A shield cutter wear prediction and adjustment method, system, medium and device

Abstract

The present application belongs to the technical field of shield construction, and provides a shield cutter wear prediction and adjustment method, system, medium and equipment. By establishing a coupling model of cutter position and tunneling motion parameters, the cumulative cutting stroke of each cutter is determined, and based on the stroke, the stratum information is mapped to a differentiated stratum working condition sequence corresponding to each cutter in space and time. A deep learning model is used to integrate the working condition sequence, construction parameters and historical data to predict the wear state. Further, by introducing a cutter group wear distribution dispersion index to evaluate the coordination state, when the triggering condition is met, a coordinated cutter replacement decision instruction is generated based on the constraint condition of controlling the relative height difference ratio of adjacent new and old cutters. Further, by constructing a comprehensive cost function, the comprehensive cost of immediate coordinated cutter replacement and delayed cutter replacement by reducing speed and wear is compared, and the optimal cutter replacement scheme is output under the premise of meeting the safety constraint, realizing the economic optimization of cutter maintenance decision.

Description

Technical Field

[0001] This invention belongs to the field of shield tunneling technology, specifically relating to a method, system, medium, and equipment for predicting and adjusting shield cutter wear. Background Technology

[0002] With the deepening development of urban underground space, the shield tunneling method has been widely used in subway and tunnel construction due to its safety and efficiency. As the core component for rock breaking, the wear condition of the cutterhead on the shield machine directly affects construction efficiency and cost.

[0003] In existing technologies, the following methods are mainly used to predict tool wear: (1) Empirical formula method based on tunneling mileage. Based on the cumulative tunneling mileage and the rock strength of the strata, the tool wear is estimated using a linear or power function relationship. For example, CN202010009506.4 discloses an empirical formula based on multiple wear mechanisms and tunneling mileage. By analyzing the microscopic wear mechanism of the tool in a specific stratum, an iterative relationship between wear and tunneling mileage is established.

[0004] (2) Theoretical calculation method based on mechanical model. By establishing a mechanical model of the interaction between the cutter and the rock, the normal force and tangential force borne by the cutter are calculated, and then the wear rate is estimated. For example, CN201910383490.0 discloses a real-time monitoring system for shield cutter wear, which uses sensors to collect parameters such as total thrust and torque, and obtains the wear prediction curve through mechanical regression analysis.

[0005] (3) Data-driven machine learning methods. In recent years, tool wear prediction based on methods such as neural networks and support vector machines has gradually emerged. For example, CN202511394537.5 discloses a method for matching similar wear curves in a historical database using the KNN algorithm.

[0006] However, existing technologies mainly suffer from the following technical problems that urgently need to be solved: First, there is the uniformity error in the description of working conditions. Existing methods typically use uniform mileage segments along the tunnel axis to match geological parameters, ignoring the significant differences in linear velocity, cutting path, and stress state between edge and center cutters. Taking a 10m diameter shield as an example, the linear velocity of the edge cutter (R=5m) is twice that of the cutter at a radius of 2.5m, and its actual cutting path (helix length) is also much larger than that of the center cutter. Using uniform spatial segmentation leads to a significantly larger error in the wear prediction of the edge cutter, seriously affecting the accuracy of cutter replacement decisions.

[0007] Second, there is a lack of a coordinated maintenance mechanism for the cutterhead group. Current cutter replacement decisions are mostly based on the wear threshold of individual cutters, neglecting the overall integrity of the cutterhead's cutting profile. Engineering practice shows that when only a few severely worn cutters are replaced, a significant height difference between the new cutter and the adjacent old cutter will occur. Due to the overall integrity of the cutterhead's cutting profile, the new cutter with the higher end face will bear more rock-breaking loads, generating a significant impact load concentration effect at the moment of cutting into the strata. This leads to an abnormally high early wear rate for the new cutter, the so-called "secondary accelerated wear" phenomenon. This phenomenon is particularly prominent in large-diameter shield tunneling (diameter > 10m) in hard rock or complex strata construction. The fundamental reason is that the edge cutters of large-diameter shield tunnels have high linear speeds and large single-cutting forces, making the load imbalance effect caused by the height difference between the end faces of new and old cutters more significant.

[0008] Third, the complexity of the formation is not adequately considered. Existing methods often use simple averaging or weighting to treat composite formations (soft on top and hard underneath) or formations with alternating soft and hard surfaces, which cannot accurately reflect the actual stress and wear state of the cutting tool in different formations.

[0009] To address the aforementioned issues, there is an urgent need for a wear prediction and adjustment method that can accurately describe differentiated cutting conditions, take into account the synergistic effect of tool groups, and fully consider the complexity of the formation. Summary of the Invention

[0010] To overcome the shortcomings of the prior art, the present invention aims to provide a method for predicting and adjusting the wear of tunnel boring machine cutters, which can accurately describe differentiated cutting conditions, take into account the synergistic effect of the cutter group, and fully consider the complexity of the formation, so as to solve the technical problems of edge cutter prediction distortion, easy breakage of new cutters after cutter replacement, and insufficient consideration of formation complexity in the prior art.

[0011] The technical solution adopted by this invention to solve its technical problem is: A method for predicting and adjusting wear of tunnel boring machine cutters includes the following steps: S1. Obtain shield tunneling construction parameters, ground parameters, and cutterhead structure parameters: The construction parameters include the feed rate, cutterhead rotation speed, total thrust, and cutterhead torque; The formation parameters include rock strength and abrasion indices; The cutter head structural parameters include the installation radius of each cutter and the initial end face height; S2. Based on the shield tunneling construction parameters and cutterhead structure parameters, establish a cutter kinematic model, synthesize the cutting paths of the rotational motion component and axial propulsion motion component of each cutter, and introduce a formation condition correction coefficient for correction, and calculate the cumulative cutting stroke of each cutter. S3. Using the tool with the smallest installation radius as a reference, based on the ratio of the cumulative cutting stroke of each tool to the cutting stroke of the reference tool, perform differentiated nonlinear resampling on the geological exploration profile data along the tunnel axis to generate a unique differentiated stratum working condition sequence for each tool that corresponds to its cutting path in time and space, so as to eliminate the spatiotemporal misalignment deviation of tools with different installation radii on the feature vector time axis. S4. Construct a multidimensional feature vector containing the differentiated formation working condition sequence, construction parameters and historical wear data, and use a time series prediction model that can handle long-term dependencies of sequence data to output the predicted wear state value of each tool. S5. Based on the wear state prediction value mentioned above, calculate the tool group wear distribution dispersion index within the same cutting trajectory band. When the preset triggering condition is met, generate a collaborative tool replacement decision command based on the normalized relative height difference ratio constraint between the new and old tools. The triggering condition includes the predicted wear amount of the target tool reaching the preset replacement threshold or the tool group wear distribution dispersion index exceeding the preset threshold. The normalized relative height difference ratio is the ratio of the maximum end face height difference between the new and old tools to the average remaining end face height of adjacent tools. S6. Perform cost optimization verification on the tool change decision, compare the combined costs of the two strategies of immediate coordinated tool change and tool change delayed due to wear and speed reduction, and select the tool change scheme to output under the premise of meeting safety constraints.

[0012] Preferably, in step S2, the method for calculating the cumulative cutting stroke is as follows: within the statistical period, the cumulative cutting stroke of each tool is obtained by vector synthesis and integration of the tool rotational motion component and the axial propulsion motion component at each sampling moment; the rotational motion component is determined by the tool installation radius and the cutterhead rotational angular velocity, and the axial propulsion motion component is determined by the cutterhead propulsion speed; the formation condition correction coefficient is determined according to the current tunneling formation type and is used to correct the deviation between the actual cutting stroke and the theoretical calculation value caused by factors such as formation slippage and vibration. The stratum condition correction coefficient is determined according to the stratum type of the excavation section in the following manner: For a single stratum, the value is taken within a preset range according to the hardness of the stratum, and the correction coefficient corresponding to hard rock layer is higher than that of soft soil layer; For a composite stratum, the weighted calculation is performed according to the thickness ratio of each stratum type in the excavation section, and the weighted coefficient is the proportion of each stratum thickness in the total height of the excavation section. The formation condition correction coefficient is adaptively iteratively updated during tunneling based on the ratio of measured wear to predicted wear. The update range is controlled by an adaptive adjustment coefficient. The adaptive adjustment is constrained so that the updated correction coefficient always remains within the preset reference range for its corresponding formation type. If the correction coefficient still exceeds the reference range after a preset number of iterations, an abnormal alarm is triggered and the adaptive update is paused. The value of the adaptive adjustment coefficient is determined according to the formation change frequency.

[0013] Preferably, in step S3, the method for generating the differentiated geological condition sequence is as follows: the geological exploration data is divided into geological slice units of unit length along the tunnel axis; the cutting stroke of the tool with the smallest installation radius is calculated as a reference reference value; the relative geological sampling density coefficient of each tool is determined according to the ratio of the cumulative cutting stroke of each tool to the reference reference value; the geological slice units are nonlinearly resampled according to the sampling density coefficient to generate a differentiated geological condition sequence for each tool; the geological condition sequences of each tool are independent of each other and do not share sampling nodes, and their sequence length is proportional to the cumulative cutting stroke of the corresponding tool.

[0014] Preferably, in step S4, the time series prediction model is a deep learning model capable of handling long-term dependencies in sequence data, including recurrent neural network models, convolutional neural network models, attention mechanism models, or combinations thereof; the feature vector also includes spatiotemporal evolution features extracted from the differentiated geological condition sequence, and dynamic response features extracted from construction parameters, wherein the dynamic response features include at least one of cutterhead torque, total thrust, and its fluctuation frequency.

[0015] Preferably, in step S5, the triggering condition includes at least one of the following: (1) The predicted wear of the target cutter reaches the preset ratio threshold of its maximum allowable wear. Different warning thresholds and forced replacement thresholds are set for the front cutter and the edge cutter, and the warning threshold for the edge cutter is lower than that for the front cutter; (2) The wear distribution dispersion index of the cutter group exceeds the preset threshold. The dispersion index is defined as the ratio of the standard deviation of the wear of all cutters in the same cutting trajectory zone to the average wear. It is only activated when the average wear of the cutter exceeds the minimum calculation threshold to eliminate interference during the break-in period after cutter replacement. The preset threshold of the dispersion index is dynamically determined according to the combination of the tunneling stratum type and the diameter of the shield machine, so that the threshold corresponding to the hard rock layer is higher than that of the soft soil layer, and the threshold corresponding to the large diameter shield is lower than that of the small diameter shield. The specific parameters of the dynamic determination rule are determined by the construction company after calibration based on the engineering data. The normalized relative height difference ratio is defined as the ratio of the maximum end face height difference between the new tool and the adjacent old tool to the average remaining end face height of the adjacent tools; the tool change decision logic is as follows: when the relative height difference ratio is lower than the lower threshold, a transition period speed reduction protection command is generated; when the relative height difference ratio is within the safe range, a command to replace the target tool individually is generated; when the relative height difference ratio exceeds the upper threshold, the tool change range is automatically expanded to the adjacent tools, and the process is iterated until the relative height difference ratio at all new and old tool junctions falls within the safe range. The transition period speed reduction protection command includes: after tool change, adopting a transition period strategy in which the feed speed and the tool head speed are reduced synchronously, continuously preset the number of loops to keep the penetration basically unchanged and reduce the initial impact load of the new tool; at the same time, investigate the cause of this abnormal wear and include the relevant data into the wear prediction model training set; The collaborative tool change decision instruction is expanded with the following priority: priority is given to replacing adjacent tools with heavier wear; if the relative height difference ratio still exceeds the standard after replacement on one side, it is expanded to adjacent tools on both sides at the same time; if the replacement range exceeds the preset number within the same trajectory zone, a cost optimization decision is triggered.

[0016] Preferably, in step S6, the cost optimization verification is achieved by constructing a comprehensive cost function, which consists of the sum of the fixed cost of a single downtime tool change and the cost of wasted remaining life of the prematurely replaced tool; the comprehensive costs of the immediate coordinated tool change strategy and the deceleration and wear-delayed tool change strategy are compared; when the life wasted cost caused by prematurely replacing the old tool exceeds a preset multiple of the fixed cost, the deceleration and wear-delay strategy is selected; otherwise, the immediate coordinated tool change strategy is selected; different cost tolerance coefficients are used for projects with high safety requirements and projects with economic priority. The cost tolerance coefficient is an adjustable parameter that is set by the construction unit according to the project's safety requirements and economic objectives. The deceleration and wear-delayed cutter replacement strategy includes: calculating the target wear amount based on the difference between the height difference between the end faces of the new and old cutters and the upper limit of the safe range; estimating the number of tunneling rings required to reach the target wear amount based on the current formation parameters and historical wear rates; generating a deceleration command to reduce the advance speed and cutterhead rotation speed to a preset ratio of normal values, and restoring normal tunneling parameters after the target wear amount is reached.

[0017] This invention also includes a shield tunneling cutter wear prediction and adjustment system, applied to the aforementioned shield tunneling cutter wear prediction and adjustment method, comprising: The parameter acquisition module is used to acquire shield tunneling construction parameters, ground parameters, and cutterhead structure parameters; The cumulative travel calculation module is used to calculate the cumulative cutting travel of each cutter by synthesizing the rotational motion component and axial propulsion motion component of each cutter and introducing the formation condition correction coefficient. Specifically, the shield tunneling trajectory is divided into multiple unit propulsion segments; the geological profile is nonlinearly resampled based on the ratio of the cumulative cutting travel of each cutter to the cutting travel of the reference cutter; and a unique formation condition sequence for each cutter is generated so that the formation sequence resolution of cutters with different installation radii corresponds proportionally to their cumulative cutting travel. The working condition mapping module is used to perform differential resampling of geological exploration profile data based on the ratio of the cumulative cutting stroke of each tool to the reference stroke, and generate a differential stratigraphic working condition sequence for each tool. The feature construction module is used to construct a feature vector containing the differentiated geological condition sequence, construction parameters, and historical wear data. The wear condition prediction module is used to output the predicted wear condition values ​​for each tool using a time series prediction model. The tool maintenance decision module is used to generate collaborative tool changing decision schemes based on the tool group wear distribution dispersion index and the normalized relative height difference ratio constraint. The cost optimization module compares the combined costs of two strategies: immediate coordinated tool change and delayed tool change due to wear and tear, and outputs the optimal solution. The sensor unit is used to collect shield tunneling construction parameters and cutterhead rotation speed data in real time; Stratigraphic database, used to store tunnel geological exploration data; Edge computing units are used to deploy the aforementioned method for predicting and adjusting shield cutter wear, or the method for mapping geological parameters to predict shield cutter wear. The decision display terminal is used to display suggested tool change plans to construction personnel and receive confirmation commands. The execution feedback unit is used to record the actual tool changing operation and the wear data after the tool changing, and feeds it back to the edge computing unit for model optimization.

[0018] Preferably, the formation parameter mapping method includes the following steps: obtaining the installation radius, rotational speed, and propulsion speed of each cutter on the shield cutterhead; synthesizing the rotational linear velocity component and axial propulsion speed component of each cutter and integrating them within a statistical period to obtain the actual cumulative cutting stroke of each cutter; using the cutting stroke of the cutter with the smallest installation radius as a reference value, calculating the formation sampling density coefficient of each cutter; based on the formation sampling density coefficient, performing nonlinear resampling on the geological exploration profile data obtained along the tunnel axis to independently generate differentiated formation working condition sequences for each cutter; the output of the method is a standardized formation working condition data sequence, which can be used as an independent data preprocessing module in combination with any wear prediction method, and can be used without relying on a specific prediction model architecture.

[0019] The present invention also includes a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the shield cutter wear prediction and adjustment method as described above, or implements the formation parameter mapping method as described above.

[0020] The present invention also includes a computer device comprising a processor and a memory, wherein the memory stores at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by the processor to implement the shield cutter wear prediction and adjustment method as described above, or to implement the formation parameter mapping method as described above.

[0021] Compared with the prior art, the beneficial effects of the present invention include: This invention establishes a coupled model between tool position and tunneling motion parameters to determine the cumulative cutting stroke of each tool, and maps formation information into a differentiated formation condition sequence corresponding to each tool in time and space based on this stroke. A deep learning model is used to fuse the condition sequence, construction parameters, and historical data to predict wear status. Furthermore, a tool group wear distribution dispersion index is introduced to evaluate the collaborative state. When triggering conditions are met, a collaborative tool replacement decision command is generated based on constraints controlling the ratio of the relative height difference between adjacent new and old tools. Further, a comprehensive cost function is constructed to compare the comprehensive costs of immediate collaborative tool replacement and decelerated wear-delayed tool replacement strategies. Under the premise of satisfying safety constraints, the optimal tool replacement scheme is output, achieving economic optimization of tool maintenance decisions. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of the shield tunneling cutter wear prediction method provided in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram illustrating the formation parameter mapping principle provided by the present invention. Figure 3 This is a schematic diagram showing the relationship between the normalized relative height difference ratio ρ and the new tool breakage rate and the overall tool replacement cost J. Figure 4 The structural block diagram of the shield cutter wear prediction and adjustment system provided by the present invention. Detailed Implementation

[0024] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. Many specific details are set forth in the following description to provide a thorough understanding of the present invention; the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0026] Example: like Figure 1-4 As shown, this embodiment provides a method for predicting and adjusting the wear of tunnel boring machine cutters, including the following steps: S1. Obtain shield tunneling construction parameters, ground parameters, and cutterhead structure parameters: Construction parameters include feed rate v, cutterhead rotation speed n, total thrust F, and cutterhead torque T; Formation parameters include uniaxial compressive strength (UCS), quartz content (Q), and rock abrasion index (CAI). The tool head structure parameters include the mounting radius R of each tool. i Tool type, initial end face height; S2. Based on the shield tunneling construction parameters and cutterhead structure parameters, establish a cutter kinematic model. By synthesizing the cutting paths of the rotational motion component and axial propulsion component of each cutter, and introducing a formation condition correction coefficient for correction, calculate the cumulative cutting stroke of each cutter. The cumulative cutting stroke is calculated as follows: within the statistical period, the cumulative cutting stroke of each tool is obtained by vector synthesis and integration of the tool rotational motion component and the axial propulsion motion component at each sampling moment; the rotational motion component is determined by the tool installation radius and the cutterhead rotational angular velocity, and the axial propulsion motion component is determined by the cutterhead propulsion speed; the formation condition correction coefficient is determined according to the current tunneling formation type and is used to correct the deviation between the actual cutting stroke and the theoretical calculation value caused by factors such as formation slippage and vibration. Specifically, this embodiment uses a vector synthesis method to calculate the cumulative cutting stroke L of the i-th tool within the statistical period T. i At each sampling time t, the instantaneous cutting stroke increment ΔL of the tool. i (t) is given by the following formula: ; Where: R i denoted as , where is the installation radius of the i-th tool (m); n(t) is the turret rotation speed at time t (rpm), and 60 is the conversion factor between minutes and seconds; v(t) is the feed speed at time t (m / s); Δt is the sampling time interval (s); and k is the formation condition correction factor (dimensionless).

[0027] When using the above formula for calculation, the units of all parameters must be consistent: n(t) is in rpm (revolutions per minute), which is converted to revolutions per second after dividing by 60; v(t) is in m / s. If the original data is in mm / min, the conversion relationship is: v(m / s) = v(mm / min) ÷ 60000.

[0028] Cumulative cutting stroke L i The sum of all increments within the statistical period T: ; Theoretical basis: The arc length of the tool's rotation within time Δt is 2πR. i ×(n(t)×Δt / 60), where v(t)Δt is the axial advance displacement. The vector combination of these two values ​​gives the actual cutting stroke increment, and then the formation condition correction factor k is introduced to correct the cutting stroke.

[0029] The stratum condition correction factor k is determined according to the stratum type of the excavation section in the following way: For a single stratum, the value is taken within a preset range according to the hardness of the stratum, and the correction factor corresponding to hard rock layer is higher than that for soft soil layer; For a composite stratum, the weighted calculation is performed according to the thickness ratio of each stratum type in the excavation section, and the weighting factor is the proportion of each stratum thickness in the total height of the excavation section. The formation condition correction coefficient is adaptively iterated and updated during the tunneling process based on the ratio of measured wear to predicted wear. The update range is controlled by the adaptive adjustment coefficient. The adaptive adjustment is constrained so that the updated correction coefficient always remains within the preset reference range of its corresponding formation type. If the correction coefficient still exceeds the reference range after a preset number of iterations, an abnormal alarm is triggered and the adaptive update is paused. The value of the adaptive adjustment coefficient is determined according to the formation change frequency.

[0030] Specifically, the actual cutting stroke of the tool in different formations deviates from the theoretical calculation value. This invention introduces a formation condition correction coefficient k to correct this deviation. k is determined through experimental calibration, historical data regression, or numerical simulation, and its determination principle is as follows: In soft soil layers, the cutting tool mainly experiences rolling friction, with smaller amplitudes of lateral slippage and normal vibration. The actual cutting stroke deviates little from the theoretical value, and the corresponding k value is close to and slightly greater than 1.0.

[0031] Hard rock layer: Under high load, the tool exhibits radial vibration and runout. The actual motion trajectory is a three-dimensional curve of helical superposition of normal vibration, which is longer than the theoretical two-dimensional helical cutting path. The corresponding k value is higher than that of soft soil layer. Composite strata: Weighted calculation based on the thickness proportion of each stratum type in the excavation section: ; Where, k jh is the baseline correction factor for the j-th stratum. j H represents the thickness of the stratum in the cross section (m), and H represents the total height of the excavation cross section (m).

[0032] The calibration method for k is as follows: In the initial stage of tunneling, the ratio of the measured cumulative wear to the theoretical predicted value is used as the correction benchmark for k. After continuously collecting data from at least 5 rings, the average value is taken as the calibration value of k corresponding to this stratum type. The construction company should confirm the k value by combining specific stratum parameters (such as water content, cohesion, UCS, etc.) through the above calibration method.

[0033] Adaptive adjustment of k: During subsequent tunneling, k is corrected based on the deviation between the measured wear amount and the predicted wear amount. ; Where λ is the adaptive adjustment coefficient, a larger λ value results in a faster correction response but reduces system stability. It is recommended to determine the value based on the frequency of formation changes: a smaller value for stable formations and a larger value for frequently changing formations. The adjusted k value must meet the baseline range for its corresponding formation type. After a preset number of iterations, k... new If the error still exceeds the range, an anomaly alarm is triggered and adaptive updates are paused. The recommended number of consecutive preset attempts is 3. For the specific range of k and the recommended value of λ, please refer to the following more specific embodiments.

[0034] S3. Using the tool with the smallest installation radius as a reference, based on the ratio of the cumulative cutting stroke of each tool to the cutting stroke of the reference tool, perform differentiated nonlinear resampling on the geological exploration profile data along the tunnel axis to generate a unique differentiated stratum working condition sequence for each tool that corresponds to its cutting path in time and space, so as to eliminate the spatiotemporal misalignment deviation of tools with different installation radii on the feature vector time axis. Specifically, the method for generating differentiated stratigraphic condition sequences is as follows: geological exploration data is divided into stratigraphic slice units of unit length along the tunnel axis; the cutting stroke of the tool with the smallest installation radius is calculated as a reference value; the relative stratigraphic sampling density coefficient of each tool is determined based on the ratio of the cumulative cutting stroke of each tool to the reference value; nonlinear resampling is performed on the stratigraphic slice units according to the sampling density coefficient to generate a differentiated stratigraphic condition sequence exclusive to each tool; the stratigraphic condition sequences of each tool are independent of each other and do not share sampling nodes, and their sequence length is proportional to the cumulative cutting stroke of the corresponding tool.

[0035] Based on the cumulative cutting stroke, the stratigraphic information in the geological exploration data is dynamically mapped into a unique stratigraphic working condition sequence corresponding to the spatiotemporal cutting path of each tool.

[0036] Calculate the cutting stroke L using the tool with the smallest installation radius as the reference. ref For the i-th tool, calculate the stroke ratio α.i : ; Based on the ratio α i Nonlinear resampling of stratigraphic geological profiles: for edge cutters (α) i >1), the original stratigraphic profile is resampled according to the stroke ratio to generate a stratigraphic working condition sequence that corresponds proportionally to the actual cutting stroke of the tool, achieving proportional alignment between the feature vector time axis and the tool cutting stroke; for the reference tool (α) i =1), with the original stratigraphic profile as the baseline sampling density. Each tool sequence is independent and does not share sampling nodes. The interpolation method used is linear interpolation or spline interpolation.

[0037] The mapping method described above eliminates the spatiotemporal misalignment error between the center and edge of the cutterhead, significantly improving prediction accuracy. The specific improvement depends on the formation conditions and model configuration, as detailed in the following more specific embodiments.

[0038] S4. Construct a multi-dimensional feature vector containing differentiated formation working condition sequences, construction parameters, and historical wear data. Use a time series prediction model that can handle long-term dependencies in sequence data to output the predicted wear status of each tool. Specifically, the time series prediction model is a deep learning model capable of handling long-term dependencies in sequence data, including recurrent neural network models, convolutional neural network models, attention mechanism models, or combinations thereof; the feature vector also includes spatiotemporal evolution features extracted from differentiated geological working condition sequences, and dynamic response features extracted from construction parameters, wherein the dynamic response features include at least one of cutterhead torque, total thrust, and its fluctuation frequency. A multidimensional feature vector containing differentiated formation working condition sequences, tunneling parameters, and historical wear data is constructed, input into a pre-trained depth time series prediction model, and outputs the current wear state prediction value of each tool.

[0039] Eigenvector composition: Formation working condition parameters (UCS) i (t), Q i (t), CAI i (t)), tunneling parameters (v(t), n(t), F(t), T(t)) and cumulative cutting stroke L of the tool i (t). This invention supports LSTM (Long Short-Term Memory Network), GRU (Gated Recurrent Unit), Transformer (Temporal Attention Model), TCN (Temporal Convolutional Network), or combinations of the above models.

[0040] S5. Based on the above-mentioned wear state prediction values, calculate the tool group wear distribution dispersion index within the same cutting trajectory band. When the preset triggering conditions are met, generate a collaborative tool replacement decision command based on the normalized relative height difference ratio constraint between the new and old tools. The triggering conditions include the predicted wear amount of the target tool reaching the preset replacement threshold or the tool group wear distribution dispersion index exceeding the preset threshold. The normalized relative height difference ratio is the ratio of the maximum end face height difference between the new and old tools to the average remaining end face height of adjacent tools. The triggering conditions include at least one of the following: (1) The predicted wear of the target tool reaches the preset ratio threshold of its maximum allowable wear. Different warning thresholds and forced replacement thresholds are set for the front hob and the edge hob, and the warning threshold of the edge hob is lower than that of the front hob; (2) The wear distribution dispersion index of the tool group exceeds the preset threshold. The dispersion index is defined as the ratio of the standard deviation of the wear of all tools in the same cutting trajectory zone to the average wear. It is only activated when the average wear of the tool exceeds the minimum calculation threshold to eliminate the interference of the break-in stage after tool replacement. The preset threshold of the dispersion index is dynamically determined according to the combination of the tunneling stratum type and the diameter of the shield machine, so that the threshold corresponding to the hard rock layer is higher than that of the soft soil layer, and the threshold corresponding to the large diameter shield is lower than that of the small diameter shield. The specific parameters of the dynamic determination rule are determined by the construction company after calibration based on the engineering data. The normalized relative height difference ratio is defined as the ratio of the maximum end face height difference between the new tool and the adjacent old tool to the average remaining end face height of the adjacent tools. The tool change decision logic is as follows: when the relative height difference ratio is lower than the lower threshold, a transition period deceleration protection instruction is generated; when the relative height difference ratio is within the safe range, an instruction to replace the target tool individually is generated; when the relative height difference ratio exceeds the upper threshold, the tool change range is automatically expanded to the adjacent tools, and the process is iterated until the relative height difference ratio at all new and old tool transition points falls within the safe range. The transition period speed reduction protection command includes: after tool change, adopting a transition period strategy of synchronously reducing the feed speed and the tool head speed, continuously presetting the number of loops to keep the penetration basically unchanged and reduce the initial impact load of the new tool; at the same time, investigating the cause of this abnormal wear and incorporating the relevant data into the wear prediction model training set; The collaborative tool change decision command expands with the following priority: prioritize replacing adjacent tools with heavier wear; if the relative height difference ratio still exceeds the standard after replacing one side, expand to adjacent tools on both sides simultaneously; if the replacement range exceeds the preset number within the same trajectory zone, trigger a cost optimization decision.

[0041] Based on the above, the tool group collaborative tool change decision in step S5 is as follows: (1) Wear distribution dispersion index η: To evaluate the uniformity of tool wear, a wear distribution dispersion index is defined: ; Where: σ is the standard deviation of wear of all tools within the same trajectory zone; μ is the average wear. When η < ηt, the tool wear distribution is considered uniform; when η ≥ ηt, the tool wear is considered uneven, triggering a coordinated tool change decision or adjusting the tunneling parameters. The dispersion index η is only activated when the average tool wear μ > the minimum calculation threshold μmin, to eliminate interference from the break-in period after tool change.

[0042] The dynamic determination rule for ηt: ηt is determined based on the combination of the tunneling strata type and the tunnel boring machine diameter. In principle, ηt is higher for hard rock layers than for soft soil layers, and ηt is lower for large-diameter tunnel boring machines than for small-diameter tunnel boring machines. Specific reference values ​​and calibration methods for ηt are detailed in the following more specific embodiments.

[0043] (2) Triggering conditions for tool change decision: A tool change decision is triggered when any of the following conditions are met: ① The predicted wear of the target tool reaches a preset replacement threshold. Different warning thresholds and forced replacement thresholds are set for front hobs and edge hobs, and the warning threshold for edge hobs is lower than that for front hobs; ② The tool group wear distribution dispersion index η exceeds a preset threshold ηt. Specific threshold settings are detailed in the following more specific embodiments.

[0044] (3) Normalized relative height difference ratio ρ and collaborative decision-making mechanism: Define the normalized relative height difference ratio ρ: ; Where: Δh is the maximum end-face height difference (mm) between the new tool and the adjacent old tool; H ref For reference height, it is defined as the average remaining end face height (mm) of adjacent tools: ; When the target tool is located on the innermost or outermost ring of the tool turret, H ref Take the remaining end face height of the actual existing single-sided adjacent tool.

[0045] The principle for setting the safe zone of ρ is as follows: Figure 3 As shown, Figure 3 This diagram illustrates the relationship between the normalized relative height difference ratio ρ, the new tool breakage rate, and the overall tool replacement cost J. Curve ① corresponds to the left vertical axis, and curve ② corresponds to the right vertical axis. The vertical solid lines indicate the positions of ρmin and ρmax and the safe range. The horizontal axis represents the normalized relative height difference ratio ρ, the left vertical axis represents the new tool breakage rate (%), which is the probability (%) of the new tool breaking during the initial stage of unbalanced impact load after tool replacement, and the right side of the vertical axis represents the overall cost J of a single tool replacement (illustrated, with the normal single tool replacement cost as a baseline of 1.0). Figure 3The chipping rate curve and the overall cost curve together define the safe operating range of ρ. The lower limit of the safe range, ρmin, is set to avoid transitional protection costs caused by excessively small differences in the end-face height between new and old tools; the upper limit of the safe range, ρmax, is set to control the chipping rate of new tools within an acceptable engineering range. ρmin and ρmax correspond to... Figure 3 The left and right inflection points of the lowest comprehensive cost range between the two curves. High-safety projects (large-diameter shield tunnels, high-risk geological formations) employ more stringent ρ... max For conventional engineering projects, the requirements can be appropriately relaxed.

[0046] (4) Collaborative tool change decision-making process: Step 1: Obtain the current remaining end face height H of the adjacent tools i-1 and i+1 of the target tool. i-1 and H i+1 (Only the height Hs on one side is obtained at the boundary location).

[0047] Step 2: Calculate the reference height H ref =(H i-1 +H i+1 ) / 2 (when one side is unilateral) Href =H s ).

[0048] Step 3: Calculate the end face height difference Δh = max(|H new -H i-1 |,|H new -H i+1 |)(When one side is unilateral, Δh=|H new -H s |).

[0049] Step 4: Calculate the relative height difference ratio ρ = Δh / H ref .

[0050] Step 5: Generate tool change command based on the value of ρ — if ρ < ρ min Generate a transition period deceleration protection command (the propulsion speed and rotational speed are reduced synchronously by a preset ratio, continuously for a preset number of cycles, while maintaining a constant penetration depth); if ρ min ≤ρ≤ρ max Generate a command to replace the target tool i individually; if ρ > ρ max The range of collaborative replacement is iteratively expanded according to priority (first the heavily worn side → both sides → extended trajectory zone) until the ρ value of all new and old tool interface surfaces falls into the safe range.

[0051] Note: The protective speed reduction strategy during the transition period differs in purpose from the active wear-down speed reduction strategy in step S6B below: the former aims to reduce the impact of the new tool and prevent breakage, with the feed speed and rotation speed decreasing proportionally to maintain a relatively constant penetration; the latter aims to actively wear down the new tool until it is flush with the old tool, allowing for a greater degree of asynchronous speed reduction. They are applicable to different scenarios, and the system automatically identifies and invokes them based on trigger conditions. See the following more specific implementation examples for the specific speed reduction ratio and number of cycles.

[0052] S6. Perform cost optimization verification on the tool change decision, compare the combined costs of the two strategies of immediate coordinated tool change and tool change delayed due to speed reduction and wear, and select the tool change scheme to output under the premise of meeting safety constraints.

[0053] Specifically, cost optimization verification is achieved by constructing a comprehensive cost function, which consists of the sum of the fixed cost of a single downtime tool change and the cost of wasted remaining life of the prematurely replaced tool. The comprehensive costs of the immediate coordinated tool change strategy and the deceleration and wear-delayed tool change strategy are compared. When the life wasted cost caused by prematurely replacing the old tool exceeds a preset multiple of the fixed cost, the deceleration and wear-delay strategy is selected; otherwise, the immediate coordinated tool change strategy is selected. Different cost tolerance coefficients are used for projects with high safety requirements and projects with economic priority. The cost tolerance coefficient is an adjustable parameter that is set by the construction unit according to the project's safety requirements and economic objectives. The strategy of reducing speed and delaying tool replacement due to wear includes: calculating the target wear amount based on the difference between the height difference between the end faces of the new and old tools and the upper limit of the safe range; estimating the number of tunneling rings required to reach the target wear amount based on the current formation parameters and historical wear rates; generating a speed reduction command to reduce the advance speed and cutterhead rotation speed to a preset ratio of normal values, and restoring normal tunneling parameters after the target wear amount is reached.

[0054] Based on the above, the cost optimization in step S6 is more specifically as follows: Before generating tool change adjustment instructions, the system performs cost verification. The comprehensive cost function J is constructed as follows: ; Where: N is the total number of tools recommended for replacement; C1 is the fixed cost of a single downtime tool replacement (including downtime loss, personnel entry costs, air pressure consumption, etc.); C w,j The remaining lifespan wasted cost of the j-th old tool that was replaced prematurely: ; Where P u H represents the unit price of a single cutting tool. rem,j H represents the remaining usable height of the j-th tool. total This represents the initial height of the cutting tool.

[0055] Compare the combined costs of the two strategies: Strategy A (Immediate Coordinated Tool Change) J A =C1+ΣC w,j ; Strategy B (Reduced tool wear and delayed tool change) JB=C d ×N rings +C1, where Cd is the time loss cost per tunneling ring, N rings The number of tunneling rings required to achieve the target wear level.

[0056] When ΣC w,j If the value is greater than β×C1, choose strategy B; otherwise, choose strategy A. The principle for setting the cost tolerance factor β: β is an adjustable parameter, set by the construction unit according to the project's safety requirements and economic objectives. Projects with high safety requirements should use a smaller value, while projects prioritizing economic efficiency should use a larger value. See the specific implementation method for detailed reference grading of β.

[0057] Strategy B's speed reduction wear scheme—Calculating the target wear amount: ; The required number of tunneling rings N is estimated based on current geological parameters and historical wear rates. rings The system generates a deceleration command to reduce the propulsion speed and cutterhead rotation speed to a preset ratio of normal values; after reaching the target wear level, it restores normal tunneling parameters. See the following more specific implementation example for the specific deceleration ratio.

[0058] The present invention also includes a shield tunneling cutter wear prediction and adjustment system, applied to the above-mentioned shield tunneling cutter wear prediction and adjustment method, comprising: The parameter acquisition module is used to acquire shield tunneling construction parameters, ground parameters, and cutterhead structure parameters; The cumulative travel calculation module is used to calculate the cumulative cutting travel of each cutter by synthesizing the rotational motion component and axial propulsion motion component of each cutter and introducing the formation condition correction coefficient. Specifically, the shield tunneling trajectory is divided into multiple unit propulsion segments; the geological profile is nonlinearly resampled based on the ratio of the cumulative cutting travel of each cutter to the cutting travel of the reference cutter; and a unique formation condition sequence for each cutter is generated so that the formation sequence resolution of cutters with different installation radii corresponds proportionally to their cumulative cutting travel. The working condition mapping module is used to perform differential resampling of geological exploration profile data based on the ratio of the cumulative cutting stroke of each tool to the reference stroke, and generate a differential stratigraphic working condition sequence for each tool. The feature construction module is used to construct feature vectors containing differential formation condition sequences, construction parameters, and historical wear data; The wear condition prediction module is used to output the predicted wear condition values ​​for each tool using a time series prediction model. The tool maintenance decision module is used to generate collaborative tool changing decision schemes based on the tool group wear distribution dispersion index and the normalized relative height difference ratio constraint. The cost optimization module compares the combined costs of two strategies: immediate coordinated tool change and delayed tool change due to wear and tear, and outputs the optimal solution. The sensor unit is used to collect shield tunneling construction parameters and cutterhead rotation speed data in real time; Stratigraphic database, used to store tunnel geological exploration data; Edge computing units are used to deploy the aforementioned methods for predicting and adjusting shield cutter wear, or methods for mapping geological parameters to predict shield cutter wear. The decision display terminal is used to display suggested tool change plans to construction personnel and receive confirmation commands. The execution feedback unit is used to record the actual tool changing operation and the wear data after the tool changing, and feeds it back to the edge computing unit for model optimization.

[0059] Preferably, the formation parameter mapping method includes the following steps: obtaining the installation radius, rotational speed, and propulsion speed of each cutter on the shield cutterhead; synthesizing the rotational linear velocity component and axial propulsion speed component of each cutter and integrating them within a statistical period to obtain the actual cumulative cutting stroke of each cutter; using the cutting stroke of the cutter with the smallest installation radius as a reference value, calculating the formation sampling density coefficient of each cutter; based on the formation sampling density coefficient, performing nonlinear resampling on the geological exploration profile data obtained along the tunnel axis to independently generate differentiated formation working condition sequences for each cutter; the output of the method is a standardized formation working condition data sequence, which can be used as an independent data preprocessing module in combination with any wear prediction method, and can be used without relying on a specific prediction model architecture.

[0060] More specific examples: Example 1: This embodiment provides reference values ​​for the formation condition correction coefficient k in step S2, the threshold parameters in step S5, and the cost tolerance coefficient β in step S6, for construction companies to use for calibration. (1) Reference range of k values: For soft soil layers, the recommended value for k is 1.0–1.05. In soft soil layers, the contact between the cutting tool and the soil is mainly characterized by rolling friction, with relatively small lateral slippage and normal vibration amplitudes. The actual cutting stroke deviates little from the theoretical value, and k is close to and slightly greater than 1.0. It should be noted that in silty soil layers with extremely high water content, tool slippage may occur. In this case, the actual cutting stroke may be lower than the theoretical value, and the k value may be close to or even lower than 1.0. If the calibration test results show k < 1.0, the construction company should analyze the cause of slippage and take corresponding measures, and simultaneously include the measured k value corresponding to this working condition in the calibration database. The above range is derived based on the kinematic characteristics of cutting tools in soft soil layers. The construction company should confirm this through calibration tests based on actual soil parameters (such as water content and cohesion).

[0061] For hard rock layers, the recommended value for k is 1.05–1.15. In hard rock formations, the cutting tool experiences radial vibration and blade runout under high normal forces. The actual trajectory is a three-dimensional curve of a helical line superimposed with normal vibration, longer than the theoretical two-dimensional helical cutting path. Based on the rock-breaking mechanics analysis of rigid roller cutters and data from domestic shield tunneling projects, when the uniaxial compressive strength (UCS) of the rock is 60–100 MPa (corresponding to medium-hard rock, referring to commonly used rock strength classifications in domestic engineering), the reference value for k is approximately 1.03–1.08; when the UCS is 100–200 MPa (corresponding to hard to extremely hard rock), the reference value for k is approximately 1.08–1.15. These ranges are derived from the characteristics of the cutting tool's motion mechanics and combined with comprehensive judgment based on engineering experience. Construction companies should verify the final value through calibration experiments of no less than 5 rings in specific projects.

[0062] The composite stratum k is calculated by weighting the thickness of each stratum. Example: A cross section is 10m high, with the upper 5m being soft soil (k1=1.02) and the lower 5m being hard rock (k2=1.10), then k=1.02×5 / 10+1.10×5 / 10=1.06.

[0063] The adaptive adjustment coefficient λ for k is recommended to range from 0.1 to 0.3. A larger λ value results in a faster response of the correction coefficient to measured deviations, but also reduces system stability; a smaller λ value results in a slower system response but stronger noise immunity. When the formation is stable, a smaller λ value (reference 0.1 to 0.15) is recommended to avoid overcorrection caused by sensor noise; when the formation changes frequently, a larger λ value (reference 0.2 to 0.3) is recommended to improve the response speed to sudden changes in formation. The initial recommended value for λ can be set to 0.2. Construction companies should dynamically calibrate λ based on the convergence of the predicted deviation during the initial stage of tunneling (no less than 10 rings) to determine the final value suitable for the geological characteristics of this project.

[0064] (2) Reference value for tool change trigger threshold: Front hobbing cutter: The warning threshold is taken as the maximum allowable wear amount W in the design.max 78%, forced replacement of threshold reference W max 90%.

[0065] Edge rolling cutter: Warning threshold reference value W max 68%, forced replacement of threshold reference W max 90%.

[0066] η Minimum computation enable threshold μ min Refer to the maximum allowable wear W of the tool with this trajectory design. max 10%.

[0067] η t Reference values: 0.30–0.40 for soft soil layers and 0.40–0.55 for hard rock layers. The uneven stress distribution on the cutting tool is more pronounced in hard rock formations, and the background value of wear distribution dispersion is inherently higher. If the same η value as that used for soft soil layers is adopted… t This will lead to frequent false triggering of tool change decisions, therefore the η corresponding to the hard rock layer t The setting is above the soft soil layer. Large-diameter shield tunneling machines have a large number of cutters and a large statistical sample size for each trajectory zone, resulting in more stable statistical results for wear uniformity and greater sensitivity to discrepancies in dispersion. Therefore, the η corresponding to large-diameter shield tunneling machines... t The setting is lower than that for small-diameter shield tunnels. For shield diameters of 10–12m, the above baseline value can be multiplied by a correction factor of 0.85–0.90; for shield diameters >12m, it can be multiplied by a correction factor of 0.80–0.85. The corrected η t The recommended value is no less than 0.30 and no more than 0.55. These reference values ​​are based on comprehensive judgment using engineering experience and should be verified through on-site data before formal application.

[0068] (3) Reference value for the safe interval of ρ: High-security range reference value: ρ min =0.12, ρ max =0.40, suitable for high-risk strata, large-diameter shield tunnels (diameter > 10m) or projects with extremely high safety requirements.

[0069] Economic range reference value: ρ min =0.15, ρ max =0.50, suitable for conventional strata, small and medium-sized shield tunnels (diameter 6-10m) or projects with high economic requirements.

[0070] The above reference values ​​are set based on the following: ρ max The upper limit is set to control the initial impact load increment of new cutting tools within a reasonable range of the tool's design rated load. High-safety engineering employs even stricter ρ... max To further reduce the risk of collapse; ρ minThe lower limit is set to balance the wasted cost of the remaining life of the old tool caused by the expansion of the tool change coordination range. When ρ is below the lower limit, it indicates that the height difference between the new and old tools is minimal, and there is no need to coordinate the replacement of adjacent tools, but transitional speed reduction protection is still required. The specific values ​​mentioned above are calculated based on tool design load analysis and engineering practice experience. Construction companies should determine the applicable ρ by combining the specific tool model and geological conditions, and calibrating it through on-site statistical data (it is recommended to collect corresponding data from no less than 3 tool change events). min and ρ max .

[0071] Transitional speed reduction protection (ρ < ρ min (At the same time): The feed speed and rotation speed are reduced by 15% to 25% simultaneously for 50 to 100 cycles, while maintaining the penetration depth, in order to reduce the initial impact load on the new tool.

[0072] (4) Reference range for cost tolerance coefficient β: For projects with high safety requirements, the reference value for β is 1.3 to 1.5; for conventional engineering projects, the reference value for β is 1.5 to 1.7; and for projects prioritizing economic efficiency, the reference value for β is 1.7 to 1.8. These grading recommendations are based on practical engineering experience, and construction units can adjust them according to actual circumstances.

[0073] Strategy B Reduction Ratio Reference Value: Propulsion speed reduced to v new = (0.6~0.8)×v normal The cutter head speed is reduced to n new = (0.7~0.9)×n normal Once the target wear level is reached, normal tunneling parameters are restored.

[0074] Example 2: This embodiment demonstrates how the formation parameter mapping method of the present invention can be used as an independent module.

[0075] Application scenario: A tunnel boring machine (TBM) construction company already has a wear prediction system based on empirical formulas, but the prediction accuracy of edge cutters is not ideal. They hope to improve the accuracy of ground parameters without modifying the existing system.

[0076] 1. Obtain parameters: cutter head diameter 10.5m; number of cutters 64; installation radius R1=0.5m, R2=1.0m, ..., R 64 =5.25m; tunneling parameters: rotation speed n=1.5rpm, advance speed v=50mm / min (corresponding to medium-speed tunneling in composite strata). This embodiment uses a simplified arrangement with equal spacing as an illustration. In actual engineering, the arrangement of cutters is determined by the cutterhead design scheme.

[0077] 2. Calculate the cumulative cutting stroke: Using one ring of tunnel segments as the statistical unit (a typical single-ring advance distance is 1500mm), i.e., L0 = 1500mm = 1.5m; advance time = 1800s; total cutterhead revolutions N = 45 revolutions; advance speed conversion v ≈ 8.33 × 10⁻⁶ -4 m / s. With edge tool R 64 Taking R=5.25m as an example (taking k=1): rotational component Lrotation = 2π × 5.25 × 45 ≈ 1485m; propulsion component Lpropulsion = 1.5m; ; The ratio of the rotational component (1485m) to the feed component (1.5m) is approximately 990:1, indicating that the rotational component is absolutely dominant, while the feed component is relatively negligible. Center tool R1 (R=0.5m): Total cutting stroke L1≈141.4m.

[0078] Note: To clearly demonstrate the geometric meaning of the stroke ratio, this step temporarily assumes k=1 for calculation (in actual engineering, the value of k for composite strata is shown in item (1) of Example 1, with a typical value of approximately 1.06). When k≠1 is introduced, the cumulative cutting stroke L of each tool is... i Scaled up proportionally, but the stroke ratio α i =L i / L ref It remains unchanged and does not affect the calculation results of the formation sampling density coefficient.

[0079] 3. Calculate the formation sampling density coefficient: using the central tool R1 as the reference (L ref =141.4m); α 64 =1485 / 141.4≈10.5; α1=141.4 / 141.4=1.0.

[0080] The above stroke ratios indicate that, if the traditional equidistant segmentation method is used, the edge tool R... 64 The corresponding formation feature vector spatiotemporal alignment deviation is approximately 10 times that of the central tool R1, i.e., R in the traditional method. 64 The generated stratigraphic sequence exhibits a misalignment on the time axis approximately 10 times that of the reference tool. This invention fundamentally eliminates this misalignment by performing nonlinear resampling of the stratigraphic profile according to the α value. This effect can be directly derived from the stroke ratio geometry and is independent of the selection of the downstream prediction model.

[0081] 4. Resampling of the geological profile: Several sampling points were obtained from the geological exploration profile data along the tunnel within the calculation section of this embodiment, at intervals of 1.5m. Edge cutter R 64 (α) 64(≈10.5) Based on this, 10,500 sequence nodes are generated through resampling. This operation transforms the travel resolution of the formation sequence from "based on the advance mileage" to "based on the actual cutting travel of each tool," eliminating spatiotemporal misalignment between tools of different radii. The center tool R1 (α1=1.0) maintains 1,000 sampling points. Resampling uses linear interpolation or spline interpolation.

[0082] 5. Output and Verification: The formation sequences corresponding to each tool are input into the existing empirical formula prediction system. Based on the parameters of the above embodiment (simulation using MATLAB platform, formation parameters constructed with reference to the measured statistical interval of UCS in typical domestic granite formations, UCS range 80–180 MPa, quartz content 25%–40%, rock abrasion index CAI value 2.5–4.0), numerical simulation is used to compare and verify the edge tool wear prediction error before and after the improvement. The results show that after introducing differentiated formation working condition sequence mapping, the edge tool wear prediction error is significantly reduced compared with the traditional equidistant segmentation method. The above improvement effect stems from the fact that the present invention eliminates the spatiotemporal alignment deviation between tools with different installation radii in the mapping stage. It can be achieved by simply introducing this mapping module without modifying the downstream prediction model. In actual engineering, the improvement in prediction accuracy is affected by factors such as formation homogeneity, sensor accuracy, and model training data volume, and may differ from the above simulation results.

[0083] In this embodiment, the differentiated formation mapping module is decoupled from the existing prediction system as an independent component, demonstrating the independent feasibility of the formation parameter mapping method. The output of this module can be interfaced with any downstream prediction method, such as empirical formula method, mechanical model method or deep learning method; the complete implementation of the present invention (Embodiment 1) combines it with a deep learning model, which can further improve the dynamic prediction capability under the condition of frequent formation changes.

[0084] Example 3: Device Example: like Figure 4 As shown, this embodiment provides a shield tunneling cutter wear prediction and adjustment system, which includes: 1. Parameter acquisition module: used to connect to the shield machine PLC system and database to collect data such as propulsion speed v, rotational speed n, thrust F, and torque T in real time; 2. Working Condition Mapping and Feature Construction Module: This module executes steps S2 and S3 of the above method to calculate the cumulative cutting stroke L of the tool. i Nonlinear resampling of the stratigraphic sequence is also performed; 3. Wear condition prediction module: It is internally equipped with a pre-trained time series prediction model (such as LSTM) to output the wear prediction value of each tool; 4. Tool maintenance decision module: It integrates a "height difference analysis unit" and a "cost optimization unit" to execute the logic of steps S5 and S6, that is, output the final tool change or parameter adjustment command based on the relative height difference ratio and cost function J.

[0085] Example 4: Computer Equipment and Media This embodiment provides a computer device, including a processor and a memory. The memory stores at least one instruction, at least one program, code set, or instruction set. The processor loads and executes the at least one instruction, at least one program, code set, or instruction set to implement the shield tunneling cutter wear prediction and adjustment method described above, or to implement the formation parameter mapping method described above. A computer-readable storage medium is also provided, storing a computer program thereon. When the computer program is executed, it implements the shield tunneling cutter wear prediction and adjustment method described above, or to implement the formation parameter mapping method described above.

[0086] In summary, the advantages of this invention are as follows: 1. Improved Edge Tool Prediction Accuracy: By mapping differentiated formation working condition sequences, the "spatiotemporal misalignment" between the cutterhead center and edge is eliminated, significantly reducing the edge tool wear prediction error. Taking the working condition described in Example 2 as an example, the cumulative cutting stroke ratio of the edge tool (R=5.25m) to the center tool (R=0.5m) is approximately 10.5:1. Under this condition, the spatiotemporal alignment deviation of the formation feature vector generated by the traditional equidistant segmentation method for the edge tool is approximately 10 times that for the center tool. This invention fundamentally eliminates the above alignment deviation from the mapping stage by performing nonlinear resampling based on the stroke ratio. The above deviation elimination effect can be directly derived from the geometric relationship of the stroke ratio without relying on simulation assumptions. The overall improvement in wear prediction accuracy varies depending on factors such as downstream prediction model configuration, formation homogeneity, and sensor accuracy. The actual engineering application effect needs to be verified with field data.

[0087] 2. Eliminates the "abrupt effect" of new cutterheads and significantly reduces the risk of breakage: A collaborative cutterhead replacement decision mechanism based on the normalized relative height difference ratio ρ is creatively proposed, which effectively solves the problem of early breakage of new cutterheads that is common in large-diameter shield tunnels.

[0088] 3. Achieved maximum economic benefits: Introduced the cost optimization model J(N)=C1+ΣC w,j The system finds a balance between "preventing breakage" and "reducing waste". It intelligently compares the two strategies of "immediate coordinated tool change" and "reducing wear" and selects the optimal solution to execute.

[0089] 4. A technological leap from "single-tool decision-making" to "tool group collaboration" has been achieved: By using the wear distribution dispersion index η and the normalized relative height difference ratio ρ, the tool replacement decision-making has been elevated from the single-tool level to the tool group level, thus achieving collaborative optimization.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for predicting and adjusting wear of tunnel boring machine cutters, characterized in that, Includes the following steps: S1. Obtain shield tunneling construction parameters, ground parameters, and cutterhead structure parameters: The construction parameters include the feed rate, cutterhead rotation speed, total thrust, and cutterhead torque; The formation parameters include rock strength and abrasion indices; The cutter head structural parameters include the installation radius of each cutter and the initial end face height; S2. Based on the shield tunneling construction parameters and cutterhead structure parameters, establish a cutter kinematic model, synthesize the cutting paths of the rotational motion component and axial propulsion motion component of each cutter, and introduce a formation condition correction coefficient for correction, and calculate the cumulative cutting stroke of each cutter. S3. Using the tool with the smallest installation radius as a reference, based on the ratio of the cumulative cutting stroke of each tool to the cutting stroke of the reference tool, perform differentiated nonlinear resampling on the geological exploration profile data along the tunnel axis to generate a unique differentiated stratum working condition sequence for each tool that corresponds to its cutting path in time and space, so as to eliminate the spatiotemporal misalignment deviation of tools with different installation radii on the feature vector time axis. The method for generating the differentiated geological condition sequence is as follows: geological exploration data is divided into geological slice units of unit length along the tunnel axis; the cutting stroke of the tool with the smallest installation radius is calculated as a reference value, using the tool's cutting stroke as the reference value; the relative geological sampling density coefficient of each tool is determined based on the ratio of the cumulative cutting stroke of each tool to the reference value; the geological slice units are nonlinearly resampled according to the sampling density coefficient to generate a differentiated geological condition sequence specific to each tool; the geological condition sequences of each tool are independent of each other and do not share sampling nodes, and their sequence length is proportional to the cumulative cutting stroke of the corresponding tool; S4. Construct a multidimensional feature vector containing the differentiated formation working condition sequence, construction parameters and historical wear data, and use a time series prediction model that can handle long-term dependencies of sequence data to output the predicted wear state value of each tool. S5. Based on the wear state prediction value mentioned above, calculate the tool group wear distribution dispersion index within the same cutting trajectory band. When the preset triggering condition is met, generate a collaborative tool replacement decision command based on the normalized relative height difference ratio constraint between the new and old tools. The triggering condition includes the predicted wear amount of the target tool reaching the preset replacement threshold or the tool group wear distribution dispersion index exceeding the preset threshold. The normalized relative height difference ratio is the ratio of the maximum end face height difference between the new and old tools to the average remaining end face height of adjacent tools. S6. Perform cost optimization verification on the tool change decision, compare the combined costs of the two strategies of immediate coordinated tool change and tool change delayed due to wear and speed reduction, and select the tool change scheme to output under the premise of meeting safety constraints.

2. The method for predicting and adjusting shield tunneling cutter wear according to claim 1, characterized in that, In step S2, the cumulative cutting stroke is calculated as follows: within the statistical period, the cumulative cutting stroke of each tool is obtained by vector synthesis and integration of the tool rotational motion component and the axial propulsion motion component at each sampling moment; the rotational motion component is determined by the tool installation radius and the cutterhead rotational angular velocity, and the axial propulsion motion component is determined by the cutterhead propulsion speed; the formation condition correction coefficient is determined according to the current tunneling formation type and is used to correct the deviation between the actual cutting stroke and the theoretical calculation value caused by formation slippage and vibration factors. The stratum condition correction coefficient is determined according to the stratum type of the excavation section in the following manner: For a single stratum, the value is taken within a preset range according to the hardness of the stratum, with the correction coefficient corresponding to hard rock layers being higher than that for soft soil layers; For a composite stratum, a weighted calculation is performed based on the proportion of each stratum type in the thickness of the excavation section, and the weighted coefficient of the weighted calculation is the proportion of each stratum thickness in the total height of the excavation section. The formation condition correction coefficient is adaptively and iteratively updated during the tunneling process based on the ratio of measured wear to predicted wear, and the update range is controlled by the adaptive adjustment coefficient.

3. The method for predicting and adjusting shield tunneling cutter wear according to claim 1, characterized in that, In step S4, the time series prediction model is a deep learning model capable of handling long-term dependencies in sequence data, including recurrent neural network models, convolutional neural network models, attention mechanism models, or combinations thereof; the feature vector also includes spatiotemporal evolution features extracted from the differentiated geological working condition sequence, and dynamic response features extracted from construction parameters, wherein the dynamic response features include at least one of cutterhead torque, total thrust, and its fluctuation frequency.

4. The method for predicting and adjusting shield tunneling cutter wear according to claim 1, characterized in that, In step S5, the triggering conditions include at least one of the following: the predicted wear of the target cutter reaches a preset percentage threshold of its maximum allowable wear; different warning thresholds and forced replacement thresholds are set for the front hob and the edge hob, and the warning threshold for the edge hob is lower than that for the front hob; the wear distribution dispersion index of the cutter group exceeds a preset threshold. The dispersion index is defined as the ratio of the standard deviation of the wear of all cutters within the same cutting trajectory zone to the average wear, and it is only activated when the average wear of the cutter exceeds the minimum calculation threshold to eliminate interference during the break-in period after cutter replacement; the preset threshold of the dispersion index is dynamically determined based on the combination of the tunneling strata type and the diameter of the tunnel boring machine, so that the threshold corresponding to hard rock layers is higher than that for soft soil layers, and the threshold corresponding to large-diameter shields is lower than that for small-diameter shields; the specific parameters of the dynamic determination rule are determined by the construction company after calibration based on engineering data. The normalized relative height difference ratio is defined as the ratio of the maximum end face height difference between the new tool and the adjacent old tool to the average remaining end face height of the adjacent tools; the tool change decision logic is as follows: when the relative height difference ratio is lower than the lower threshold, a transition period speed reduction protection command is generated; when the relative height difference ratio is within the safe range, a command to replace the target tool individually is generated; when the relative height difference ratio exceeds the upper threshold, the tool change range is automatically expanded to the adjacent tools, and the process is iterated until the relative height difference ratio at all new and old tool junctions falls within the safe range. The transition period speed reduction protection command includes: after tool change, adopting a transition period strategy in which the feed speed and the tool head speed are reduced synchronously, continuously preset the number of loops to keep the penetration basically unchanged and reduce the initial impact load of the new tool; at the same time, investigate the cause of this abnormal wear and include the relevant data into the wear prediction model training set; The collaborative tool change decision instruction is expanded with the following priority: priority is given to replacing adjacent tools with heavier wear; if the relative height difference ratio still exceeds the standard after replacement on one side, it is expanded to adjacent tools on both sides at the same time; if the replacement range exceeds the preset number within the same trajectory zone, a cost optimization decision is triggered.

5. The method for predicting and adjusting shield tunneling cutter wear according to claim 3, characterized in that, In step S6, the cost optimization verification is achieved by constructing a comprehensive cost function, which consists of the sum of the fixed cost of a single downtime tool change and the cost of wasted remaining life of the tool that was replaced prematurely. The comprehensive costs of the immediate coordinated tool change strategy and the deceleration and wear-delayed tool change strategy are compared. When the life wasted cost caused by prematurely replacing the old tool exceeds a preset multiple of the fixed cost, the deceleration and wear-delay strategy is selected; otherwise, the immediate coordinated tool change strategy is selected. Projects with high safety requirements and projects prioritizing economic efficiency adopt different cost tolerance coefficients. These cost tolerance coefficients are adjustable parameters that can be set by the construction unit based on the project's safety requirements and economic objectives. The deceleration and wear-delayed cutter replacement strategy includes: calculating the target wear amount based on the difference between the height difference between the end faces of the new and old cutters and the upper limit of the safe range; estimating the number of tunneling rings required to reach the target wear amount based on the current formation parameters and historical wear rates; generating a deceleration command to reduce the advance speed and cutterhead rotation speed to a preset ratio of normal values, and restoring normal tunneling parameters after the target wear amount is reached.

6. A shield tunneling cutter wear prediction and adjustment system, characterized in that, The method for predicting and adjusting shield cutter wear according to any one of claims 1-5 includes: The parameter acquisition module is used to acquire shield tunneling construction parameters, ground parameters, and cutterhead structure parameters; The cumulative travel calculation module is used to calculate the cumulative cutting travel of each cutter by synthesizing the rotational motion component and axial propulsion motion component of each cutter and introducing the formation condition correction coefficient. Specifically, the shield tunneling trajectory is divided into multiple unit propulsion segments; the geological profile is nonlinearly resampled based on the ratio of the cumulative cutting travel of each cutter to the cutting travel of the reference cutter; and a unique formation condition sequence for each cutter is generated so that the formation sequence resolution of cutters with different installation radii corresponds proportionally to their cumulative cutting travel. The working condition mapping module is used to perform differential resampling of geological exploration profile data based on the ratio of the cumulative cutting stroke of each tool to the reference stroke, and generate a differential stratigraphic working condition sequence for each tool. The feature construction module is used to construct a feature vector containing the differentiated geological condition sequence, construction parameters, and historical wear data. The wear condition prediction module is used to output the predicted wear condition values ​​for each tool using a time series prediction model. The tool maintenance decision module is used to generate collaborative tool changing decision schemes based on the tool group wear distribution dispersion index and the normalized relative height difference ratio constraint. The cost optimization module compares the combined costs of two strategies: immediate coordinated tool change and delayed tool change due to wear and tear, and outputs the optimal solution. The sensor unit is used to collect shield tunneling construction parameters and cutterhead rotation speed data in real time; Stratigraphic database, used to store tunnel geological exploration data; Edge computing units are used to deploy a ground parameter mapping method for predicting shield cutter wear; The decision display terminal is used to display suggested tool change plans to construction personnel and receive confirmation commands. The execution feedback unit is used to record the actual tool changing operation and the wear data after the tool changing, and feeds it back to the edge computing unit for model optimization.

7. The shield tunneling cutter wear prediction and adjustment system according to claim 6, characterized in that, The geological parameter mapping method includes the following steps: obtaining the installation radius, cutterhead rotation speed, and propulsion speed of each cutter on the shield cutterhead; synthesizing the rotational linear velocity component and axial propulsion speed component of each cutter and integrating them within a statistical period to obtain the actual cumulative cutting stroke of each cutter; using the cutting stroke of the cutter with the smallest installation radius as a reference value, calculating the geological sampling density coefficient of each cutter; based on the geological sampling density coefficient, performing nonlinear resampling on the geological exploration profile data obtained along the tunnel axis to independently generate differentiated geological working condition sequences for each cutter; the output of the method is a standardized geological working condition data sequence.

8. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed, implements a method for predicting and adjusting shield cutter wear as described in any one of claims 1-5.

9. A computer device, characterized in that, The computer device includes a processor and a memory, the memory storing at least one instruction, at least one program, code set or instruction set, the at least one instruction, at least one program, code set or instruction set being loaded and executed by the processor to implement the shield cutter wear prediction and adjustment method as described in any one of claims 1 to 5.

Images