectral and resistivity signals, particle size distribution of the excavated soil, liquid and plastic limit parameters of the excavated soil, and parameters for the injection of amendments. Step 2: Based on the moisture content of the slag, cutterhead torque, and resistivity signals, extract the rate of change, acceleration term, and abrupt change amount using time-series difference, sliding window regression, filtered differentiation, and anomaly detection methods, and calculate the abrupt change sensitivity factor. The results are compared with the formation mutation judgment threshold mth to determine whether the current working condition is a normal working condition or a formation mutation working condition. Mutation-sensitive factors The method for obtaining the data is as follows: based on the moisture content data Wt of the slag soil, time-series difference and sliding window regression techniques are used to calculate the first derivative of the moisture content change in the continuous time series, thereby obtaining the rate of change of moisture content. ; Based on the cutterhead torque data Tp, a time-series filtering and numerical differentiation method is used to first smooth and denoise the torque sequence, and then calculate the second derivative to obtain the acceleration term of the torque. Based on resistivity signal data Sns, anomaly detection and temporal abrupt change analysis methods are employed to compare the differences in resistivity signals between adjacent time periods, identifying significant electrical abrupt changes. The reflectance shift of the spectral response signal data Snir is then used to correct the validity of these resistivity signal abrupt changes, yielding the final resistivity abrupt change amount. The rate of change of moisture content obtained acceleration term of torque and resistivity abrupt change After dimensionless processing, the mutation sensitivity factor was calculated and obtained. The formula is as follows: In the formula, w1, w2, and w3 represent weighting coefficients; Step 3: A convolutional neural network-based normal prediction sub-model library trained on normal operating condition data is used for prediction of different strata categories; a rapid response model for sudden changes is constructed by combining convolutional neural networks and recurrent neural networks for emergency prediction of sudden operating conditions. The strata model fit index CMI is calculated by fitting historical data and the consistency of strata identification, and compared with the strata model fit threshold Cth to determine whether the current model is suitable. If it is not suitable, the first strategy is given. The formation model fit index (CMI) is obtained by using the goodness-of-fit index (Rmodel) of each model and the consistency probability (Pmatch) between the formation identification results and the category of the called prediction model. After dimensionless processing, the formation model fit index (CMI) is calculated, as shown in the following formula: In the formula, a1 and a2 represent weight coefficients; historical construction data are collected, and the goodness of fit between the normal prediction sub-model library and the rapid response model for sudden changes is evaluated. The sum of squared residuals is compared with the actual collected data to obtain the goodness of fit index Rmodel for each model; based on the real-time stratum identification results, the probability distribution after feature fusion and clustering is extracted to obtain the consistency probability Pmatch between the stratum identification results and the category of the called prediction model. Step 4: Construct a working condition correlation map based on tunneling parameters, formation parameters and amendment addition amount, establish a working condition coupling prediction model using graph neural network, extract key nonlinear coupling features, calculate the working condition coupling index GCI, and compare it with the working condition coupling threshold Gth to determine whether the prediction is stable. If it is unstable, a second strategy is given. The Condition Coupling Index (GCI) is obtained by analyzing the correlation between cutterhead torque data and slag moisture content data through nonlinear coupling characteristics extracted from the condition coupling prediction model. Correlation between thrust data and formation pressure distribution data Correlation between in-cabin earth pressure data and spoil permeability data After dimensionless processing, the working condition coupling index GCI is calculated and obtained, as shown in the following formula: In the formula, s1, s2 and s3 represent weighting coefficients; Step 5: Collect actual construction performance data and compare it with the predicted value of the working condition coupling prediction model. Calculate the prediction correction index AEI and compare it with the prediction correction threshold Ath to determine whether the prediction deviation of the working condition coupling prediction model is within a reasonable range. If it is not within a reasonable range, then the third strategy is given. The prediction correction index (AEI) is obtained by collecting actual construction performance data, including soil flowability (Lf), tool wear rate (Mpred), and energy consumption (Ec) as construction result indicators; simultaneously, it collects the corresponding parameter values ​​predicted by the coupled prediction model, including predicted soil flowability (Lpred), predicted tool wear rate (Mpred), and predicted energy consumption (Epred); the collected actual construction performance data is compared with the predicted parameter values ​​to calculate the prediction correction index AEI, as shown in the following formula: In the formula, d1, d2 and d3 represent weighting coefficients.

2. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step one includes: S11. Collect cutterhead torque data Tp by installing torque sensors on the main drive system of the tunnel boring machine; collect propulsion thrust data F by installing thrust sensors on the propulsion cylinders of the tunnel boring machine; collect earth pressure data P by installing earth pressure sensors on the inner wall of the earth pressure chamber; establish a tunneling parameter data set. S12. Collect soil moisture content data Wt by setting a near-infrared spectral sensor at the soil sampling port; obtain soil dry density data Gd by calling the formation database through the construction log acquisition unit; and establish a formation parameter data group. S13. Collect formation spectral response signal data Snir using a near-infrared spectral sensor; collect formation resistivity signal data Sres using a resistivity probe; establish a multimodal sensor signal data set; S14. Collect the particle size distribution data Dp of the slag soil by arranging a micro-particle online analysis device at the outlet of the soil pressure chamber; collect the liquid limit and plastic limit parameter data Lp of the slag soil by arranging a micro-particle online analysis device at the outlet of the soil pressure chamber; and establish a data set of physical properties of the slag soil. S15. By installing flow sensors, pressure sensors and concentration detection devices on the injection device, the modifier injection parameters, including flow rate, pressure, concentration and injection time, are collected in real time to form a modifier injection parameter set.

3. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step two also includes: The formation mutation determination threshold mth is preset, and mutation sensitivity factors are... A comparative analysis was performed with the formation abrupt change threshold mth to obtain the first evaluation results, including: When mutation-sensitive factors When the current working condition is ≤ the formation change judgment threshold mth, it is determined to be a normal working condition; When mutation-sensitive factors When the formation mutation determination threshold mth is reached, the current working condition is determined to be a formation mutation working condition.

4. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step three includes: S31. Construct an initial convolutional neural network model using a convolutional neural network. Use the cutterhead torque data Tp, slag moisture content data Wt, resistivity signal data Sres, spectral response signal data Snir, and the corresponding formation type and amendment injection parameters collected under normal operating conditions as training features to train and test the initial convolutional neural network model. Use the trained initial convolutional neural network model as a normal prediction sub-model library, where each sub-model corresponds to a formation type. During runtime, based on the output of the formation identification module, call the prediction sub-model corresponding to the current formation type and output the prediction result. S32. For the sample data of sudden change conditions, a rapid response initial model for sudden change is constructed using a convolutional neural network combined with a recurrent neural network, and the rate of change of water content under sudden change conditions is used as the starting point. acceleration term of torque and resistivity abrupt change As training features, the model is incrementally trained and tested; the trained model is used as a rapid response prediction model for sudden changes. During runtime, when the working condition is determined to be a sudden change, the rapid response model for sudden changes is automatically switched and invoked to generate emergency prediction results.

5. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step three also includes: By setting a predefined stratigraphic model adaptation threshold Cth and comparing the stratigraphic model adaptation index CMI with the stratigraphic model adaptation threshold Cth, the second evaluation results are obtained, including: When the formation model fit index CMI is greater than or equal to the formation model fit threshold Cth, it indicates that the current model is well-fitted and the prediction results output by the current model are maintained. When the formation model fit index (CMI) is less than the formation model fit threshold (Cth), it indicates that the current model is not compatible, triggering the first warning instruction and generating the first strategy: A multi-model weighting mechanism is adopted, calling pre-trained backup models under different scenarios, and outputting temporary prediction results through weighted averaging to prevent the results from becoming completely invalid; Emergency amendment injection reference parameters for the corresponding formation category are called from the database as protective compensation inputs for the model output, ensuring that even if the model is not compatible, it can still provide a reasonable basis for injection control; The prediction results are dynamically corrected by comparing the model prediction results with the emergency reference values, correcting the prediction results, and forming a temporarily corrected prediction output; While correcting the prediction, the current working condition is marked as "model incompatible state," and key parameters during the correction process, including the CMI value, the corrected prediction value, and the reference value call status, are written to the log.

6. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step four includes: S41. Based on the cutterhead torque data Tp, thrust data F, and internal earth pressure data P from the tunneling parameter group; the slag moisture content data Wt from the formation parameter data group; and the slag particle size distribution data Dp and liquid / plastic limit parameter data Lp from the slag physical property data group, and combined with the actual amount of amendment added, a working condition correlation map is established. An initial working condition coupling model is constructed based on a graph neural network (GNN), using tunneling parameters, formation parameters, and amendment addition amount as input nodes to train and test the initial working condition coupling model. The trained initial working condition coupling model is used as a working condition coupling prediction model to extract nonlinear coupling features during operation, including the correlation between cutterhead torque data and slag moisture content data. Correlation between thrust data and formation pressure distribution data Correlation between in-cabin earth pressure data and spoil permeability data .

7. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step four also includes: By setting a preset operating condition coupling threshold Gth, and comparing the operating condition coupling index GCI with the operating condition coupling threshold Gth, the third evaluation results are obtained, including: When the working condition coupling index GCI is greater than or equal to the working condition coupling threshold Gth, the prediction is determined to be stable, the output of the working condition coupling prediction model is maintained, and the predicted value of the slag soil improvement parameter is directly generated. When the working condition coupling index GCI is less than the working condition coupling threshold Gth, the prediction is determined to be unstable, triggering a second early warning instruction and generating a second strategy: performing adaptive weighted correction on the input parameters, including adding weight compensation to the torque, pressure, and moisture content signals, and recalculating the prediction results by combining the optimal correction factor from the historical construction database; at the same time, the GCI value, the prediction values ​​before and after correction, and the parameter weighting are recorded in the log for subsequent model retraining and optimization.

8. The method for predicting parameters of soil improvement in shield tunnels according to claim 1, characterized in that, Step five includes: By setting a preset prediction correction threshold Ath and comparing the prediction correction index AEI with the prediction correction threshold Ath, the fourth evaluation result is obtained, including: When the prediction correction index AEI is less than the prediction correction threshold Ath, the prediction deviation of the working condition coupled prediction model is determined to be within a reasonable range, no correction is made, and continuous monitoring is carried out. When the prediction correction index AEI is greater than or equal to the prediction correction threshold Ath, it is determined that the prediction deviation of the working condition coupled prediction model is not within a reasonable range and the prediction deviation is large. This triggers the third early warning instruction and generates the third strategy: to start the online correction and transfer learning mechanism, including incrementally updating the parameters of the working condition coupled prediction model and adjusting the model weights and deviations; and to retrain the model in batches based on the AEI values ​​and model correction records recorded in the construction database.

9. A shield tunnel spoil improvement parameter prediction system, applied to the shield tunnel spoil improvement parameter prediction method according to any one of claims 1 to 8, characterized in that, include: The multimodal data acquisition module is used to deploy various sensors in key parts of the tunnel boring machine to collect data in real time, including cutterhead torque, thrust, earth pressure inside the chamber, moisture content of the slag, dry density of the stratum, spectral and resistivity signals, particle size distribution of the slag, liquid and plastic limit parameters of the slag, and parameters for the injection of amendments. The mutation detection and feature extraction module is used to extract the rate of change, acceleration term, and mutation amount based on the moisture content of the slag, cutterhead torque, and resistivity signals, using methods such as time-series difference, sliding window regression, filtered differentiation, and anomaly detection, and to calculate the mutation sensitivity factor. The results are compared with the formation mutation judgment threshold mth to determine whether the current working condition is a normal working condition or a formation mutation working condition. The model prediction and adaptation evaluation module is used for the normal prediction sub-model library of convolutional neural networks trained on normal working condition data, which is used for prediction of different strata categories; the rapid response model for sudden changes is constructed by combining convolutional neural networks and recurrent neural networks, which is used for emergency prediction of sudden working conditions. The strata model fit index CMI is calculated by fitting historical data and the consistency of strata identification, and compared with the strata model fit threshold Cth to determine whether the current model is suitable. If it is not suitable, the first strategy is given. The working condition coupling analysis and prediction module is used to construct a working condition correlation map based on tunneling parameters, formation parameters and amendment addition amount, establish a working condition coupling prediction model using graph neural network, extract key nonlinear coupling features, calculate the working condition coupling index GCI, and compare it with the working condition coupling threshold Gth to determine whether the prediction is stable. If it is unstable, a second strategy is given. The construction performance evaluation and correction module is used to collect actual construction performance data and compare it with the predicted values ​​of the working condition coupling prediction model, calculate the prediction correction index AEI, and compare and analyze it with the prediction correction threshold Ath to determine whether the prediction deviation of the working condition coupling prediction model is within a reasonable range. If it is not within a reasonable range, a third strategy is given.

Images