A shield grinding pile cutting performance evaluation method based on working condition partitioning agent model

Abstract

The present application relates to the technical field of shield construction risk rapid assessment, and particularly relates to a shield grinding pile cutting performance assessment method based on a working condition partition agent model. The method comprises the following steps: obtaining a basic data set of shield grinding pile working conditions and a constraint set for cutting performance assessment; constructing a parameterized cutting simulation model according to the basic data set and the constraint set; constructing a shield grinding pile cutting simulation sample library according to a master control input parameter set and the parameterized cutting simulation model; constructing and training a working condition partition agent model by using the shield grinding pile cutting simulation sample library; obtaining a new shield grinding pile working condition based on the working condition partition agent model; rapidly assessing the new shield grinding pile working condition, and outputting a shield grinding pile cutting performance assessment result after simulation review. The method solves the problems in the prior art, such as fitting all working conditions into a single mapping relationship, large prediction error of edge working conditions or high-risk working conditions, inability to meet the comprehensive comparison and selection demand of a construction scheme, and lack of a prediction reliability judgment mechanism.

Description

Technical Field

[0001] This invention relates to the field of rapid risk assessment technology for tunnel boring machine (TBM) construction, and in particular to a method for evaluating the cutting performance of TBM pile grinding based on a working condition zoning proxy model. Background Technology

[0002] With the rapid development of urban rail transit, municipal tunnels, and underground space engineering, shield tunneling has become one of the important methods for urban tunnel construction. During shield pile grinding construction, there are complex contact, crushing, shearing, and energy dissipation processes between the cutterhead and the pile. If the pile grinding performance cannot be accurately judged before construction, it may lead to problems such as abnormally increased cutterhead torque, rapid cutter wear, rebar entanglement, poor slag removal, shield attitude disturbance, or even cutterhead jamming, thereby affecting construction safety and efficiency.

[0003] Existing methods for evaluating the cutting performance of tunnel boring machines (TBMs) for piling mainly include empirical judgment, field testing, indoor model testing, and numerical simulation. In recent years, surrogate models and machine learning methods have been increasingly used for response prediction in underground engineering. However, existing methods are mostly focused on predicting TBM tunneling parameters, predicting cutterhead torque time series, predicting surface settlement, or evaluating the stress on tunnel structures. They typically use construction monitoring data or equipment operation data as the main input, and the predicted objects are mostly single output indicators.

[0004] For the specific construction problem of shield tunneling pile grinding, existing methods still have the following shortcomings: First, there is a lack of a master control parameter system for the cutting mechanism of shield tunneling pile grinding; second, there is a lack of a method for constructing a parameterized cutting simulation sample library for pile grinding conditions; third, traditional shallow surrogate models or ordinary neural network models usually fit all working conditions into a single mapping relationship and lack a prediction reliability judgment mechanism, making it difficult to distinguish different local response modes such as orthogonal, eccentric piles, and edge pile grinding, resulting in increased prediction errors under edge or high-risk working conditions; fourth, existing prediction methods often only output a single index, making it difficult to simultaneously evaluate cutting load, instantaneous overload risk, cutting efficiency, and pile damage degree, and failing to meet the needs of comprehensive comparison and selection of construction schemes.

[0005] Therefore, there is an urgent need to propose a rapid evaluation method for cutting performance in shield tunneling pile construction scenarios. While retaining the physical information of numerical simulation, this method can achieve rapid, stable, and verifiable evaluation of cutting torque, cutting energy consumption, and pile damage under new working conditions through a parametric cutting simulation sample library and a working condition partition proxy model. This will provide technical support for the selection of shield tunneling pile construction schemes, optimization of cutterhead layout, and risk control. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model. This method solves the problems of existing technologies that fit all working conditions to a single mapping relationship, leading to increased prediction errors in edge or high-risk working conditions, failure to meet the comprehensive comparison and selection requirements of construction schemes, and lack of a prediction reliability judgment mechanism.

[0007] To achieve the above objectives, in a first aspect, the present invention provides a method for evaluating the cutting performance of shield tunneling pile grinding based on a working condition partitioning proxy model, comprising: acquiring a basic dataset of shield tunneling pile grinding working conditions and a constraint set for cutting performance evaluation; constructing a parametric cutting simulation model based on the basic dataset and the constraint set; determining a set of master control input parameters and constructing a shield tunneling pile grinding cutting simulation sample library based on the master control input parameter set and the parametric cutting simulation model; constructing and training a working condition partitioning proxy model using the shield tunneling pile grinding cutting simulation sample library; obtaining new shield tunneling pile grinding working conditions based on the working condition partitioning proxy model; rapidly evaluating the new shield tunneling pile grinding working conditions, and outputting a shield tunneling pile grinding cutting performance evaluation report after simulation verification.

[0008] This invention achieves rapid evaluation of shield tunneling pile cutting performance by constructing a parametric cutting simulation model and a shield tunneling pile grinding cutting simulation sample library, and training a working condition partition proxy model based on the shield tunneling pile grinding cutting simulation sample library. By constructing a working condition partition proxy model, the predictive stability under complex pile grinding conditions is improved, and a comprehensive evaluation of cutting safety, economy and crushing effect is supported. This provides a basis for construction scheme comparison, cutterhead parameter pre-evaluation and high torque risk identification, and ultimately forms a reliable application closed loop of prediction-verification-updating, improving the reliability and sustainable updating capability of the rapid evaluation results of shield tunneling pile grinding.

[0009] Optionally, constructing the parametric cutting simulation model based on the basic dataset and the constraint set includes: constructing the parametric cutting simulation model based on numerical simulation methods and in combination with the basic dataset and the constraint set.

[0010] This invention constructs a parametric cutting simulation model by combining a basic dataset, a set of constraints, and numerical simulation methods. This parametric cutting simulation model can automatically generate shield tunneling pile grinding simulation conditions according to different parameter combinations, complete geometric reconstruction, material assignment, contact definition, mesh generation, cutting calculation, and extract the corresponding cutting response results.

[0011] Optionally, the step of constructing a shield tunneling pile grinding simulation sample library based on the master control input parameter set and the parameterized cutting simulation model includes: obtaining different parameter combinations according to the master control input parameter set and multiple sampling methods; inputting the parameter combinations into the parameterized cutting simulation model to obtain a set of cutting performance output indicators; and storing the parameter combination input-cutting performance output indicator sample pairs as the shield tunneling pile grinding simulation sample library.

[0012] This invention improves the prediction stability under complex pile grinding conditions by determining the set of master control input parameters, and constructs a shield pile grinding cutting simulation sample library by combining a parameterized cutting simulation model, providing a data foundation for the subsequent construction and training of the working condition partition proxy model.

[0013] Optionally, the step of constructing and training the working condition partition proxy model using the shield tunneling pile cutting simulation sample library includes: constructing the working condition partition proxy model using the shield tunneling pile cutting simulation sample library; constructing a target loss function; dividing the shield tunneling pile cutting simulation sample library into a training set, a validation set, and a test set, and training the working condition partition proxy model in combination with the target loss function; and verifying the working condition partition proxy model.

[0014] This invention constructs a working condition partitioned proxy model through a shield tunneling pile cutting simulation sample library, enabling evaluation results to be obtained quickly without repeating complete and detailed simulation calculations for new working conditions. This significantly reduces the calculation costs of multi-scheme comparison and initial risk screening, and improves the efficiency of scheme evaluation before construction.

[0015] Optionally, constructing the working condition partitioning proxy model includes: constructing the working condition partitioning proxy model including a gated expert partitioning module; the gated expert partitioning module completes the weight allocation of expert sub-models based on the grinding pile contact morphology and cutting working condition characteristics, wherein the expert sub-models correspond to the physical working conditions of the center, eccentric, and edge grinding piles, respectively.

[0016] This invention expresses the coupling relationship between scale, eccentricity, material, reinforcement and tool arrangement through a gating expert partitioning mechanism. The gating expert partitioning module is deeply bound to the physical mechanism of shield tunneling pile grinding, and is used to distinguish the response modes of orthogonal pile grinding, eccentric pile grinding, edge pile grinding and high resistance pile grinding.

[0017] Optionally, the construction of the target loss function includes: constructing a physical consistency constraint loss that includes concrete strength, positive reinforcement ratio, cutterhead structural parameters, and abrupt changes in edge grinding piles; and constructing the target loss function based on the physical consistency constraint loss.

[0018] This invention constructs a physical consistency constraint loss function, which transforms the physical prior laws of shield tunneling pile grinding into quantifiable gradient penalty terms, ensuring the rationality of the physical meaning of the output of the working condition partitioning proxy model and preventing the network from falling into the overfitting trap of pure data-driven approaches.

[0019] Optionally, the rapid evaluation of the new shield tunneling pile grinding condition includes: using the new shield tunneling pile grinding condition to calculate a new parameter combination; inputting the new parameter combination into the condition zoning proxy model to predict and output a new set of cutting performance output indicators and the uncertainty of the indicators; and evaluating whether the current condition is a low-risk, medium-risk, or high-risk pile grinding condition based on the new set of cutting performance output indicators.

[0020] This invention rapidly evaluates new shield tunneling pile grinding conditions by calculating new parameter combinations. It is not limited to single torque prediction, but outputs multiple indicators simultaneously. It can evaluate shield tunneling pile grinding schemes from multiple dimensions such as equipment safety, construction energy consumption and obstacle pile breaking effect, and reduce prediction errors in edge or high-risk conditions.

[0021] Optionally, the shield tunneling pile cutting performance evaluation report output after simulation verification includes: establishing a fine simulation verification mechanism; if the new shield tunneling pile working condition meets any condition in the fine simulation verification mechanism, then triggering the fine simulation verification mechanism; re-inputting the new shield tunneling pile working condition that has triggered verification into the parameterized cutting simulation model to obtain the verified input-output sample pair and storing it in the shield tunneling pile cutting simulation sample library; calculating a comprehensive score based on the new cutting performance output index set to rank the optional construction schemes; and outputting the shield tunneling pile cutting performance evaluation report including the new parameter combination, the new cutting performance output index set, the uncertainty of the index, the working condition risk level, whether fine simulation verification is required, and the ranking of the optional construction schemes.

[0022] This invention establishes a refined simulation verification mechanism. When the uncertainty of the new working condition prediction is high, the prediction result is close to the control threshold, or the input parameters exceed the coverage of the sample library, the verification mechanism can be triggered and the verified sample increment is written into the shield tunneling pile cutting simulation sample library. This mechanism can avoid the working condition partition proxy model from being directly used for engineering decision-making when the extrapolation risk is high, thereby improving the reliability and sustainable updating capability of the shield tunneling pile rapid evaluation results.

[0023] Optionally, the step of calculating a comprehensive score and ranking the available construction schemes based on the new set of cutting performance output indicators includes: the new set of cutting performance output indicators includes a new peak cutting torque, a new energy consumption per unit volume, and a new pile damage rate, and the comprehensive score satisfies the following expression:

[0024] In the formula, For comprehensive scoring, For the new peak cutting torque, For the allowable torque of the tunnel boring machine, The new energy consumption per unit volume. To allow for energy consumption per unit volume, The damage rate of the new pile body, To normalize the forecast uncertainty, These are the weighting coefficients.

[0025] This invention ranks different cutterhead parameters, cutter arrangement, and grinding pile position schemes by setting a comprehensive score, providing a basis for construction scheme comparison, cutterhead parameter pre-evaluation, and high torque risk identification. The smaller the comprehensive score, the better the corresponding grinding pile scheme is in terms of cutting safety, economy, crushing effect, and predictive reliability.

[0026] Secondly, the present invention also provides a shield tunneling pile cutting performance evaluation system based on a working condition partitioning proxy model, comprising: a processor, an input device, an output device, and a memory, wherein the processor, the input device, the output device, and the memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the shield tunneling pile cutting performance evaluation method based on a working condition partitioning proxy model as described in any of the preceding aspects of the present invention.

[0027] The present invention provides a shield tunneling pile cutting performance evaluation system based on a working condition partition proxy model. This system is compact, stable, highly integrated, and simple in construction. It can stably execute the shield tunneling pile cutting performance evaluation method based on a working condition partition proxy model provided in the preceding part of the present invention, further enhancing the overall applicability and practical application capability of the present invention. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart illustrating a method for evaluating the cutting performance of a tunnel boring machine pile based on a working condition partitioning proxy model, according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the numerical simulation of shield tunneling pile cutting according to an embodiment of the present invention; Figure 3 This is a diagram of the proxy model structure according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a shield tunneling pile cutting performance evaluation system based on a working condition partitioning proxy model, according to an embodiment of the present invention. Detailed Implementation

[0030] Specific embodiments of the present invention will now be described in detail. It should be noted that the embodiments described herein are for illustrative purposes only and are not intended to limit the invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be apparent to those skilled in the art that these specific details are not necessary to practice the invention. In other instances, well-known circuits, software, or methods have not been specifically described to avoid obscuring the invention.

[0031] Throughout this specification, references to "an embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in connection with that embodiment or example is included in at least one embodiment of the invention. Therefore, the phrases "in an embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily refer to the same embodiment or example. Furthermore, specific features, structures, or characteristics can be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. Moreover, those skilled in the art will understand that the illustrations provided herein are for illustrative purposes and are not necessarily drawn to scale.

[0032] It should be noted in advance that, in one alternative embodiment, except for independent descriptions, the same symbols or letters appearing in all formulas have the same meaning.

[0033] In one optional embodiment, please refer to Figure 1 This invention provides a method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model. The method includes the following steps: S1. Obtain the basic dataset of shield tunneling pile grinding conditions and the constraint set for cutting performance evaluation.

[0034] This example uses a tunnel boring machine (TBM) section of an urban rail transit system adjacent to the pile foundation of an existing high-speed railway bridge as an example. In a localized area, the TBM line is in close proximity to the existing high-speed railway bridge pile foundations, with some bridge piles located within the sweeping influence range of the TBM cutterhead. Before on-site construction, it is necessary to determine whether the TBM cutterhead will experience problems such as excessive cutting torque, high energy consumption per unit volume, insufficient pile breakage, or excessive response from adjacent bridge piles at different pile grinding positions.

[0035] In this embodiment, the working conditions to be evaluated are divided into three categories: orthogonal grinding piles, offset grinding piles, and edge grinding piles. Orthogonal grinding piles indicate that the center of the cutterhead and the center of the bridge pile are nearly coincident; offset grinding piles indicate that there is a certain lateral offset between the center of the cutterhead and the center of the bridge pile; and edge grinding piles indicate that the bridge pile is located at the edge of the cutterhead sweep range.

[0036] A thorough investigation was conducted to obtain basic data on the working conditions of the shield tunneling pile grinding operation, including: shield equipment and cutterhead parameters: cutterhead diameter, cutterhead area, cutterhead opening area, cutterhead rotation speed range, cutter installation radius, cutter type, number of cutters, and cutter spatial arrangement; bridge pile geometric parameters: pile diameter, pile length, pile top elevation, pile bottom elevation, pile spatial coordinates, and relative position between the pile and the shield tunnel axis; pile material parameters: concrete strength grade, concrete axial compressive strength, elastic modulus, damage evolution parameters, steel reinforcement grade, steel reinforcement diameter, and steel reinforcement... Spacing and reinforcement ratio; relative relationship between shield and pile: lateral offset distance between cutterhead center and pile center, spatial angle between shield tunneling direction and pile axis, overlap range between cutterhead rotation sweep area and pile cross-section; numerical simulation parameters: contact parameters, material constitutive model, mesh size, calculation step size, cutterhead rotation boundary, cutting path, damage criteria and energy output rules; engineering control requirements: maximum allowable cutting torque, allowable peak torque, cutterhead jamming risk control requirements, unit volume energy consumption control requirements, and pile damage degree evaluation requirements.

[0037] The main data used in this embodiment are shown in Table 1. The diameter of the shield cutterhead, the cutterhead opening ratio, and the cutter arrangement are from the shield equipment and cutterhead design data; the diameter of the bridge pile, the concrete strength, and the reinforcement ratio are from the bridge pile design data; and the offset distance between the shield cutterhead and the bridge pile is calculated from the plane relationship of the line and the coordinates of the bridge pile.

[0038] Table 1 Basic data for shield tunneling pile grinding operation

[0039] Furthermore, the set of constraints for evaluating the cutting performance of shield tunneling piles is clarified, including: the cutterhead cutting torque does not exceed the upper limit of the allowable torque of the shield equipment; the peak cutting torque does not exceed the preset overload control threshold. The overload control threshold is based on the allowable torque of the tunnel boring machine. With safety control coefficient Confirmed, satisfied Safety control coefficient Based on the performance of the tunnel boring machine and construction safety requirements, the value can be taken as 0.85 to 0.95; in this embodiment, 0.90 is used. The allowable energy consumption per unit volume does not exceed the construction economic control threshold. The construction economic control threshold is determined based on the cutting energy consumption level of similar projects, preliminary simulation or test results, and construction economic control requirements. In this embodiment, the allowable energy consumption per unit volume is taken as 22 MJ / m³.3 The pile damage rate meets the target breakage level required for construction; the cutterhead opening ratio, cutter arrangement density, and effective cutting trajectory meet the equipment construction constraints; the pile eccentricity and contact coverage are within the coverage range of the parametric simulation sample library; when the new working condition exceeds the coverage range of the sample library or the prediction uncertainty exceeds the threshold, a fine cutting simulation review should be triggered.

[0040] S2. Construct a parametric cutting simulation model based on the basic dataset and the constraint set.

[0041] Specifically, in this embodiment, a parametric cutting simulation model is constructed based on the basic dataset and the set of evaluation constraints, using finite element software, discrete element software, material point method programs, or other numerical simulation methods that can describe the interaction between the cutterhead and the pile.

[0042] The parametric cutting simulation model includes at least: a shield cutterhead geometry model, used to characterize the shield cutterhead diameter, opening area, cutter position, and effective radius of the cutter; a bridge pile geometry model, used to characterize the pile size, spatial position of the pile, and its relative relationship with the shield cutterhead; a pile material model, used to characterize concrete damage and the reinforcement effect of steel bars; a contact cutting model, used to describe the contact, friction, breakage, and energy dissipation processes between the cutter, cutterhead, and pile; and post-processing extraction rules, used to automatically extract the cutting torque time history, total cutting work, failure volume, and damage volume from the simulation results.

[0043] Parametric cutting simulation models can automatically complete geometric reconstruction, material assignment, contact definition, mesh generation, cutting calculation, and result extraction based on different parameter combinations.

[0044] S3. Determine the set of main control input parameters, and construct a shield tunneling pile cutting simulation sample library based on the set of main control input parameters and the parameterized cutting simulation model. Step S3 includes the following sub-steps: S31. Determine the set of main control input parameters.

[0045] Specifically, in this embodiment, based on the shield tunneling pile cutting mechanism, the pile-to-cutterhead dimensional ratio, pile eccentricity, cutterhead opening ratio, and equivalent cutter arrangement coefficient are obtained. In order to enhance the characterization ability of eccentric pile grinding and edge pile grinding conditions, this embodiment introduces the pile contact coverage rate, and simultaneously introduces the equivalent deformation modulus of the stratum and the equivalent bending stiffness of the bridge pile.

[0046] The pile-to-cutterhead dimensional ratio is used to characterize the geometric dimensional relationship of the pile relative to the cutterhead.

[0047] In the formula, The pile-to-cutterhead dimensional ratio, The diameter of the bridge pier. The diameter of the shield cutterhead.

[0048] The eccentricity of the grinding pile is used to characterize the grinding conditions of orthogonal grinding, eccentric grinding, and edge grinding.

[0049] In the formula, For the eccentricity of the grinding pile, This is the lateral offset distance between the center of the cutterhead and the center of the pile. The diameter of the shield cutterhead.

[0050] The cutterhead aperture ratio is used to characterize the slag discharge space and structural permeability of the cutterhead.

[0051] In the formula, The cutterhead aperture ratio, The area of ​​the cutter head opening. This represents the total area of ​​the cutter head.

[0052] Dimensionless normalized equivalent tool arrangement coefficient:

[0053] In the formula, This is the equivalent tool arrangement coefficient. The number of cutting tools involved in the grinding process is determined by the contact range of the cutter head. The cutting capability weight for the j-th type of tool, with a value ranging from 0 to 1, is obtained through rock-breaking tests. The effective cutting width of the j-th type of tool is taken from the tool design parameters. The effective cutting trajectory length of the j-th tool in the contact area of ​​the grinding pile is obtained by geometric intersection calculation. The lever arm correction factor for the j-th tool is calculated from the ratio of the mounting radius to the maximum radius of the tool head. This is the nominal cutting cross-sectional area of ​​the cutter head.

[0054] The contact coverage of the grinding pile can enhance the characterization of eccentric grinding pile and edge grinding pile conditions:

[0055] In the formula, For the contact coverage of the grinding pile, This represents the overlap area between the area swept by the rotating cutterhead and the cross-section of the bridge pier. This represents the cross-sectional area of ​​the bridge pier.

[0056] Equivalent bending stiffness of bridge piles:

[0057] In the formula, The equivalent bending stiffness of the bridge pier. The equivalent elastic modulus of the bridge pile. Let be the moment of inertia of the bridge pile section.

[0058] Furthermore, the set of master control input parameters for rapid evaluation is determined:

[0059] In the formula, The pile-to-cutterhead dimensional ratio, For the eccentricity of the grinding pile, For the contact coverage of the grinding pile, For concrete strength, For reinforcement ratio, The cutterhead aperture ratio, This is the equivalent tool arrangement coefficient. The equivalent deformation modulus of the formation. This is the equivalent bending stiffness of the bridge pile.

[0060] S32. Based on the set of main control input parameters and various sampling methods, different parameter combinations are obtained.

[0061] Specifically, in this embodiment, different parameter combinations are generated using Latin hypercube sampling, orthogonal experimental sampling, uniform sampling, or typical engineering working condition sampling methods, based on the value range of the master control input parameter set: This embodiment selects three typical working conditions for illustration, and the combination of input parameters is shown in Table 2.

[0062] Table 2 Parameter combinations for three typical operating conditions

[0063] It can be seen that the cutterhead makes the best contact with the bridge pile in working condition 1, there is obvious eccentricity in working condition 2, and working condition 3 is cutting at the edge of the cutterhead with a low contact coverage rate. At the same time, the concrete strength and reinforcement ratio are high, which is a relatively unfavorable pile grinding working condition.

[0064] S33. Input the parameter combination into the parameterized cutting simulation model to obtain a set of cutting performance output indicators.

[0065] Specifically, in this embodiment, a corresponding automated parametric simulation program is developed. This automated parametric simulation program can be implemented based on the built-in command tools, script interfaces, remote call interfaces of numerical simulation software, or a self-developed calculation program.

[0066] The functions of the parameterized simulation automation program include: automatically generating geometric models of the shield cutterhead and bridge piles based on the combination of input parameters; automatically assigning material parameters based on concrete strength and reinforcement ratio; automatically generating the cutterhead cutting area based on the cutterhead opening ratio and equivalent tool arrangement coefficient; automatically setting contact relationships, boundary conditions, rotational cutting process, and calculation step size; automatically executing cutting simulation calculations; automatically extracting cutting torque time history, total cutting work, damage volume, and failure volume; and automatically writing the calculation results into the shield pile cutting simulation sample library.

[0067] In this embodiment, the parameterized simulation automation program automatically completes the following operations: generating the spatial relationship between the cutterhead and the bridge pile, setting orthogonal, offset, or edge grinding pile conditions, assigning parameters to concrete and steel reinforcement materials, setting the cutterhead rotation, advance, and contact conditions, extracting cutting torque, cutting energy consumption, pile damage, and bridge pile response, and writing the calculation results into the shield tunneling pile grinding cutting simulation sample library.

[0068] The parameterized simulation automation program can input parameter combinations into the formation of a simulation sample library for shield tunneling pile cutting.

[0069] Furthermore, based on the parameterized simulation automation program, each set of parameters is combined... Input a parametric cutting simulation model, automatically complete the simulation calculation for the corresponding shield tunneling pile grinding condition, extract cutting performance output indicators, and obtain a set of cutting performance output indicators:

[0070] In the formula, This is the set of cutting performance output indicators corresponding to the parameter combinations. This represents the average cutting torque corresponding to the parameter combination. The peak cutting torque corresponding to the parameter combination. The energy consumption per unit volume corresponding to the parameter combination. The damage rate of the pile body corresponding to the parameter combination.

[0071] Among them, the average cutting torque is used to evaluate the overall cutting load level:

[0072] In the formula, The average cutting torque, To stabilize the cutting stage, This represents the cutting torque time history.

[0073] Peak cutting torque is used to evaluate the risk of instantaneous overload and jamming.

[0074] In the formula, This is the peak cutting torque. This represents the statistical time interval for peak cutting torque. This represents the cutting torque time history.

[0075] Energy consumption per unit volume, used to evaluate cutting efficiency and construction costs:

[0076] In the formula, Energy consumption per unit volume The total mechanical work done throughout the entire cutting process. The volume of material that is cut and stripped from the original pile body by the tool when the pile body reaches the pile body failure threshold is geometrically equal to the volume of solid material completely removed from the pile body after the effective cutting trajectory of the tool sweeps. The pile body failure threshold is the critical mechanical state in which the stress of the pile body material reaches the uniaxial ultimate compressive strength and the overall bearing capacity of the pile body completely fails.

[0077] Pile damage rate is used to evaluate the degree of damage caused by cutting within the pile body.

[0078] In the formula, For pile damage rate, To meet the pile damage threshold, the total volume of the damaged area is geometrically the volume of the entire spatial region where cutting force induces mechanical degradation and crack propagation. The target pile volume participating in the cutting is geometrically defined as the initial volume of the complete pile segment to be processed within the cutting stroke coverage area of ​​the cutterhead. It is obtained by multiplying the original cross-sectional area of ​​the pile by the cutting axial coverage length. The pile damage threshold is the critical state in which the stiffness attenuation of the pile body in the cutting disturbance area is not less than 15% of the initial stiffness and the internal microcracks enter the stable propagation stage.

[0079] S34. Store the parameter combination input-cutting performance output index sample pair as the shield tunneling pile cutting simulation sample library.

[0080] Specifically, in this embodiment, all input-output sample pairs are automatically stored as a simulation sample library for shield tunneling pile cutting:

[0081] In the formula, This serves as a sample library for shield tunneling pile cutting simulation. For parameter combinations, This is the set of cutting performance output indicators corresponding to the parameter combinations.

[0082] This embodiment generated a total of 520 candidate parameter combinations, of which 480 sets completed effective simulation calculations. The remaining 40 sets were marked as abnormal or high-risk samples due to severe contact instability, significantly excessive peak torque, or insufficient pile breakage. The final shield tunneling pile cutting simulation sample library includes 480 sets of effective input-output samples. A simulation dataset was constructed using the cutting values ​​from this library; please refer to [link to relevant documentation]. Figure 2 .

[0083] S4. Using the aforementioned shield tunneling pile cutting simulation sample library, construct and train a working condition partitioned proxy model. Step S4 includes the following sub-steps: S41. Using the shield tunneling pile cutting simulation sample library, construct the working condition partition proxy model.

[0084] Specifically, in this embodiment, the shield tunneling pile cutting simulation sample library is used to construct a working condition partitioning proxy model. The working condition partitioning proxy model includes an input feature enhancement module, an explicit interactive modeling module, a gated expert partitioning module, a residual fusion backbone module, a multi-task output module, and an uncertainty assessment module.

[0085] The input feature enhancement module is used to map low-dimensional input parameters to a high-dimensional feature space.

[0086] In the formula, For the first One enhancement feature, For parameter-specific nonlinear mapping networks, For the first One input parameter.

[0087] The explicit interactive modeling module is used to construct interactive features between parameters, including scale ratio-eccentricity interactive features, strength-reinforcement ratio interactive features, and cutterhead opening ratio-equivalent cutter arrangement coefficient interactive features, to express the multi-factor coupling effect during shield tunneling pile grinding:

[0088] In the formula, For the first The first enhancement feature and the first Explicit interactive features comprised of several enhancement features. For the first One enhancement feature, For the first One enhancement feature.

[0089] The gated expert partitioning module is deeply integrated with the physical mechanism of tunnel boring machine (TBM) pile grinding. Based on the pile contact morphology and cutting condition characteristics, it automatically assigns weights to each expert sub-model. The gated expert partitioning module is used to distinguish at least the response modes of orthogonal pile grinding, eccentric pile grinding, edge pile grinding, and high-resistance pile grinding.

[0090] In the formula, For the first A specialist sub-model center grinds the piles. For the first An expert sub-model for eccentric grinding piles For the first An expert sub-model edge grinding piling Let be the nonlinear feature mapping function corresponding to the k-th expert sub-model, and have a total of An expert sub-model, The shared operating condition feature vector is used as the input to the gating expert partitioning module. For gating networks, For the first The weights of each expert sub-model The center of the k-th expert sub-model is the grinding pile.

[0091] The residual fusion backbone module is used to fuse input enhancement features, explicit interactive features, and gated expert features:

[0092] In the formula, This is a comprehensive feature vector formed by fusing input enhancement features, explicit interactive features, and expert fusion features; The input feature enhancement module provides the enhanced feature set. The interaction feature vector obtained by the explicit interaction modeling module. Output the expert fusion features obtained by weighting and fusing according to the gating weights for each expert sub-model.

[0093] And deep feature extraction is performed using residual structures:

[0094] In the formula, Let l be the input feature vector of the l-th residual layer. This is the nonlinear feature transformation function corresponding to the l-th residual layer, used for further extraction and mapping of input features. For the first The output feature vector obtained after processing by each residual layer is used as the input feature of the next residual layer.

[0095] The multi-task output module is used to synchronously output the average cutting torque, peak cutting torque, energy consumption per unit volume, and pile damage rate.

[0096] The uncertainty assessment module is used to output the uncertainty corresponding to each forecast indicator:

[0097] In the formula, For the first The prediction uncertainty variance corresponding to each cutting performance output index For the first Uncertainty estimation function for each output index, This is the final feature vector obtained by the residual fusion backbone module after multi-layer residual feature extraction.

[0098] The working condition zoning proxy model internally includes orthogonal pile grinding experts, offset pile grinding experts, edge pile grinding experts, and high-resistance pile grinding experts, enabling the model to select different local response laws for different pile grinding working conditions. This setting corresponds to the engineering characteristics of this project where different pile grinding methods have different effects on cutterhead stress and bridge pile response. For the structure of the working condition zoning proxy model, please refer to [link to relevant documentation]. Figure 3 .

[0099] S42. Construct the target loss function. Step S42 includes the following sub-steps: S421. Construct physical consistency constraint losses including concrete strength, reinforcement ratio, cutterhead structural parameters, and edge grinding pile mutations.

[0100] Specifically, in this embodiment, a physical consistency constraint loss is constructed, including concrete strength, reinforcement ratio, cutterhead structural parameters, and abrupt changes in edge grinding piles.

[0101] The monotonicity loss of concrete strength and the constraint cutting torque are monotonically non-increasing with the concrete compressive strength.

[0102] In the formula, For the monotonic loss of concrete strength, To constrain cutting torque, This refers to the compressive strength of concrete.

[0103] If the model predicts a situation that violates the laws of physics, such as an increase in concrete strength or an abnormal rise in cutting torque, then a penalty will be imposed for the loss of monotonicity in concrete strength.

[0104] The positive monotonicity loss of the reinforcement ratio constrains the peak cutting torque and the energy consumption per unit volume, which are monotonically non-decreasing with the reinforcement ratio.

[0105]

[0106] In the formula, For the positive monotonicity loss of reinforcement ratio, This is the peak cutting torque. Energy consumption per unit volume This represents the reinforcement ratio.

[0107] If the reinforcement ratio increases while the two indicators decrease abnormally, the positive monotonicity loss of the reinforcement ratio will result in a penalty.

[0108] The monotonicity loss of the cutter head structural parameters constrains the unit volume energy consumption to monotonically and non-decreasing with the cutter head opening ratio and the equivalent tool arrangement coefficient.

[0109]

[0110] In the formula, For the monotonicity loss of the cutter head structural parameters, Energy consumption per unit volume The cutterhead aperture ratio, This is the equivalent tool arrangement coefficient.

[0111] If the cutterhead opening ratio or equivalent tool arrangement coefficient decreases, or the specific energy consumption decreases abnormally, then the monotonicity loss of the cutterhead structural parameters will result in a penalty.

[0112] Edge grinding pile abrupt change constraint loss: A range of edge grinding pile eccentricity is defined, and the gradient magnitude of the peak cutting response within this range is constrained to be no less than the abrupt change threshold.

[0113] In the formula, For edge grinding pile abrupt change constraint loss, As an indicator function, the constraint only takes effect when the eccentricity falls within the edge grinding interval. For the edge grinding pile eccentricity range, The mutation threshold, This represents the peak cutting response within the eccentricity range of the edge grinding pile.

[0114] If the peak cutting response gradient within the interval does not reach the mutation threshold, the edge grinding pile mutation constraint loss will incur a penalty.

[0115] Furthermore, the physical prior laws of shield tunneling pile grinding are transformed into quantifiable gradient penalty terms, and a physical consistency constraint loss function is constructed. The total loss is the weighted sum of the losses of each constraint term:

[0116] In the formula, For the loss due to physical consistency constraints, For the monotonic loss of concrete strength, For the positive monotonicity loss of reinforcement ratio, For the monotonicity loss of the cutter head structural parameters, For edge grinding pile abrupt change constraint loss, These are the weight hyperparameters for each sub-constraint.

[0117] S422. Based on the physical consistency constraint loss, construct the target loss function.

[0118] Specifically, in this embodiment, based on the physical consistency constraint loss, and combined with the multi-task regression loss and uncertainty constraint loss, a target loss function is constructed:

[0119] In the formula, Loss to the target For the first The regression loss for each output metric, As task weight, For the loss due to physical consistency constraints, The weights for the loss due to physical consistency constraints. Loss due to uncertainty constraints This represents the weight of the loss due to uncertainty constraints.

[0120] S43. Divide the shield tunneling pile cutting simulation sample library into a training set, a validation set, and a test set, and train the working condition partition proxy model in combination with the target loss function.

[0121] Specifically, in this embodiment, the shield tunneling pile cutting simulation sample library is divided into a training set, a validation set, and a test set, and a working condition partition proxy model is trained in combination with the objective loss function.

[0122] S44. Verify the working condition partition agent model.

[0123] Specifically, in this embodiment, the trained working condition partitioning proxy model is verified through cross-validation, error statistics, and typical working condition verification to ensure that it has stable predictive ability within the training sample coverage area.

[0124] S5. Based on the aforementioned working condition partitioning agent model, obtain the new shield tunneling pile grinding working condition.

[0125] Specifically, in this embodiment, the input to the working condition partitioning agent model is:

[0126] In the formula, The pile-to-cutterhead dimensional ratio, For the eccentricity of the grinding pile, For the contact coverage of the grinding pile, For concrete strength, For reinforcement ratio, The cutterhead aperture ratio, This is the equivalent tool arrangement coefficient. The equivalent deformation modulus of the formation. This is the equivalent bending stiffness of the bridge pile.

[0127] The output of the working condition partitioning agent model is:

[0128] In the formula, This is the predicted value of the average cutting torque. This is the predicted peak cutting torque value. This is the predicted energy consumption per unit volume. This is the predicted value of pile damage rate. This is the predicted value of the maximum horizontal displacement of the bridge pier. This represents the predicted maximum settlement of the bridge piers. This is the predicted value of the maximum bending moment of the bridge pier.

[0129] Input the three sets of typical working conditions into the trained working condition partitioning proxy model to obtain the predicted values ​​of the new shield tunneling pile grinding working conditions. Please refer to Table 3.

[0130] Table 3 Predicted values ​​for the new shield tunneling pile grinding conditions

[0131] S6. Perform a rapid evaluation of the new shield tunneling pile grinding condition, and output a shield tunneling pile grinding performance evaluation report after simulation verification. Step S6 includes the following sub-steps: S61. Calculate the new parameter combination using the new shield tunneling pile grinding condition.

[0132] Specifically, in this embodiment, the new parameter combination is calculated using the predicted values ​​of the new shield tunneling pile grinding conditions:

[0133] In the formula, The ratio of new pile to cutterhead dimensions. For the eccentricity of the new grinding pile, For the new concrete strength, For the new reinforcement ratio, For the new cutter head opening ratio, This is the new equivalent tool arrangement coefficient.

[0134] S62. Input the new parameter combination into the working condition partition proxy model to predict and output a new set of cutting performance output indicators and the uncertainty of each indicator.

[0135] Specifically, in this embodiment, the new parameter combination is input into the trained and calibrated working condition partitioning proxy model to quickly obtain a set of predicted new cutting performance output indicators:

[0136] In the formula, This is a set of new cutting performance output indicators. For the new average cutting torque, For the new peak cutting torque, The new energy consumption per unit volume. This represents the damage rate of the new pile body.

[0137] Simultaneously output the corresponding uncertainty index:

[0138] In the formula, This is the set of uncertainty indicators corresponding to the predicted cutting performance indicators for the new shield tunneling pile grinding conditions. The standard deviation of the uncertainty of the new average cutting torque. The standard deviation of the uncertainty of the new peak cutting torque. The standard deviation of the uncertainty in the new unit volume energy consumption. This represents the standard deviation of the uncertainty in the damage rate of the new pile.

[0139] S63. Based on the new set of cutting performance output indicators, assess whether the current working condition is a low-risk, medium-risk, or high-risk grinding condition.

[0140] Specifically, in this embodiment, based on the new set of cutting performance output indicators, the current working condition is assessed as a low-risk, medium-risk, or high-risk grinding condition: When the new peak cutting torque Exceeding the allowable torque of the tunnel boring machine When the condition is determined to be a high-torque risk condition; when the new unit volume energy consumption Energy consumption per unit volume exceeding the allowable limit When the damage rate of the new pile body is high, it is determined to be a high-energy-consumption working condition; Damage rate below the target pile body When the maximum horizontal displacement of the bridge pier is predicted, it is determined to be an insufficient cutting condition; Not exceeding the maximum allowable horizontal displacement of the bridge pile and the predicted maximum settlement of the bridge pile If the settlement does not exceed the maximum allowable settlement of the bridge pile, the response of the adjacent bridge pile is considered to be controllable; if any prediction result is close to the allowable threshold, or the prediction uncertainty exceeds the threshold, a fine simulation review is triggered.

[0141] In this embodiment, the peak cutting torque of working condition 1 is lower than the allowable torque of the shield tunneling equipment, the energy consumption per unit volume is lower than the allowable energy consumption per unit volume, the pile damage rate is higher than the target pile damage rate, and the bridge pile displacement and settlement are small. Therefore, it is determined to be a controllable working condition.

[0142] In Condition 2, both the peak cutting torque and energy consumption per unit volume were higher than in Condition 1, and the horizontal displacement and settlement of the bridge piers also increased. Specifically, the peak cutting torque was 4860 kN·m, lower than the overload control threshold of 5400 kN·m set in this embodiment; the energy consumption per unit volume was 20.6 MJ / m. 3 Below the allowable energy consumption per unit volume of 22 MJ / m 3 The pile damage rate was 0.74, higher than the target pile damage rate of 0.70. Therefore, the main cutting performance indicators have not yet exceeded the corresponding control thresholds. However, since the peak cutting torque and energy consumption per unit volume are significantly higher than in condition 1, and the relevant indicators are close to the control thresholds, this condition is identified as a condition requiring attention. For this type of condition, it is recommended to appropriately optimize the feed speed, cutter head speed, and tool arrangement before construction, and to strengthen the monitoring of bridge piles.

[0143] The peak cutting torque in working condition 3 reaches 5680 kN·m, close to the allowable torque of 6000 kN·m for tunnel boring machines; the energy consumption per unit volume is 25.2 MJ / m³. 3 This exceeds the allowable energy consumption per unit volume of 22 MJ / m³ set in this embodiment. 3 The pile damage rate was 0.61%, lower than the target pile damage rate of 0.70; meanwhile, the maximum horizontal displacement and settlement of the bridge piles were significantly higher than the previous two working conditions. Therefore, this working condition was determined to be one requiring verification.

[0144] According to the above rules, Condition 1 can be directly used as a feasible pile grinding scheme; Condition 2 requires stronger control over construction parameters and monitoring arrangements; Condition 3 should enter the fine simulation verification process.

[0145] S64. Establish a refined simulation review mechanism.

[0146] Specifically, in this embodiment, to improve the prediction reliability of the working condition partitioning proxy model in critical working conditions and sparse sample regions, a refined simulation verification mechanism is established. This refined simulation verification mechanism differs from the working condition risk level evaluation. The working condition risk level evaluation is mainly determined based on the relationship between the cutting performance prediction results and the corresponding engineering control thresholds, while the refined simulation verification mechanism further considers whether the prediction results are close to the control thresholds, prediction uncertainty, and the coverage of the working condition to be evaluated in the shield tunneling pile cutting simulation sample library.

[0147] When the new shield tunneling pile grinding condition meets any of the following conditions, a detailed simulation verification is triggered: the predicted peak cutting torque reaches or exceeds the overload control threshold. The predicted energy consumption per unit volume reaches or exceeds the allowable energy consumption per unit volume. The predicted pile damage rate is lower than the target pile damage rate. The normalized forecast uncertainty of any forecasting indicator exceeds the uncertainty control threshold. Alternatively, the normalized distance between the input parameters corresponding to the working condition to be evaluated and the nearest neighbor sample in the shield tunneling pile cutting simulation sample library exceeds the sample coverage threshold. .

[0148] In this embodiment, the energy consumption per unit volume is allowed. Take 22 MJ / m 3 The uncertainty control threshold is determined by calibration based on the predicted uncertainty distribution on the validation set, and can be taken as the 95th quantile of the normalized predicted uncertainty of the validation set. The sample coverage threshold is determined by calculation based on the nearest neighbor distance distribution within the simulation sample library, and can be taken as the 95th quantile of the normalized nearest neighbor distance of each valid sample in the sample library. Therefore, both the uncertainty control threshold and the sample coverage threshold are obtained by statistical calculation from existing sample data, rather than being arbitrarily set.

[0149] S65. If the new shield tunneling pile grinding condition meets any of the conditions in the fine simulation verification mechanism, then the fine simulation verification mechanism is triggered.

[0150] Specifically, in this embodiment, if the new shield tunneling pile grinding condition meets any condition of the fine simulation verification mechanism, the fine simulation verification mechanism is triggered.

[0151] S66. Re-input the new shield tunneling pile grinding working condition that triggers the review into the parameterized cutting simulation model, obtain the reviewed input-output sample pair, and store it in the shield tunneling pile grinding cutting simulation sample library.

[0152] Specifically, in this embodiment, the new shield tunneling pile grinding conditions that trigger the review are re-input into the parametric cutting simulation model, and the parametric simulation automation program performs fine simulation calculations. The reviewed input-output sample pairs are then written into the shield tunneling pile grinding cutting simulation sample library.

[0153] S67. Based on the new set of cutting performance output indicators, calculate the comprehensive score and rank the optional construction schemes.

[0154] Specifically, in this embodiment, for multiple candidate construction schemes, a comprehensive score is calculated based on the new peak cutting torque, new energy consumption per unit volume, and new pile damage rate to rank the available construction schemes:

[0155] In the formula, For comprehensive scoring, For the new peak cutting torque, For the allowable torque of the tunnel boring machine, The new energy consumption per unit volume. To allow for energy consumption per unit volume, The damage rate of the new pile body, To normalize the forecast uncertainty, These are the weighting coefficients.

[0156] Among them, normalized prediction uncertainty The following calculations were made based on the uncertainties corresponding to the three prediction indicators involved in the comprehensive scoring: peak cutting torque, energy consumption per unit volume, and pile damage rate:

[0157] In the formula, To normalize the forecast uncertainty, The standard deviation of the uncertainty corresponding to the predicted value of the new peak cutting torque. For the allowable torque of the tunnel boring machine, The standard deviation of the uncertainty in the new unit volume ratio energy consumption prediction. To allow for energy consumption per unit volume, The standard deviation of the uncertainty in the predicted damage rate of the new pile is given. The target pile damage rate.

[0158] The lower the overall score, the better the corresponding grinding pile solution is in terms of cutting safety, economy, crushing effect and prediction reliability.

[0159] S68. Output the shield tunneling pile cutting performance evaluation report, which includes the new parameter combination, the new set of cutting performance output indicators, the uncertainty of each indicator, the working condition risk level, whether fine simulation verification is required, and the ranking of the optional construction schemes.

[0160] Based on the rapid assessment results of the new shield tunneling pile grinding conditions, a shield tunneling pile cutting performance assessment report is output. The report should include at least: new parameter combinations, new average cutting torque prediction, new peak cutting torque prediction, new unit volume energy consumption prediction, new pile damage rate prediction, uncertainty of each prediction value, working condition risk level, whether detailed simulation verification is required, and ranking of optional construction schemes.

[0161] In one optional embodiment, please refer to Figure 4The present invention also provides a shield tunneling pile cutting performance evaluation system based on a working condition partition proxy model. This system includes a processor, an input device, an output device, and a memory. The processor, input device, output device, and memory are interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to call the program instructions to execute relevant steps in a relevant embodiment of the shield tunneling pile cutting performance evaluation method based on a working condition partition proxy model.

[0162] This invention provides a shield tunneling pile cutting performance evaluation system based on a working condition partitioning proxy model. The functional components can be integrated into a single processing unit, exist as separate physical entities, or be integrated into a single unit. These integrated components can be implemented in hardware or as software functions.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model, characterized in that, include: Obtain the basic dataset of shield tunneling pile grinding conditions and the constraint set for cutting performance evaluation; Based on the aforementioned basic dataset and the aforementioned constraint set, a parametric cutting simulation model is constructed. Determine the set of main control input parameters, and construct a shield tunneling pile cutting simulation sample library based on the set of main control input parameters and the parameterized cutting simulation model; Using the aforementioned shield tunneling pile cutting simulation sample library, a working condition partition proxy model was constructed and trained; Based on the aforementioned working condition partitioning agent model, a new shield tunneling pile grinding working condition is obtained; A rapid evaluation of the new shield tunneling pile grinding conditions is conducted, and a shield tunneling pile cutting performance evaluation report is output after simulation verification.

2. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 1, characterized in that, The step of constructing a parametric cutting simulation model based on the basic dataset and the constraint set includes: The parameterized cutting simulation model is constructed based on numerical simulation and in combination with the basic dataset and the constraint set.

3. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 1, characterized in that, The construction of the shield tunneling pile cutting simulation sample library based on the main control input parameter set and the parameterized cutting simulation model includes: Based on the set of main control input parameters and various sampling methods, different parameter combinations are obtained; Input the parameter combination into the parameterized cutting simulation model to obtain a set of cutting performance output indicators; The parameter combination input and cutting performance output index sample pairs are stored as the shield tunneling pile cutting simulation sample library.

4. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 1, characterized in that, The process of constructing and training a working condition partitioned proxy model using the shield tunneling pile cutting simulation sample library includes: The working condition partition proxy model is constructed using the shield tunneling pile cutting simulation sample library. Construct the target loss function; The shield tunneling pile cutting simulation sample library is divided into a training set, a validation set, and a test set, and the working condition partition proxy model is trained in combination with the target loss function; Verify the described working condition partition agent model.

5. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 4, characterized in that, The construction of the working condition partition agent model includes: Construct the working condition partitioning agent model, which includes the gating expert partitioning module; The gated expert partitioning module completes the weight allocation of expert sub-models based on the contact morphology and cutting condition characteristics of the grinding piles. The expert sub-models correspond to the physical conditions of the center, eccentric, and edge grinding piles, respectively.

6. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 4, characterized in that, The constructed target loss function includes: Construct physical consistency constraint losses including concrete strength, positive reinforcement ratio, cutterhead structural parameters, and abrupt changes in edge grinding piles; Based on the physical consistency constraint loss, the target loss function is constructed.

7. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 1, characterized in that, The rapid assessment of the new shield tunneling pile grinding conditions includes: Calculate the new parameter combination using the new shield tunneling pile grinding condition; The new parameter combination is input into the working condition partition proxy model to predict and output a new set of cutting performance output indicators and the uncertainty of the indicators; Based on the new set of cutting performance output indicators, the current working condition is assessed as a low-risk, medium-risk, or high-risk grinding condition.

8. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 7, characterized in that, The simulation verification output of the shield tunneling pile cutting performance evaluation report includes: Establish a sophisticated simulation verification mechanism; If the new shield tunneling pile grinding condition meets any of the conditions in the fine simulation verification mechanism, then the fine simulation verification mechanism is triggered; The new shield tunneling pile grinding working condition that triggers the review is re-input into the parameterized cutting simulation model to obtain the reviewed input-output sample pair and store it in the shield tunneling pile cutting simulation sample library. Based on the new set of cutting performance output indicators, a comprehensive score is calculated to rank the available construction schemes. The output includes the new parameter combination, the new set of cutting performance output indicators, the uncertainty of the indicators, the working condition risk level, whether fine simulation verification is required, and the ranking of the optional construction schemes, which constitutes the shield tunneling pile cutting performance evaluation report.

9. The method for evaluating the cutting performance of shield tunneling piles based on a working condition partitioning proxy model according to claim 8, characterized in that, The step of calculating a comprehensive score and ranking the available construction schemes based on the new set of cutting performance output indicators includes: The new set of cutting performance output indicators includes the new peak cutting torque, the new energy consumption per unit volume, and the new pile damage rate. The comprehensive score satisfies the following expression: In the formula, For comprehensive scoring, For the new peak cutting torque, For the allowable torque of the tunnel boring machine, The new energy consumption per unit volume. To allow for energy consumption per unit volume, The damage rate of the new pile body, To normalize the forecast uncertainty, These are the weighting coefficients.

10. A shield tunneling pile cutting performance evaluation system based on a working condition partitioning proxy model, characterized in that, include: The system includes a processor, an input device, an output device, and a memory, all interconnected. The memory stores a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute a shield tunneling pile cutting performance evaluation method based on a working condition partitioning proxy model, as described in any one of claims 1 to 9.

Images