A method for coupled analysis of cutterhead mud cake and tunneling parameters

Abstract

This invention discloses a method for coupled analysis of cutterhead mud cake and tunneling parameters, belonging to the field of intelligent cutterhead control technology. It aims to solve the problems of delayed early warning of mud cake formation on tunnel boring machine (TBM) cutterheads and significant blindness in tunneling parameter adjustment. The method involves constructing a dynamic digital twin cutterhead model based on a 3D cutterhead model; generating representative working points in a 3D parameter space; obtaining a set of simulation results; extracting a mud cake severity index; and constructing a standard simulation result library containing multiple parameters and the mud cake severity index. Based on actual working environment parameters and cutter wear ranges, a set of representative working points is retrieved and matched. Preliminary target tunneling parameters are generated by combining the mud cake severity index with similarity. The cutterhead is driven to operate, and the mud cake severity index is quantified in real time. A three-level correlation feature is constructed, and the probability of mud cake formation is predicted based on its correlation with a probability mapping library. When the probability exceeds a threshold, the preliminary target tunneling parameters are dynamically adjusted. This achieves advanced early warning of cutterhead mud cake risk and precise adaptation of tunneling parameters.

Description

Technical Field

[0001] This invention relates to the field of intelligent cutterhead control technology, and more specifically to a method for coupled analysis of cutterhead mud cake and tunneling parameters. Background Technology

[0002] In the field of tunnel boring machine construction, the cutterhead is the core component for cutting soil, and its operating status directly affects construction efficiency and safety. Mud cake formation on the cutterhead is a frequent technical problem in complex strata. At best, it can cause a sharp drop in tunneling speed and a surge in equipment energy consumption; at worst, it can cause the cutterhead to jam, forcing a shutdown for maintenance, increasing construction costs and project time risks.

[0003] Existing cutterhead simulation models are mostly based on idealized settings and rely on fixed cutter states, failing to fully consider the evolution of cutter wear during long-term cutting. This results in limited simulation accuracy for actual cutting conditions and makes it difficult to reflect the soil cutting and flow characteristics at different wear stages. Furthermore, the characterization of the construction environment is relatively simplistic, focusing primarily on a few physical parameters of the strata, while neglecting the comprehensive influence of hydrological conditions, spatial orientation, and other factors, thus failing to fully capture the cohesion and retention patterns of soil in complex environments. The methods for assessing the risk of mud cake formation are also relatively simple, relying mainly on threshold alarms for single parameters such as torque and thrust, lacking consideration of soil retention time and cohesion range. The integrated assessment of multiple dimensions of features leads to delayed and inaccurate risk identification, making it difficult to achieve early warning. The matching mechanism for tunneling parameters lacks flexibility; traditional simulation result libraries only cover limited preset working conditions. When actual geological parameters or cutter conditions exceed the preset range, it is difficult to quickly generate suitable parameter schemes, and parameter adjustments rely heavily on manual experience, failing to achieve real-time optimization in response to dynamic changes in working conditions. Simultaneously, the integration between simulation models and actual construction data is low, lacking an effective feedback mechanism to feed field data back to model updates. This results in a gradual decline in model adaptability as construction progresses, making it difficult to continuously support parameter optimization in complex geological formations. These problems mean that existing technologies are largely reactive in dealing with cutterhead mud cake in complex geological formations, lacking optimization capabilities and hindering the efficiency and safety of shield tunneling. Therefore, to overcome these limitations, this invention proposes a coupled analysis method for cutterhead mud cake and tunneling parameters. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a coupling analysis method for cutterhead mud cake and tunneling parameters. This method solves the technical problems of delayed early warning of cutterhead mud cake in complex geological formations, lack of precise basis for adjusting tunneling parameters, and difficulty in adapting to dynamically changing working environments and cutter wear conditions, resulting in high risk of cutterhead mud cake and low tunneling efficiency.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for coupled analysis of cutterhead mud cake and tunneling parameters, comprising:

[0007] A basic geometric model of a digital twin cutterhead is constructed based on the three-dimensional model of the cutterhead. A tool wear calculation module is embedded in the basic geometric model. The tool wear calculation module is used to update the tool wear amount according to the tool wear parameters to form a dynamic digital twin cutterhead model.

[0008] A three-dimensional parameter space for the cutterhead is constructed to generate representative working points of the cutterhead. A set of simulation results of representative working points is generated based on the dynamic digital twin cutterhead model. These results are used to extract cutterhead mud cake parameters, quantify the mud cake degree index, and construct a standard simulation result library that includes working environment parameters, tunneling parameters, tool wear, and mud cake degree index.

[0009] Based on the actual working environment parameters of the cutterhead and the updated tool wear range, the hierarchical index structure of the standard simulation result library is used for layer-by-layer retrieval. Based on the stability of the working environment parameters, a set of representative working conditions is matched to generate preliminary target tunneling parameters by combining the mud cake degree index and similarity.

[0010] Based on the initial target tunneling parameters, the cutterhead is driven to operate, the mud cake degree index is quantified in real time, and a three-level correlation feature is constructed. This feature is used to predict the probability of mud cake in the future prediction period based on the correlation rules between the three-level correlation feature and the probability mapping relationship library. When the probability of mud cake is greater than the mud cake probability threshold, the initial target tunneling parameters are dynamically adjusted.

[0011] Specifically, the steps for generating preliminary target tunneling parameters include:

[0012] Extract the mud cake degree index and tunneling parameters corresponding to the representative working point from the representative working point set, configure the mud cake degree threshold, screen out the representative working point with the mud cake degree index less than the mud cake degree threshold, and construct the initial candidate parameter set;

[0013] When the current working environment is determined to be stable, the target mud cake degree index is determined based on the mud cake degree index of the representative working point in the initial candidate parameter set, so as to locate the target representative working point. The corresponding tunneling parameters are obtained based on the target representative working point as the preliminary target tunneling parameters.

[0014] When the current working environment is determined to be unstable, the midpoint of the fluctuation range of the current fluctuating environmental parameters is used as the reference point, and combined with the current non-fluctuating environmental parameters, a comprehensive working environment parameter is constructed.

[0015] Calculate the distance between the working environment parameters and the comprehensive working environment parameters of each representative working point in the initial candidate parameter set, and set the fitting weight of the representative working points based on the distance values;

[0016] According to the fitting weight, the tunneling parameters of each representative working point in the initial candidate parameter set are weighted and fitted to obtain the comprehensive tunneling parameters, which are used as the initial target tunneling parameters.

[0017] The initial target tunneling parameters are input into the dynamic digital twin cutterhead model for simulation verification. If the mud cake degree index output by the dynamic digital twin cutterhead model is less than the preset verification threshold, the initial target tunneling parameters are confirmed to be valid; otherwise, the verification is deemed to have failed, and the fitting early warning mechanism is triggered.

[0018] Specifically, the steps for predicting the probability of mud cake formation within a future prediction period include:

[0019] Configure the prediction cycle, collect current working environment parameters, tunneling parameters, tool wear, and quantify the real-time mud cake degree index based on the cutterhead mud cake parameters, and construct a three-level correlation feature, which includes: parameter coordination feature, trend evolution feature and risk warning feature;

[0020] At the beginning of each prediction cycle, the probability prediction process for mud cake formation is triggered. The time range of the current prediction cycle is defined based on the start point of the current prediction cycle and the preset prediction cycle duration.

[0021] Configure a probability mapping relationship library between three-level correlation features and mud cake probability. The probability mapping relationship library is constructed based on the correspondence between the combination of three-level correlation feature values ​​and the actual mud cake probability.

[0022] The parameter coordination feature value, trend evolution feature value, and risk warning feature value are used as input dimensions. Each dimension is divided into feature levels according to the numerical range. The combination of feature levels corresponds to the preset mud cake probability benchmark value in the probability mapping relationship library.

[0023] Three-level correlation feature values ​​are extracted, the feature level to which each correlation feature value belongs is determined, and feature level combinations are formed. The corresponding mud cake probability benchmark value is matched according to the feature level combination, and the mud cake probability benchmark value is adjusted by real-time correction coefficient to obtain the mud cake probability of the prediction period. The real-time correction coefficient is calculated based on the deviation between the current working environment parameters and the working environment parameters under the same feature level combination in history.

[0024] Specifically, the parameter coordination feature is used to reflect the degree of matching between working environment parameters, tunneling parameters and tool wear. It is obtained by calculating the adaptation coefficient between working environment parameters and tunneling parameters, and combining the coupling index between tool wear and tunneling parameters.

[0025] Trend evolution characteristics are used to capture the dynamic change pattern of the actual mud cake degree index. They are obtained by extracting the slope, fluctuation amplitude and mean drift of the actual mud cake degree index within the sliding window.

[0026] Risk warning features are used to characterize the growth trend of mud cake risk under the current parameter combination. They are obtained by calculating the difference rate between the actual mud cake degree index and the preset safety threshold, combined with the fluctuation rate of working environment parameters.

[0027] Specifically, the steps for dynamically adjusting the initial target tunneling parameters include:

[0028] Retrieve the correlation data table of working environment parameters, tunneling parameters, tool wear and mud cake degree index from the standard simulation result library;

[0029] Based on the mapping slope data table of each tunneling parameter and the mud cake degree index under the current working environment parameters in the standard simulation result library, the initial adjustment weights of each tunneling parameter are set in descending order of the values ​​in the mapping slope data table; by monitoring the rate of change of the trend evolution characteristic value in real time, the initial adjustment weights are linearly corrected according to the rate of change.

[0030] The deviation between the simulated mud cake degree index and the actual mud cake degree index is calculated. The deviation value is multiplied by the adjustment weight to obtain the weighted deviation value. The ratio of the weighted deviation value to the current value of the initial target tunneling parameter is determined as the single parameter adjustment ratio. The product of the single parameter adjustment ratio and the current value of the corresponding tunneling parameter is used as the target adjustment amount of the corresponding tunneling parameter.

[0031] The initial target tunneling parameters are adjusted in steps according to the target adjustment amount, and the interval between two adjacent adjustment steps is set according to the risk warning characteristic value.

[0032] During the adjustment process, real-time data of the actual mud cake degree index and working environment parameters are continuously collected. The change trend of the actual mud cake degree index before and after the adjustment is used to quantify the improvement range.

[0033] Configure the evaluation period. If the improvement within the evaluation period is less than the preset threshold after the tunneling parameters are adjusted, the target adjustment amount is recalculated and an adjustment warning is issued; otherwise, the multi-dimensional mapping relationship table in the standard simulation result library is updated through clustering algorithm.

[0034] Specifically, the steps for updating the tool wear amount based on tool wear parameters include:

[0035] When the tool turret is in operation, tool wear parameters are collected to construct a tool wear parameter set.

[0036] After preprocessing the tool wear parameter set, a tool wear model is constructed. The tool wear model reflects the cutting intensity by multiplying the formation hardness coefficient and the tool pressure, and reflects the cumulative effect of the cutting action by combining the product of the contact time and the cutting line velocity, so as to calculate the tool wear increment per unit time.

[0037] The cumulative tool wear is obtained by integrating the tool wear increment per unit time; the actual wear detection value sequence collected by the tool wear monitoring device is obtained; based on the collection time of the actual wear detection value sequence, the cumulative tool wear calculated by the corresponding tool wear model is obtained; and the coefficient weights in the tool wear model are dynamically adjusted by the deviation correction algorithm.

[0038] The calibrated cumulative tool wear is fed back to the tool wear calculation module to update the tool wear in the dynamic digital twin tool head model.

[0039] Specifically, the steps for constructing a standard simulation results library include:

[0040] The range of working environment parameters of the cutterhead, the combination of tunneling parameters and the tool wear range are obtained, a three-dimensional parameter space is constructed, and representative working points are generated in the three-dimensional parameter space by using a stratified sampling method.

[0041] For each representative working condition, simulations were repeated based on a dynamic digital twin toolhead model to obtain a set of simulation results;

[0042] Extract cutterhead cake parameters from the simulation results set to form the original cutterhead cake sequence;

[0043] The original cutterhead cake sequence was time-series aligned, and a filtering algorithm was used to remove abnormal fluctuation data to construct a standard cutterhead cake sequence.

[0044] The cluster centers of each cutterhead cake parameter in the standard cutterhead cake sequence are calculated and used as typical cutterhead cake parameters to construct a typical cutterhead cake sequence.

[0045] Specifically, the steps for constructing the standard simulation results library also include:

[0046] Set the cake threshold for each cutterhead cake parameter, and calculate the relative deviation between each typical cutterhead cake parameter and the corresponding cake threshold in a typical cutterhead cake sequence.

[0047] The influence weights of each typical cutterhead cake formation parameter are set, and the cake formation degree index is calculated by weighted summation based on the relative deviation and the influence weights.

[0048] Based on the tool wear range, typical cutterhead mud cake sequences are classified, and corresponding working environment parameters, tunneling parameters and mud cake degree index are associated to construct a multi-dimensional mapping table;

[0049] By configuring a hierarchical index structure, the multi-dimensional mapping relationship table is stored in the standard simulation result library.

[0050] Specifically, the steps for matching a representative set of operating conditions include:

[0051] Collect actual working environment parameters and actual tool wear parameters in the cutterhead construction area;

[0052] Configure the fluctuation assessment period, obtain the working environment parameter sequence within the fluctuation assessment period, calculate the fluctuation amplitude of each working environment parameter sequence, and respond to whether the current working environment is stable.

[0053] When the current working environment is determined to be stable, the tool wear amount is updated according to the current actual tool wear parameters, the tool wear range is located, and the simulation subset is obtained by first-level filtering according to the tool wear range in the standard simulation result library.

[0054] In the simulation subset, the similarity between the current actual working environment parameters and the working environment parameters of the representative working points in the simulation subset is calculated. Representative working points with similarity greater than a preset similarity threshold are selected to construct a set of representative working points.

[0055] Specifically, the steps for matching a representative set of operating conditions also include:

[0056] When the current working environment is determined to be unstable, if the fluctuation amplitude of the working environment parameter sequence is greater than the corresponding fluctuation threshold, it is marked as a fluctuating environment parameter, and its fluctuation range is obtained based on the working environment parameter sequence of the fluctuating environment parameter; otherwise, it is marked as a non-fluctuating environment parameter, and the current actual working environment parameter is used as a fixed parameter.

[0057] Update the tool wear amount and locate the tool wear range based on the current actual tool wear parameters, and select the simulation subset from the standard simulation result library;

[0058] In the simulation subset, the similarity between the current non-fluctuating environment parameters and the non-fluctuating environment parameters of the representative working points of the simulation subset is calculated. Initial representative working points with similarity greater than a preset similarity threshold are selected to construct an initial representative working point set.

[0059] Select representative operating point sets from the initial representative operating point set that are within the fluctuation range of the current fluctuation environment parameter.

[0060] The beneficial effects of this invention are:

[0061] This invention addresses the problems of delayed early warning of mud cake formation on tunnel boring machine (TBM) cutterheads, lack of precise basis for adjusting tunneling parameters, and difficulty in adapting to dynamic working environments and tool wear conditions. It constructs a dynamic digital twin cutterhead model with an embedded tool wear calculation module, combining representative working point data generated by stratified sampling with a standard simulation result library of multi-dimensional mappings to achieve accurate simulation of the actual cutterhead state, solving the problem of the disconnect between the static model and actual working conditions. Through a differentiated working condition matching strategy for stable and unstable working environments, combined with a probability prediction mechanism of three-level correlation features and feature level combinations, it achieves advanced and accurate early warning of mud cake probability, solving the problem of delayed warnings. Relying on parameter adjustment logic based on mapping slope weighting and dynamic correction, combined with iterative optimization of the standard simulation result library, it ensures that tunneling parameters can adapt to dynamic environments and tool wear changes in real time, solving the problem of blind parameter adjustment, thereby effectively reducing the risk of cutterhead mud cake formation and improving tunneling efficiency and construction stability. Attached Figure Description

[0062] Figure 1 This is a flowchart of a method for coupling analysis of cutterhead mud cake and tunneling parameters according to the present invention;

[0063] Figure 2 This is a flowchart illustrating the specific steps of updating the tool wear amount based on tool wear parameters according to the present invention.

[0064] Figure 3 A flowchart outlining the specific steps involved in constructing the standard simulation result library of this invention;

[0065] Figure 4 This is a flowchart illustrating the specific steps involved in matching a representative set of operating conditions according to the present invention.

[0066] Figure 5 This is a flowchart illustrating the specific steps involved in generating preliminary target tunneling parameters according to the present invention.

[0067] Figure 6 This is a flowchart illustrating the specific steps involved in determining whether to dynamically adjust the initial target tunneling parameters according to the present invention. Detailed Implementation

[0068] Please see Figure 1 This embodiment introduces a method for coupled analysis of cutterhead mud cake and tunneling parameters, including:

[0069] Step S1: Construct a basic geometric model of the digital twin cutterhead based on the 3D model of the cutterhead. Embed a tool wear calculation module into the basic geometric model to update the tool wear amount according to the tool wear parameters, forming a dynamic digital twin cutterhead model that can dynamically reflect the actual state of the cutterhead. For the construction area involved by the cutterhead, quantify the working environment parameters of the cutterhead, construct a 3D parameter space of the cutterhead, and generate representative working condition points of the cutterhead. Generate a set of simulation results based on the dynamic digital twin cutterhead model. Through coupled simulation of discrete element method and computational fluid dynamics, simulate the motion state and interaction force of soil particles under the cutting action of the cutterhead under different combinations of working environment parameters and different combinations of tunneling parameters. Extract the mud cake parameters of the cutterhead and quantify the mud cake degree index. After multi-condition simulation, form a standard simulation result library containing the correspondence between working environment parameters, tunneling parameters, tool wear amount, and mud cake degree index.

[0070] In this embodiment, the tool wear calculation module updates the tool wear amount based on tool wear parameters, including the contact time between the cutterhead and the stratum, the stratum hardness coefficient, the pressure on the tool, and the cutting speed. Geological survey data, hydrological monitoring data, and engineering design data of the construction area involved in the cutterhead are collected, transforming the physical characteristics of the stratum, hydrological monitoring features, and spatial conditions of tunnel construction into quantifiable working environment parameters. The discrete element method is used to simulate the motion state and interaction forces of soil particles under the cutting action of the cutterhead during operation. Combined with fluid dynamics calculations, the flow characteristics of the soil within the space between the cutterhead and the soil chamber are analyzed. The calculation results of both are coupled to construct a simulation engine that reflects the entire process of soil movement from cutting to slag removal. Inputting tunneling parameters such as cutterhead rotation speed, thrust, torque, amendment injection rate, and foam mixing ratio into the simulation engine, the engine's internal algorithms calculate and output parameters related to soil cake formation, including the soil residence time at the cutterhead opening, the ratio of actual slag discharge to theoretical cutting volume per unit time, the percentage of the cutterhead surface area adhered to by soil, and the ratio of the difference between real-time torque and initial torque. These parameters characterize the soil flow state and soil cake formation risk trend in the cutterhead region. By selecting core indicators characterizing the soil cake formation state, the weight coefficients of each indicator in the comprehensive evaluation are determined. Based on the ratio of the measured value of each indicator to its corresponding threshold, combined with the weight coefficients, a weighted calculation is performed to obtain a soil cake formation degree index that quantifies the risk of soil cake formation. The index range covers the entire range from no soil cake formation risk to severe soil cake formation. Based on the range of geological parameters, combinations of tunneling parameters, and cutterhead wear status involved in the cutterhead operation, a simulation engine is used to perform multi-condition simulation calculations. The correspondence between working environment parameters, tunneling parameters, cutter wear amount, and mud cake degree index under different working conditions is recorded. These correspondences are classified and stored according to different intervals of cutter wear amount, forming a standard simulation result library containing multi-dimensional mapping relationships.

[0071] Please see Figure 2Preferably, the specific steps for updating the tool wear amount based on tool wear parameters include:

[0072] When the cutter head is in operation, tool wear parameters are collected, including the contact time between the cutter head and the formation during operation, the formation hardness coefficient obtained by geological exploration equipment, the real-time tool pressure during the tool cutting process of the cutter head using a pressure sensor, and the cutting linear velocity calculated based on the cutter head rotation speed and the tool installation radius to construct a tool wear parameter set.

[0073] The collected tool wear parameter set is preprocessed, and a filtering algorithm is used to remove abnormal fluctuation data and perform time series alignment to ensure the stability and consistency of the input parameters.

[0074] Based on the theory of material wear, a tool wear model is constructed. The tool wear model reflects the cutting intensity by multiplying the formation hardness coefficient and the tool pressure, and reflects the cumulative effect of the cutting action by combining the product of the contact time and the cutting linear velocity, thereby calculating the tool wear increment per unit time.

[0075] The cumulative tool wear is obtained by integrating the incremental tool wear per unit time. The actual wear detection value sequence collected by the tool wear monitoring device is obtained, and the cumulative tool wear calculated by the corresponding tool wear model is obtained according to the collection time of the actual wear detection value. The coefficient weights in the tool wear model are dynamically adjusted by the deviation correction algorithm so that the deviation between the calculated cumulative tool wear and the actual wear detection value is less than the preset wear deviation threshold, thereby further improving the accuracy of wear calculation.

[0076] The calibrated cumulative tool wear is fed back to the tool wear calculation module, which updates the tool wear in the dynamic digital twin cutterhead model in real time. This enables the dynamic digital twin cutterhead model to accurately reflect the actual wear state of the cutterhead at different tunneling stages, providing reliable cutterhead state input for subsequent multi-condition simulation of the simulation engine, and effectively improving the authenticity and credibility of the cutterhead mud cake risk assessment.

[0077] Please see Figure 3 Preferably, the specific steps for constructing a standard simulation results library include:

[0078] Based on historical construction data and geological survey reports, typical working environment parameter ranges, tunneling parameter combinations, and tool wear intervals of the cutterhead are obtained, and a three-dimensional parameter space is constructed. By configuring parameter boundaries and variation gradients, the parameter space is ensured to cover various working conditions that may be encountered in actual construction, providing a comprehensive input basis for subsequent multi-condition simulations. Specifically, the tool wear interval is divided into initial interval, light wear interval, moderate wear interval, and heavy wear interval by setting wear thresholds for different tool types.

[0079] To address the heterogeneity of soil cutting by the tunnel boring machine cutterhead, a stratified sampling method is used to generate representative working points in a three-dimensional parameter space. For each representative working point, based on a dynamic digital twin cutterhead model, a coupled simulation engine of discrete element method and computational fluid dynamics is used to simulate the complete process of cutterhead rotation cutting, soil breaking, amendment mixing, and soil chamber slag discharge. The simulation is repeated for each representative working point to eliminate errors caused by soil discreteness, and a set of simulation results for each stable representative working point is obtained.

[0080] Extract cutterhead mud cake parameters from the simulation results set of each representative working point, including soil residence time at the cutterhead opening, proportion of adhesive area on the cutterhead surface, slag discharge efficiency, torque growth rate and uniformity of amendment mixing, etc., which are directly related to mud cake, to form the original cutterhead mud cake sequence.

[0081] The original cutterhead mud cake sequence was time-series aligned by configuring a sliding window that matches the cutterhead rotation cycle. The original cutterhead mud cake sequences from different simulations were aligned to the same time dimension to ensure coverage of the complete cutterhead cutting cycle. A density-based filtering algorithm was used to remove abnormal fluctuation data caused by soil dispersion in the original cutterhead mud cake sequence, while retaining effective information reflecting the mud cake trend and improving sequence stability to construct a standard cutterhead mud cake sequence.

[0082] The cluster centers of the cutterhead cake parameters of each cutterhead cake sequence in the standard cutterhead cake sequence are calculated as their typical cutterhead cake parameters. Typical cutterhead cake sequences for each representative working point are constructed to ensure that the common cake characteristics under the representative working points can be characterized.

[0083] Based on the cutterhead cake parameters in historical cases of cutterhead cake failure, we statistically analyzed the critical values ​​of each cutterhead cake parameter that cause cake formation under different working conditions. In this way, we set the cake formation threshold for each cutterhead cake parameter, such as the soil retention time threshold at the cutterhead opening and the percentage threshold of the adhesive area on the cutterhead surface.

[0084] For each typical cutterhead cake parameter in a typical cutterhead cake sequence, the relative deviation between the parameter and the corresponding cake threshold is calculated to obtain a dimensionless deviation value, thereby achieving comparability between different parameters.

[0085] Based on the influence of each cutterhead cake parameter on cutterhead cake failure, the influence weights of each typical cutterhead cake parameter are set using the analytic hierarchy process and historical cases of cutterhead cake failure. Based on the relative deviations of each typical cutterhead cake parameter and their corresponding influence weights, a cake severity index is calculated by weighted summation: the relative deviation of each parameter is multiplied by its influence weight and then summed. The resulting index value ranges from 0 to 1, thus achieving a quantitative characterization of cake risk.

[0086] Based on the tool wear range, the typical cutterhead mud cake sequence of each representative working condition point is classified, and the corresponding working environment parameters, tunneling parameters and mud cake degree index are associated to construct a multi-dimensional mapping relationship table.

[0087] By configuring a hierarchical index structure, the multi-dimensional mapping relationship table is stored in the standard simulation result library, ensuring that similar working conditions can be quickly matched during subsequent searches, and providing accurate reference for adjusting actual construction parameters.

[0088] Step S2: Real-time acquisition of actual working environment parameters in the cutterhead construction area and acquisition of actual tool wear parameters. Inputting the actual tool wear parameters into the tool wear calculation module to update the tool wear amount, thereby adjusting the tool state of the dynamic digital twin cutterhead model; Based on the actual working environment parameters and the updated tool wear amount, a representative set of working conditions is retrieved and matched in the standard simulation result library through a hierarchical index structure to extract the corresponding tunneling parameters; Combining the actual parameters and the retrieved working condition parameters, preliminary target tunneling parameters are generated. These parameters can adapt to the current actual working environment and tool wear state, providing basic parameter references for the safe and efficient tunneling of the cutterhead.

[0089] In this embodiment, actual working environment parameters are collected in real time through an on-site sensor array, covering the real-time clay content of the strata in the construction area, groundwater pressure, soil moisture content, and spatial constraints. The collection frequency is synchronized with the cutterhead rotation cycle to ensure that the parameters dynamically reflect the real-time changes in the construction environment. Actual tool wear parameters are obtained through a tool wear monitoring system, including the actual contact time between the cutterhead and the current stratum, the real-time cutting pressure of the tool, and the actual cutting linear velocity calculated based on the cutterhead rotation speed. This is synchronously correlated with the real-time stratum hardness coefficient fed back by geological exploration equipment to form a complete set of actual tool wear parameters. The actual tool wear parameter set is input into the tool wear calculation module, and the cumulative tool wear is updated according to the tool wear amount model. The updated wear amount is then synchronized to the dynamic digital twin cutterhead model, ensuring that the tool size decay and surface morphology changes in the model are consistent with the actual cutterhead, providing an accurate tool state benchmark for parameter matching. Based on actual working environment parameters and updated tool wear range, a hierarchical indexing structure of the standard simulation result library is used to search layer by layer to select a set of representative working conditions that are highly similar to the actual working conditions. The tunneling parameter combinations corresponding to these working conditions are extracted as initial tunneling parameters, including reference values ​​for cutterhead speed, propulsion force, modifier injection rate and foam mixing ratio.

[0090] Please see Figure 4 Preferably, the specific steps for matching a representative set of operating conditions include:

[0091] Collect actual working environment parameters of the cutterhead construction area, including the physical properties, hydrological characteristics and spatial constraints of the formation, and obtain actual tool wear parameters, covering the contact time between the cutterhead and the formation, the pressure on the tool, the cutting speed and the formation hardness coefficient, to ensure that the parameters can reflect the real state of the current construction.

[0092] The construction environment may experience sudden parameter changes due to abrupt geological shifts, groundwater pressure fluctuations, etc. Using a uniform matching logic would lead to insufficient matching accuracy under different working environment parameters. To address this, a fluctuation assessment period is configured, and the working environment parameter sequence within that period is obtained. The fluctuation amplitude of each working environment parameter sequence is calculated. If the fluctuation amplitude of any working environment parameter sequence exceeds a preset fluctuation threshold, the current working environment parameter is determined to be unstable; otherwise, the current working environment is considered stable. This solves the problem of a single matching strategy caused by differences in environmental characteristics, providing a basis for subsequent scenario-based matching and ensuring that the set of working point values ​​is appropriately matched to different environmental states. The fluctuation threshold is defined based on the normal fluctuation range of working environment parameters under stable working conditions, ensuring effective differentiation between normal fluctuations and abnormal sudden changes in environmental parameters.

[0093] When the current working environment is determined to be stable, the tool wear amount is updated based on the actual tool wear parameters to locate the corresponding tool wear interval. In the standard simulation result library, a primary filtering is performed according to the tool wear interval to obtain a subset of simulations consistent with the current tool wear state. Then, within this subset, the similarity between the current working environment parameters and the working environment parameters of representative working points in the simulation subset is calculated using the Euclidean distance algorithm. Representative working points with similarity greater than a preset similarity threshold are selected to construct a set of representative working points. Since real-time working environment parameters and simulation working point parameters may not correspond one-to-one, the working environment parameters of representative working points in the simulation subset are located based on the currently collected working environment parameters, and then similarity calculations are performed to accurately pinpoint the set of representative working points closest to the current working environment and tool state.

[0094] When the current working environment is determined to be unstable, some working environment parameters will fluctuate significantly, while others will remain relatively constant. If the fluctuation range of the working environment parameter sequence is greater than the corresponding fluctuation threshold, the working environment parameter is marked as a fluctuating environment parameter, and its fluctuation range is obtained based on the working environment parameter sequence of the fluctuating environment parameter; otherwise, it is marked as a non-fluctuating environment parameter, and the current actual working environment parameter is used as the fixed parameter.

[0095] The tool wear amount is updated based on the current actual tool wear parameters, and the tool wear range is located. The simulation results are then filtered in the standard simulation result library to obtain a simulation subset. Within the simulation subset, the similarity between the current non-fluctuating environment parameters and the non-fluctuating environment parameters of the representative working points of the simulation subset is calculated using the Euclidean distance algorithm. Initial representative working points with similarity greater than a preset similarity threshold are selected to construct an initial representative working point set. By calculating the similarity of non-fluctuating environment parameters, working points that match the basic conditions can be selected even when the parameters do not correspond completely, avoiding fundamental adaptation errors caused by non-fluctuating environment parameter mismatch.

[0096] Select representative operating point sets from the initial representative operating point set that are within the fluctuation range of the current fluctuation environment parameter.

[0097] Please see Figure 5 Preferably, the specific steps for generating preliminary target tunneling parameters include:

[0098] Extract the mud cake degree index and tunneling parameters corresponding to each representative working point from the representative working point set. Configure the mud cake degree threshold based on the historical data cluster analysis in the standard simulation result library, and screen out representative working points with mud cake degree index less than the mud cake degree threshold to construct the initial candidate parameter set. If the screening result is empty, the mud cake degree threshold is reduced by the preset adjustment step size and the screening is repeated until the initial candidate parameter set contains at least one set of representative working points.

[0099] When the current working environment is determined to be stable with minimal parameter fluctuations, priority should be given to ensuring low mud cake risk in the tunneling parameters. Simultaneously, when the degree of mud cake formation is similar, selecting parameters that best match the actual environment can improve adaptability and avoid efficiency reduction or mud cake formation due to parameter-environment deviations. Based on the mud cake degree index of representative working points in the initial candidate parameter set, a target mud cake degree index is determined to locate the target representative working point. The corresponding tunneling parameters are then obtained from the target representative working point as preliminary target tunneling parameters. For example, the representative working point with the smallest mud cake degree index is selected from the initial candidate parameter set. If multiple representative working points with the same mud cake degree index exist, the similarity between these representative working points and the current working environment parameters is calculated using the Euclidean distance algorithm. The tunneling parameters corresponding to the representative working point with the highest similarity are selected as the preliminary target tunneling parameters.

[0100] When the current working environment is determined to be unstable, a single parameter value is difficult to represent the actual environment. Taking the midpoint of the fluctuation range of the current fluctuating environment parameter as the benchmark point, and combining it with the current non-fluctuating environment parameter, a comprehensive working environment parameter is constructed to balance the uncertainty brought about by the fluctuation and provide a reasonable reference point for subsequent parameter fitting.

[0101] The distance between the working environment parameters and the comprehensive working environment parameters of each representative working point in the initial candidate parameter set is calculated using the Euclidean distance algorithm. Based on the distance values, the fitting weights of the representative working points are set, and corresponding fitting weight coefficients are assigned to each representative working point in the initial candidate parameter set.

[0102] According to the fitting weight, the tunneling parameters of each representative working point in the initial candidate parameter set are weighted and fitted. Each tunneling parameter component is multiplied by its corresponding weight coefficient and then summed to obtain the comprehensive tunneling parameters, which serve as the initial target tunneling parameters. Through weighted fitting, the advantages of multiple highly adaptable working points can be integrated to generate comprehensive parameters that can cover the fluctuation range, thus avoiding parameter bias.

[0103] The initial target tunneling parameters are input into the dynamic digital twin cutterhead model for simulation verification. If the mud cake degree index output by the dynamic digital twin cutterhead model is less than the preset verification threshold, the initial target tunneling parameters are confirmed to be valid. Otherwise, the verification is deemed to have failed, triggering a fitting early warning mechanism to indicate that the standard simulation result library may have insufficient coverage of matching working point or accuracy deviation, and suggesting supplementary simulation or parameter calibration. At the same time, the deviation value between the mud cake degree index of each representative working point in the initial candidate parameter set and the corresponding index of the comprehensive tunneling parameters is calculated. Working points with deviation values ​​greater than the fitting anomaly threshold are removed to update the initial candidate parameter set. Weighted fitting and verification are performed again until valid parameters are generated or the early warning is completed.

[0104] Step S3: Transmit the initial target tunneling parameters to the cutterhead drive assembly to drive the cutterhead in tunneling operations. Simultaneously, collect the cutterhead mud cake parameters in real time and quantify the real-time mud cake degree index. Extract the actual mud cake degree index sequence within a continuous time period and construct a three-level correlation feature based on working environment parameters and tool wear. Predict the mud cake probability within the future prediction period based on the correlation rules of the three-level correlation feature and the probability mapping relationship library. If the predicted mud cake probability is greater than the preset mud cake probability threshold, extract the simulated mud cake degree index corresponding to the initial target tunneling parameters and calculate the deviation between the simulated mud cake degree index and the actual mud cake degree index. Based on this deviation, dynamically adjust the initial target tunneling parameters to make the actual mud cake degree index approach the low-risk range, thus obtaining the final target tunneling parameters. Enter the actual working environment parameters, real-time cutterhead wear, final target tunneling parameters, and corresponding actual mud cake degree indices into the standard simulation result library. Analyze the correlation features between the new data and the existing data in the standard simulation result library to update the mapping relationship within the standard simulation result library.

[0105] In this embodiment, by real-time acquisition of cutterhead sludge cake parameters, the actual sludge cake degree index is quantified, achieving dynamic monitoring and quantitative characterization of the cutterhead sludge cake state. A sequence of actual sludge cake degree indices over continuous time periods is extracted, and a three-level correlation feature is constructed by integrating working environment parameters and tool wear. Based on the correlation rules between the three-level correlation feature values ​​and a probability mapping relationship library, the probability of sludge cake formation in future prediction periods is predicted. The probability mapping relationship library is constructed based on the correspondence between combinations of three-level correlation feature values ​​and actual sludge cake probabilities in historical data. By matching the feature level combination with the benchmark value and applying real-time correction coefficients, the timeliness and accuracy of the prediction are improved. Accuracy: When the predicted probability of mud cake formation exceeds the mud cake probability threshold, the simulated mud cake severity index corresponding to the initial target tunneling parameters is called, and its deviation from the actual value is calculated. Based on the deviation, the initial target tunneling parameters are dynamically adjusted so that the actual mud cake severity index quickly converges to the low-risk range. Actual working environment parameters, real-time cutterhead wear, final target tunneling parameters, and the corresponding actual mud cake severity index are stored together. Through clustering algorithms and association rule mining, the multi-dimensional mapping relationship table in the standard simulation result library is updated, realizing dynamic iteration of the standard simulation result library and improving the accuracy of subsequent parameter matching and the cutterhead's adaptive capability. Simultaneously, based on the graded threshold of the actual mud cake severity index, graded prevention operations are triggered. In low-risk situations, parameters are kept stable and periodic flushing is performed; in medium-risk situations, the amendment injection ratio is increased and the cutterhead rotation is controlled in reverse; in high-risk situations, advance is paused and the cutterhead is rotated alternately in conjunction with flushing. Through the combination of graded operations and dynamic adjustments, proactive prevention and timely intervention against cutterhead mud cake formation are strengthened.

[0106] Please see Figure 6 Preferably, the specific steps for determining whether to dynamically adjust the initial target tunneling parameters include:

[0107] Configure the prediction period to define the prediction time range of the probability of mud cake formation. This prediction period needs to be adapted to the rhythm of the cutterhead tunneling operation, covering a sufficiently long period of time to capture the cumulative trend of mud cake formation risk, while also meeting the timeliness requirements for real-time adjustment.

[0108] By collecting current working environment parameters, tunneling parameters, and tool wear, and based on cutterhead mud cake parameters, a real-time mud cake degree index is quantified, and a three-level correlation feature is constructed:

[0109] The primary correlation feature is the parameter coordination feature, which reflects the degree of matching between working environment parameters, tunneling parameters, and tool wear. It is obtained by calculating the adaptation coefficient between working environment parameters and tunneling parameters, combined with the coupling index between tool wear and tunneling parameters. Specifically, a standard adaptation matrix between working environment parameters and tunneling parameters is configured, and the deviation of the measured values ​​of the current working environment parameters and tunneling parameters from the standard adaptation matrix is ​​calculated to obtain the adaptation coefficient. A coupling function between tool wear and tunneling parameters is configured, and the current tool wear and tunneling parameters are input to obtain the coupling index. The adaptation coefficient and the coupling index are weighted and summed according to a preset ratio to obtain the parameter coordination feature value.

[0110] The secondary correlation feature is a trend evolution feature, used to capture the dynamic change pattern of the actual mud cake degree index. It is obtained by extracting the slope, fluctuation amplitude, and mean drift of the actual mud cake degree index within a sliding window. Specifically, a sliding window of fixed duration is configured to segment the actual mud cake degree index sequence within a continuous period. The linear regression slope of the actual mud cake degree index sequence within each window is calculated as the slope. The difference between the maximum and minimum values ​​of the index within each window is calculated as the fluctuation amplitude. The difference between the mean index within the current window and the mean index within the previous window is calculated as the mean drift. The slope, fluctuation amplitude, and mean drift are integrated according to preset weights to obtain the trend evolution feature value.

[0111] The third-level correlation feature is a risk warning feature used to characterize the potential growth trend of mud cake risk under the current parameter combination. It is obtained by calculating the difference rate between the actual mud cake degree index and the preset safety threshold, combined with the fluctuation rate of the working environment parameters. Specifically, the difference between the actual mud cake degree index and the preset safety threshold is calculated, and the ratio of this difference to the preset safety threshold is taken as the difference rate; the change of the working environment parameters per unit time is calculated as the fluctuation rate; the difference rate and the fluctuation rate are weighted and summed according to preset coefficients to obtain the risk warning feature value.

[0112] At the beginning of each prediction cycle, the mud cake probability prediction process is triggered, and the currently constructed third-level correlation feature values ​​are called synchronously. Taking the beginning point as the time reference, the time range of the current prediction cycle is defined based on the preset prediction cycle duration, and mud cake probability prediction is carried out for the time range.

[0113] Based on the association rules between three-level correlation feature values ​​and a probability mapping relationship library, a probability mapping relationship library is configured between three-level correlation features and mud cake probabilities. This library is constructed based on the correspondence between combinations of three-level correlation feature values ​​in historical data and actual mud cake probabilities. Parameter coordination feature values, trend evolution feature values, and risk warning feature values ​​are used as input dimensions. Each dimension is divided into several feature levels according to numerical ranges. Different combinations of feature levels correspond to preset mud cake probability benchmark values ​​in the probability mapping relationship library.

[0114] Extract the current three-level correlation feature values, determine the feature level to which each correlation feature value belongs, and form a feature level combination. Based on the feature level combination, match the corresponding mud cake probability benchmark value in the probability mapping relationship library, and introduce a real-time correction coefficient to adjust the mud cake probability benchmark value. The real-time correction coefficient is calculated based on the deviation between the current working environment parameters and the working environment parameters under the same feature level combination in history, and finally obtain the mud cake probability for the prediction period.

[0115] Based on historical data clustering analysis in the standard simulation results library, a mud cake probability threshold is configured. When the mud cake probability in the prediction period exceeds the threshold, it is determined that the initial target tunneling parameters need to be adjusted; otherwise, the initial target tunneling parameters remain unchanged. The determination result serves as the basis for initiating subsequent dynamic adjustment procedures, ensuring that adjustment measures are triggered only when there is a risk of mud cake formation, thus avoiding ineffective adjustments that could affect construction stability.

[0116] Preferably, the specific steps for dynamically adjusting the initial target tunneling parameters include:

[0117] Retrieve the correlation data table of working environment parameters, tunneling parameters, tool wear and mud cake degree index from the standard simulation result library. Based on the current working environment parameters, calculate the actual value range of each tunneling parameter under different working conditions to clarify the adjustable range of the initial target tunneling parameters. Set the upper and lower limits of adjustment for each tunneling parameter to ensure that the adjustment is within the safe operating range of the equipment.

[0118] Based on the mapping slope data table of each tunneling parameter and the mud cake degree index under the current working environment parameters in the standard simulation result library, the initial adjustment weights of each tunneling parameter are set in descending order of the values ​​in the mapping slope data table; and by monitoring the rate of change of the trend evolution characteristic value in real time, the initial adjustment weights are linearly corrected according to the rate of change, and the sum of the adjustment weights of each tunneling parameter after correction is 1.

[0119] The deviation between the simulated mud cake degree index and the actual mud cake degree index is calculated. The deviation value is multiplied by the adjustment weight to obtain the weighted deviation value. The ratio of the weighted deviation value to the current value of the initial target tunneling parameter is determined as the single parameter adjustment ratio. The product of the single parameter adjustment ratio and the current value of the corresponding tunneling parameter is the target adjustment amount of the tunneling parameter.

[0120] The initial target tunneling parameters are adjusted in steps according to the target adjustment amount. Each adjustment amount is determined by the product of the target adjustment amount and the preset step ratio. This is to avoid sudden changes in the cutterhead operating state due to excessive adjustment in a single step, and to ensure the stability of the tunneling process. The interval between two adjacent adjustment steps is set according to the risk warning characteristic value. For example, when the risk warning characteristic value is in the high-risk range, the interval is shorter; when it is in the low-risk range, the interval is longer. This is to accelerate the adjustment response speed to quickly reduce the risk when the risk is high, and reduce the adjustment frequency to maintain the continuity of operation when the risk is low.

[0121] During the adjustment process, real-time data of the actual mud cake degree index and working environment parameters are continuously collected. The improvement range is quantified based on the changing trend of the actual mud cake degree index before and after adjustment. For example, the difference between the actual mud cake degree index at the beginning of adjustment and the actual mud cake degree index at the end of the evaluation period is calculated, and the ratio of this difference to the actual mud cake degree index at the beginning of adjustment is taken as the improvement range. If the index shows a downward trend after adjustment, the improvement range is positive, and the more significant the downward trend, the larger the positive value. If the index does not change significantly or increases, the improvement range is zero or negative.

[0122] An evaluation cycle is configured to centrally analyze the comprehensive effects of adjustment measures, avoiding misjudgments of adjustment effectiveness due to short-term fluctuations. If the improvement in tunneling parameters within the evaluation cycle is less than the preset threshold, the target adjustment amount is recalculated, and an adjustment warning is issued to prompt operators to check the parameter association logic or equipment operating status. Otherwise, the adjusted tunneling parameters, along with the current working environment parameters, real-time cutterhead wear, and mud cake index, are updated using a clustering algorithm to create a multi-dimensional mapping relationship table in the standard simulation result library. This enables dynamic iteration of the standard simulation result library, improving subsequent parameter matching accuracy and cutterhead self-adaptation capabilities.

[0123] Preferably, the specific steps for triggering graded prevention operations based on the mud cake degree index include:

[0124] A grading threshold for the mud cake degree index is set, dividing the mud cake degree index into three level intervals: low risk, medium risk, and high risk. The grading threshold for each interval is determined based on the index critical value corresponding to historical mud cake events in the standard simulation result library. The upper limit of the low risk interval is lower than the lower limit of the medium risk interval, and the upper limit of the medium risk interval is lower than the lower limit of the high risk interval.

[0125] The actual mud cake degree index is monitored in real time. When the actual mud cake degree index is at a low risk level, the tunneling parameters are kept unchanged, and the high-pressure water flushing device on the cutterhead surface is activated at the set time interval to remove the small amount of soil attached to the surface. At the same time, the tool wear and working environment parameters are continuously collected, and the trend evolution characteristic value in the three-level correlation feature is updated to ensure that the actual mud cake degree index is stably within the low risk range.

[0126] When the actual mud cake degree index enters the medium risk level, while maintaining the tunneling parameters, the injection ratio of the amendment is increased, and the cutterhead is controlled to rotate in the reverse direction for a short time to remove the soil stuck in the edge area using the reverse cutting force; at the same time, the parameter coordination characteristic value analysis is called to analyze the adaptability of the current tunneling parameters and the working environment, the frequency of cutterhead status monitoring is increased, and the changing trend of risk warning characteristic values ​​is captured in real time.

[0127] If the actual mud cake degree index reaches a high-risk level, the tunnel boring machine's advance will be suspended, the cutterhead will be controlled to rotate alternately in the forward and reverse directions, and high-pressure water flow will be used to flush the cutterhead and the inside of the soil chamber; the actual mud cake degree index will be continuously monitored until it drops to a low-risk level, at which point normal tunneling operations will resume; if the actual mud cake degree index does not decrease after a preset time, a deep investigation process will be triggered to check the wear condition of the cutters and the internal structure of the soil chamber.

[0128] The execution records of graded prevention operations, including trigger time, operation content, actual mud cake degree index change curve, and tunneling parameter adjustment records, will be stored in the standard simulation result library in real time. The graded threshold system and parameter matching rules of each level of operation will be optimized by combining the historical operation effect data. By combining graded operations with dynamically optimized tunneling parameters, the advanced prevention and timely intervention of cutterhead mud cake can be achieved.

[0129] Working principle and its effects:

[0130] This invention addresses the problems of delayed early warning of mud cake formation on the cutterhead of tunnel boring machines, lack of precise basis for adjusting tunneling parameters, and difficulty in adapting to dynamic working environments and cutter wear conditions. It achieves efficient tunneling control of the cutterhead through a collaborative mechanism of dynamic modeling, intelligent matching, and real-time adjustment.

[0131] A dynamic digital twin cutterhead model with an embedded tool wear calculation module was constructed. Combined with representative working condition points generated by stratified sampling, a multi-dimensional mapping standard simulation result library was built, encompassing working environment parameters, tunneling parameters, tool wear amount, and mud cake degree index. This process not only achieves precise synchronization between the virtual model and the actual cutterhead state by updating tool wear amount in real time, solving the problem of the static model being out of sync with working conditions, but also improves the basic accuracy of subsequent parameter matching through multi-parameter correlation storage, providing data support for efficient tunneling.

[0132] Based on the stability of the actual working environment, differentiated working condition matching is implemented. When stable, suitable working condition points are identified through similarity screening. When unstable, accurate matching is achieved by separating fluctuating and non-fluctuating environmental parameters. Combined with a three-level correlation feature construction and probability mapping relationship library, the probability of mud cake formation is predicted. This mechanism not only improves the timeliness and accuracy of prediction by using feature level combination and real-time correction coefficients to solve the problem of traditional early warning lag, but also ensures high adaptability between the initial target tunneling parameters and the current working conditions through a scenario-based matching strategy, providing a reliable benchmark for parameter adjustment.

[0133] Dynamic adjustment is triggered based on a mud cake probability threshold. Adjustment weights are determined by mapping slope and corrected using real-time feature values. The tunneling parameters are adjusted step-by-step to bring the mud cake severity index to a low-risk range. Simultaneously, actual data is fed back to update the standard simulation result library. This process ensures the accuracy and stability of parameter adjustment through weighted deviation calculation and hierarchical adjustment, addressing the problem of blind parameter adjustment. Furthermore, iterative optimization of the standard simulation result library continuously improves adaptability, ultimately achieving a significant reduction in cutterhead mud cake risk and a comprehensive improvement in tunneling efficiency and construction stability.

[0134] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for coupling analysis of cutterhead balling-up and tunneling parameters, characterized in that, The method comprises the following steps: constructing a basic geometric model of a digital twin cutterhead based on a three-dimensional model of the cutterhead, embedding a tool wear calculation module in the basic geometric model, the tool wear calculation module being used to update tool wear according to tool wear parameters to form a dynamic digital twin cutterhead model; constructing a three-dimensional parameter space of the cutterhead to generate representative working condition points of the cutterhead, generating a simulation result set of the representative working condition points according to the dynamic digital twin cutterhead model, and using the simulation result set to extract cutterhead mud cake parameters, quantify a mud cake degree index, and construct a standard simulation result library containing working environment parameters, tunneling parameters, tool wear, and the mud cake degree index; based on actual working environment parameters of the cutterhead and an updated tool wear interval, performing layer-by-layer retrieval based on a hierarchical index structure of the standard simulation result library, matching a representative working condition point set based on the stability of the working environment parameters, and generating preliminary target tunneling parameters based on the mud cake degree index and similarity; driving the cutterhead to operate according to the preliminary target tunneling parameters, quantifying the mud cake degree index in real time, and constructing three-level correlation features, which are used to predict a mud cake probability in a future prediction period according to the three-level correlation features and an association rule of a probability mapping relationship library, and dynamically adjusting the preliminary target tunneling parameters in response to the mud cake probability being greater than a mud cake probability threshold; extracting the mud cake degree index and the tunneling parameters corresponding to the representative working condition points from the representative working condition point set, configuring a mud cake degree threshold, screening out the representative working condition points with a mud cake degree index less than the mud cake degree threshold, and constructing an initial candidate parameter set; when it is determined that the current working environment is stable, determining a target mud cake degree index based on the mud cake degree index of the representative working condition points in the initial candidate parameter set, locating a target representative working condition point, and obtaining corresponding tunneling parameters of the target representative working condition point as the preliminary target tunneling parameters; when it is determined that the current working environment is unstable, constructing comprehensive working environment parameters based on a midpoint of a fluctuation interval of a current fluctuating environment parameter and current non-fluctuating environment parameters; calculating distance values of the working environment parameters of each representative working condition point in the initial candidate parameter set and the comprehensive working environment parameters, and setting fitting weights of the representative working condition points based on the distance values; performing weighted fitting operation on the tunneling parameters of each representative working condition point in the initial candidate parameter set according to the fitting weights to obtain comprehensive tunneling parameters as the preliminary target tunneling parameters; inputting the preliminary target tunneling parameters into the dynamic digital twin cutterhead model for simulation verification, and confirming the preliminary target tunneling parameters to be effective if the mud cake degree index output by the dynamic digital twin cutterhead model is less than a preset verification threshold; otherwise, determining that the verification fails, and triggering a fitting warning mechanism.

2. The method of claim 1, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of predicting the mud cake probability in the future prediction period comprise: configuring a prediction period, collecting current working environment parameters, tunneling parameters, tool wear, and quantifying a real-time mud cake degree index according to cutterhead mud cake parameters, and constructing three-level correlation features, the three-level correlation features including parameter coordination features, trend evolution features, and risk warning features; At the starting point of each prediction period, a mud cake probability prediction process is triggered, and a time range of a current prediction period is determined based on a preset prediction period length and a starting point of the current prediction period as a time reference; A probability mapping relationship library of three-level correlation features and mud cake probability is configured, and the probability mapping relationship library is constructed based on a corresponding relationship between a three-level correlation feature value combination and an actual mud cake probability; Parameter coordination feature values, trend evolution feature values and risk warning feature values are taken as input dimensions, each dimension is divided into feature levels according to a numerical interval, and a feature level combination corresponds to a preset mud cake probability reference value in the probability mapping relationship library; Three-level correlation feature values are extracted, feature levels to which the correlation feature values belong are determined, a feature level combination is formed, a corresponding mud cake probability reference value is matched according to the feature level combination, the mud cake probability reference value is adjusted through a real-time correction coefficient, and a mud cake probability of a prediction period is obtained, wherein the real-time correction coefficient is calculated according to a deviation between a current working environment parameter and a historical working environment parameter under the same feature level combination.

3. The method of claim 2, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The parameter coordination feature is used to reflect a matching degree of the working environment parameter, the tunneling parameter and the tool wear amount, and is obtained by calculating an adaptation coefficient of the working environment parameter and the tunneling parameter and combining a coupling index of the tool wear amount and the tunneling parameter; The trend evolution feature is used to capture a dynamic change law of an actual mud cake degree index, and is obtained by extracting a change slope, a fluctuation amplitude and a mean shift amount of the actual mud cake degree index in a sliding window; The risk warning feature is used to represent a growth trend of a mud cake risk under a current parameter combination, and is obtained by calculating a difference rate of the actual mud cake degree index and a preset safety threshold and combining a fluctuation rate of the working environment parameter.

4. The method of claim 1, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of dynamically adjusting the preliminary target tunneling parameter include: A correlation relationship data table of the working environment parameter, the tunneling parameter, the tool wear amount and the mud cake degree index in a standard simulation result library is called; According to a mapping slope data table of each tunneling parameter and the mud cake degree index under a current working environment parameter in the standard simulation result library, initial adjustment weights of the tunneling parameters are set in a descending order of values in the mapping slope data table; the initial adjustment weights are linearly corrected according to a change rate of the trend evolution feature value by real-time monitoring; A deviation value of a simulation mud cake degree index and an actual mud cake degree index is calculated, a weighted deviation value is obtained by multiplying the deviation value by an adjustment weight, a single parameter adjustment proportion is determined as a ratio of the weighted deviation value to a current value of the preliminary target tunneling parameter, and a target adjustment amount of a corresponding tunneling parameter is a product of the single parameter adjustment proportion and the current value of the corresponding tunneling parameter; The preliminary target tunneling parameter is step by step adjusted according to the target adjustment amount, and an interval time of adjacent two steps of adjustment is set according to the risk warning feature value; Real-time data of the actual mud cake degree index and the working environment parameter are continuously collected during the adjustment process, and an improvement amplitude is quantified according to a change trend of the actual mud cake degree index before and after the adjustment. An evaluation period is configured, if the improvement range in the evaluation period is less than a preset range threshold after the adjustment of the tunneling parameters, the target adjustment amount is recalculated, and an adjustment warning is given; otherwise, a clustering algorithm is used to update the multi-dimensional mapping relationship table in the standard simulation result library.

5. The method of claim 1, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of updating the tool wear parameter according to the tool wear parameter include: When the cutter head is in a working state, the tool wear parameter is collected to construct a tool wear parameter set; After data preprocessing of the tool wear parameter set, a tool wear amount model is constructed, the tool wear amount model reflects the cutting strength through the product of the formation hardness coefficient and the tool pressure, and reflects the cumulative effect of the cutting action through the product of the contact time and the cutting linear speed, to calculate the tool wear increment per unit time; The tool wear increment per unit time is integrated to obtain the cumulative tool wear amount; an actual wear detection value sequence collected by a tool wear monitoring device is obtained, and the cumulative tool wear amount calculated by the tool wear amount model corresponding to the actual wear detection value sequence is obtained according to the collection time of the actual wear detection value sequence; the coefficient weight in the tool wear amount model is dynamically adjusted through a deviation correction algorithm; The calibrated cumulative tool wear amount is fed back to the tool wear calculation module to update the tool wear amount in the dynamic digital twin cutter head model.

6. The method of claim 1, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of constructing the standard simulation result library include: Obtain the cutter head working environment parameter range, the tunneling parameter combination and the tool wear interval, construct a three-dimensional parameter space, and generate representative working condition points in the three-dimensional parameter space by using a hierarchical sampling method; For each representative working condition point, a dynamic digital twin cutter head model is used to repeatedly simulate to obtain a simulation result set; From the simulation result set, cutter head mud cake parameters are extracted to form an original cutter head mud cake sequence; The original cutter head mud cake sequence is subjected to time sequence alignment processing, and abnormal fluctuation data is removed by using a filtering algorithm to construct a standard cutter head mud cake sequence; The clustering centers of each cutter head mud cake parameter in the standard cutter head mud cake sequence are calculated as typical cutter head mud cake parameters to construct a typical cutter head mud cake sequence.

7. The method of claim 6, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of constructing the standard simulation result library further include: Set the mud cake threshold of each cutter head mud cake parameter, calculate the relative deviation of each typical cutter head mud cake parameter in the typical cutter head mud cake sequence and the corresponding mud cake threshold; Set the influence weight of each typical cutter head mud cake parameter, and calculate the mud cake degree index by weighted summation based on the relative deviation and the influence weight; According to the tool wear interval, the typical cutter head mud cake sequence is classified, and the corresponding working environment parameters, tunneling parameters and mud cake degree index are associated to construct a multi-dimensional mapping relationship table; The multi-dimensional mapping relationship table is stored in the standard simulation result library by configuring a hierarchical index structure.

8. The method of claim 1, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of matching the representative working condition point set include: Collect the actual working environment parameters and the actual tool wear parameters of the cutter head construction area; An evaluation period is configured, if the improvement range in the evaluation period is less than a preset range threshold after the adjustment of the tunneling parameters, the target adjustment amount is recalculated, and an adjustment warning is given; otherwise, a clustering algorithm is used to update the multi-dimensional mapping relationship table in the standard simulation result library. When it is determined that the current working environment is stable, updating the tool wear amount according to the current actual tool wear parameter, positioning the tool wear interval to which the tool wear amount belongs, performing one-level screening in the standard simulation result library according to the tool wear interval to obtain a simulation subset; In the simulation subset, calculating the similarity between the current actual working environment parameter and the working environment parameter of the representative working condition point in the simulation subset, screening the representative working condition point with a similarity greater than a preset similarity threshold, and constructing a representative working condition point set.

9. The method of claim 8, wherein the coupling analysis of the cutterhead clay cake and the excavation parameters is characterized by, The specific steps of matching the representative working condition point set further include: When it is determined that the current working environment is unstable, if the fluctuation amplitude of the working environment parameter sequence is greater than the corresponding fluctuation threshold, the working environment parameter is marked as a fluctuating environment parameter, and the fluctuation interval of the fluctuating environment parameter is obtained according to the working environment parameter sequence of the fluctuating environment parameter; otherwise, the working environment parameter is marked as a non-fluctuating environment parameter, and the current actual working environment parameter is taken as a fixed parameter; According to the current actual tool wear parameter, the tool wear amount is updated and the tool wear interval to which the tool wear amount belongs is positioned, and a simulation subset is screened from the standard simulation result library; In the simulation subset, calculating the similarity between the current non-fluctuating environment parameter and the non-fluctuating environment parameter of the representative working condition point in the simulation subset, screening the initial representative working condition point with a similarity greater than a preset similarity threshold, and constructing an initial representative working condition point set; In the initial representative working condition point set, the initial representative working condition point in which the fluctuating environment parameter is located in the fluctuation interval of the current fluctuating environment parameter is screened, and a representative working condition point set is constructed.

Images