TBM master control parameter progressive step-by-step decision method and system

Abstract

This invention belongs to the field of TBM construction technology in underground tunnel engineering, and provides a progressive step-by-step decision-making method and system for TBM master control parameters. The progressive step-by-step decision-making method for TBM master control parameters includes: calculating the breaking energy under different rock mass strengths and penetration depths based on linear rock cutting test data, with the optimal breaking energy as the first decision objective; obtaining the optimal penetration depth range under different rock mass strengths based on the optimal breaking energy constraint range; obtaining the feasible region of TBM master control parameters based on a partitioned weighted TBM rock-machine relationship model and the optimal penetration depth range under different rock mass strengths; within the feasible region of TBM master control parameters, setting the optimal tunneling speed as the second decision objective; and obtaining the final master control parameter decision values ​​based on the known relationship between TBM penetration depth, cutterhead rotation speed, and tunneling speed, thereby achieving progressive step-by-step decision-making for TBM master control parameters.

Description

Technical Field

[0001] This invention belongs to the field of TBM construction technology in underground tunnel engineering, and particularly relates to a progressive step-by-step decision-making method and system for TBM main control parameters. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Tunnel Boring Machines (TBMs) are widely used in the construction of long, deep-buried tunnels due to their high excavation efficiency, minimal disturbance to the surrounding rock, high-quality tunnel completion, good safety, and high degree of automation and informatization. However, the selection of current TBM excavation control parameters lacks scientific basis and relies mainly on human experience. This makes it difficult to adjust and respond promptly and reasonably to changes in rock geological conditions, often resulting in mismatches or inadequacies between TBM control parameters and rock conditions. Consequently, this leads to serious consequences such as slow excavation speed, soaring excavation costs, abnormal wear of cutterhead tools, and even machine jamming and damage. Therefore, research on TBM excavation control parameter optimization and decision-making methods is of great significance for ensuring the safe and efficient excavation of TBMs and has become a research hotspot in the international TBM construction field.

[0004] Most existing TBM tunneling parameter optimization decision-making methods employ either goal optimization or data mining. Goal optimization methods, based on rock breaking physics or rock-machine data mapping, seek the optimal master control parameters within the permissible range of selectable TBM master control parameters, aiming for high tunneling speed or low energy consumption. This yields theoretically optimal tunneling parameters with strong scientific validity, but it doesn't fully leverage excellent driving experience, and its interpretability and practicality need further improvement. Data mining, on the other hand, is a driver behavior learning method. It utilizes machine learning or deep learning algorithms to mine massive amounts of past driving data, using artificial intelligence to learn and imitate the decision-making methods of excellent drivers. It relies heavily on past driving experience, especially the quality of the data and the proportion of excellent driving strategies.

[0005] The inventors discovered that the TBM master control parameter decision results obtained by a single objective optimization method or data mining method cannot simultaneously possess both scientific validity and practicality. Summary of the Invention

[0006] To address the technical problems mentioned above, this invention provides a progressive step-by-step decision-making method and system for TBM master control parameters, which integrates the advantages of both target optimization and driving experience learning, making the decision results both scientific and practical.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] The first aspect of the present invention provides an incremental step-by-step decision-making method for TBM master control parameters.

[0009] In one or more embodiments, a progressive step-by-step decision-making method for TBM master parameters includes:

[0010] Based on rock mass test data, the specific energy of breakage under different rock mass strengths and penetration depths is calculated. Taking the optimal specific energy of breakage as the first decision objective, the optimal penetration depth range under different rock mass strengths is obtained according to the constraint range of the optimal specific energy of breakage.

[0011] Based on the partition-weighted TBM rock-machine relationship model and the optimal penetration range under different rock mass strengths, the feasible region of the TBM master control parameters is obtained; wherein, the partition-weighted TBM rock-machine relationship model is composed of several machine learning models with adaptive weights connected in series; the TBM master control parameters include the TBM penetration and cutterhead rotation speed.

[0012] Within the feasible region of the TBM master control parameters, the optimal tunneling speed is taken as the second decision objective. Then, based on the known relationship between the TBM penetration depth, cutterhead rotation speed and tunneling speed, the final master control parameter decision value is obtained, thereby realizing the incremental step-by-step decision of the TBM master control parameters.

[0013] As one implementation method, the process for determining the adaptive weights of the partition-weighted TBM rock-machine relationship model is as follows:

[0014] Based on the distribution characteristics of rock mass parameters, rock mass parameters are partitioned, and their performance on samples in different partitions is weighted differently.

[0015] As one implementation method, the process for determining the optimal penetration range under different rock mass strengths is as follows:

[0016] Based on rock mass test data, the actual rock breaking physical laws of the cutting tool and the rock mass are obtained. Then, through multi-parameter regression analysis, the formula for the change of the breaking energy of the two variables of rock mass strength and penetration is obtained.

[0017] Based on the definition of the optimal fracture energy range, the optimal penetration range under different rock mass strengths is obtained.

[0018] As one implementation method, the machine learning model includes support vector regression, random forest, and decision tree algorithms.

[0019] As one implementation method, the training sample data of the partition-weighted TBM rock-machine relationship model is the actual data of TBM tunneling sites under different geological conditions; the actual data of TBM tunneling sites includes rock mass parameters characterized by uniaxial compressive strength, joint frequency and long axis size of rock debris, and tunneling parameters characterized by cutterhead thrust, torque, cutterhead speed and penetration.

[0020] A second aspect of the present invention provides an incremental step-by-step decision-making system for TBM master control parameters.

[0021] In one or more embodiments, a progressive step-by-step decision-making system for TBM master parameters includes:

[0022] The optimal penetration range determination module is used to calculate the fracture energy under different rock mass strengths and penetrations based on rock mass test data. Taking the optimal fracture energy as the first decision objective, the optimal penetration range under different rock mass strengths is obtained according to the optimal fracture energy constraint range.

[0023] The feasible region determination module for master control parameters is used to obtain the feasible region of TBM master control parameters based on the partition-weighted TBM rock-machine relationship model and the optimal penetration range under different rock mass strengths; wherein, the partition-weighted TBM rock-machine relationship model is composed of several machine learning models with adaptive weights connected in series; the TBM master control parameters include the penetration of TBM and the cutterhead rotation speed.

[0024] The main control parameter decision determination module is used to determine the optimal tunneling speed as the second decision objective within the feasible region of the TBM main control parameters. Then, based on the known relationship between the TBM penetration depth, cutterhead rotation speed and tunneling speed, the final decision value of the main control parameters is obtained, thereby realizing the progressive step-by-step decision of the TBM main control parameters.

[0025] As one implementation method, in the main control parameter feasible region determination module, the process for determining the adaptive weights of the partition-weighted TBM rock-machine relationship model is as follows:

[0026] Based on the distribution characteristics of rock mass parameters, rock mass parameters are partitioned, and their performance on samples in different partitions is weighted differently.

[0027] As one implementation method, in the optimal penetration range determination module, the process for determining the optimal penetration range under different rock mass strengths is as follows:

[0028] Based on rock mass test data, the actual rock breaking physical laws of the cutting tool and the rock mass are obtained. Then, through multi-parameter regression analysis, the formula for the change of the breaking energy of the two variables of rock mass strength and penetration is obtained.

[0029] Based on the definition of the optimal fracture energy range, the optimal penetration range under different rock mass strengths is obtained.

[0030] As one implementation, in the master control parameter feasible region determination module, the machine learning model includes support vector regression, random forest and decision tree algorithms.

[0031] As one implementation method, in the main control parameter feasible domain determination module, the training sample data of the partition-weighted TBM rock-machine relationship model are actual data from TBM tunneling sites under different geological conditions; the actual data from TBM tunneling sites include rock mass parameters characterized by uniaxial compressive strength, joint frequency, and long axis size of rock debris, and tunneling parameters characterized by cutterhead thrust, torque, cutterhead rotation speed, and penetration depth.

[0032] A third aspect of the present invention provides a computer-readable storage medium.

[0033] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the TBM master control parameter progressive step-by-step decision-making method as described above.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] This invention constructs a partitioned weighted TBM rock-machine relationship model based on an adaptive weighted machine learning model, obtaining the rock-machine mutual feedback law. It then sequentially optimizes TBM tunneling parameters under different geological conditions, using optimal specific energy and efficient tunneling as optimization objectives. This leads to the establishment of a progressive step-by-step decision-making method for the main control parameters based on the rock-machine mutual feedback law. This method integrates the advantages of both objective optimization and driver experience learning, making the decision results both scientific and practical, achieving high efficiency and low energy consumption, and promoting the optimization of TBM tunneling parameters from theory to practical application.

[0036] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0037] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0038] Figure 1 This is a flowchart of the progressive step-by-step decision-making method for TBM master control parameters according to an embodiment of the present invention;

[0039] Figure 2 It is the penetration value calculated according to the embodiments of the present invention.

[0040] Figure 3 This is a schematic diagram of the TBM master control parameter progressive step-by-step decision-making system in an embodiment of the present invention. Detailed Implementation

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0042] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, 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.

[0043] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0044] Reference Figure 1 This paper presents a flowchart of a progressive step-by-step decision-making method for TBM master control parameters. This embodiment first explores the interaction between rock mass mechanical properties and TBM mechanical properties. Based on this, it constructs the rock-machine interaction law and relationship, which is fundamental to improving the adaptability of TBM tunneling parameters to rock mass changes and optimizing parameters. This embodiment proposes to combine the advantages of theoretical analysis and data statistics to explore the rock-machine law. The theoretical analysis relies on full-scale or scaled-down linear / rotary rock breaking tests to discover and reveal the rock breaking mechanism of the cutter head under different working conditions, providing physical laws and prior information for exploring the rock-machine relationship. The empirical statistics rely on massive amounts of actual field data under different geological conditions, using regression methods to explore the response laws between data.

[0045] This implementation constructs the rock-machine interaction law and relationship, namely the correlation between rock mass parameters and TBM mechanical parameters. Currently, research on the rock-machine mapping relationship in TBM tunneling has gradually shifted from single-indicator to multi-indicator indicators, and the construction methods have evolved from simple regression to artificial intelligence. To meet the needs of TBM control parameter optimization decision-making, this embodiment improves the TBM rock-machine mapping relationship in the following two aspects:

[0046] (1) Machine learning models exhibit significant "black box" characteristics. When the data samples used to train the model are relatively discrete, the model trained solely through machine learning methods is easily affected by "outliers," often leading to overfitting and poor generalization. To address this, this embodiment introduces the aforementioned qualitative rock-machine data patterns as constraints for the machine learning model, and analyzes and filters the data samples. For "outliers" with poor rationality, their sample weights during training are reduced to minimize their negative impact on model accuracy and generalization, thereby improving the generalization and accuracy of the trained model.

[0047] (2) The variation patterns of TBM rock-machine data show significant differences within different rock mass parameter ranges. Single-method machine learning models can usually only achieve good mining results under local rock mass conditions, but their applicability to rock mass conditions is limited. When rock mass conditions change, the computational accuracy of the model is also affected. To address this, this embodiment assigns adaptive weights to multiple machine learning models and concatenates them. Models that perform better within the current rock mass condition range automatically receive higher weights, thereby improving the adaptability and computational accuracy of the rock-machine data model. For example, based on support vector regression, random forest, and decision tree algorithms, sub-models are established with the main control parameters (penetration degree and cutterhead speed) as outputs. Rock mass parameters are partitioned according to their distribution characteristics. The performance of the three sub-models on samples in different partitions is used to differentiate the weights, thus establishing a partition-weighted TBM rock-machine relationship model.

[0048] The following is combined Figure 1 The TBM master control parameter progressive step-by-step decision-making method of this embodiment specifically includes the following steps:

[0049] Step 1: Based on rock mass test data, calculate the fracture energy under different rock mass strengths and penetration depths. Taking the optimal fracture energy as the first decision objective, and based on the optimal fracture energy constraint range, obtain the optimal penetration depth range under different rock mass strengths.

[0050] In the specific implementation of step 1, the process of determining the optimal penetration range under different rock mass strengths is as follows:

[0051] Based on rock mass test data, the actual rock breaking physical laws of the cutting tool and the rock mass are obtained. Then, through multi-parameter regression analysis, the formula for the change of the breaking energy of the two variables of rock mass strength and penetration is obtained.

[0052] Based on the definition of the optimal fracture energy range, the optimal penetration range under different rock mass strengths is obtained.

[0053] It should be noted that the rock mass test data are based on model tests such as linear cutting tests, rotary cutting tests, and data simulations. Among them, the linear cutting test of rock mass includes full-scale or scaled-down linear / rotary rock breaking tests.

[0054] Step 2: Based on the partition-weighted TBM rock-machine relationship model and the optimal penetration range under different rock mass strengths, obtain the feasible region of the TBM master control parameters; wherein, the partition-weighted TBM rock-machine relationship model is composed of several machine learning models with adaptive weights connected in series; the TBM master control parameters include the TBM penetration and cutterhead rotation speed.

[0055] The process for determining the adaptive weights of the partition-weighted TBM rock-machine relationship model is as follows:

[0056] Based on the distribution characteristics of rock mass parameters, rock mass parameters are partitioned, and their performance on samples in different partitions is weighted differently.

[0057] It should be noted that the machine learning models mentioned here include, but are not limited to, support vector regression, random forest, and decision tree algorithms.

[0058] In the specific implementation process, the training sample data of the partition-weighted TBM rock-machine relationship model are the actual data of TBM tunneling sites under different geological conditions; the actual data of TBM tunneling sites include rock mass strength, TBM penetration depth and cutterhead rotation speed.

[0059] Step 3: Within the feasible region of the TBM master control parameters, the optimal tunneling speed is taken as the second decision objective. Then, based on the known relationship between the TBM penetration depth, cutterhead rotation speed and tunneling speed, the final master control parameter decision value is obtained, thereby realizing the progressive step-by-step decision of the TBM master control parameters.

[0060] The known relationship between the penetration depth, cutterhead rotation speed, and tunneling speed of the TBM is: Tunneling speed = penetration depth × cutterhead rotation speed.

[0061] Here is a specific example:

[0062] The model tests involved rock samples with three strengths: 60.78 MPa, 120.47 MPa, and 179.65 MPa. Therefore, the regression equation relating fracture energy (SE) and penetration depth was divided into three ranges. When the rock mass strength (UCS) is within this range, the fracture energy (SE) and penetration depth (p) satisfy the following relationship: UCS < 90 MPa:

[0063] SE = 1.8p 2 -17.53p +84.06(1)

[0064] 90MPa <UCS<150MPa:

[0065] SE = 3.31p 2 -18.26p +102.89(2)

[0066] 150MPa <UCS:

[0067] SE = 25.43p 2 -91.57p+170.53(3)

[0068] Within the aforementioned intervals, each interval has an optimal specific energy value and a corresponding penetration value. To make this decision-making method more applicable in practical operations, an optimal specific energy range is defined within each interval, which corresponds to an optimal penetration range. The parameter decision in the second step is performed within this range. The optimal specific energies for the three intervals are 41.38 MJ / m³, 77.87 MJ / m³, and 88.09 MJ / m³, respectively. The optimal specific energy and penetration range for each interval are defined as shown in Table 1.

[0069] Table 1 Penetration constraint range under different rock mass strength ranges

[0070]

[0071] Assuming a certain working condition, the average values ​​of various statistical parameters are used as the reference values: compressive strength 52.4 MPa, joint frequency 2.5 m⁻¹, and rock fragment particle size 30 mm. A floating range is given to the joint frequency and rock fragment particle size under this working condition, namely 2.1-2.9 m⁻¹ and 26-34 mm, with step sizes of 0.2 m⁻¹ and 2 mm, respectively. After permutation and combination, a total of 25 sets of input parameters are substituted into the partitioned weighted model to obtain 25 sets of penetration depth and cutterhead rotation speed values. The calculated penetration depth values ​​are as follows: Figure 2 As shown.

[0072] Figure 2 The horizontal and vertical axes represent the joint frequency and rock fragment particle size of the selected samples, respectively. The colors of the 25 sample points on the graph represent the penetration values ​​calculated by the model; lighter colors indicate darker penetration values, and darker colors indicate greater penetration values. Based on the formula for the variation of the fracture specific energy of the two variables (compressive strength and penetration) and the optimal specific energy range, the penetration range is constrained to be 2.68–7.06 mm. Based on this, the 25 sets of main control parameter combinations can be screened. For example... Figure 2As shown by the black dashed line, the seven sets of values ​​in the lower left corner of the dashed line all satisfy the above constraints, resulting in seven sets of values ​​that meet the requirements. These seven sets of samples represent the control parameter combinations that meet the specific energy requirements, determined by the driver's tunneling experience under the example working condition. Based on this, the tunneling speed corresponding to the above seven sets of main control parameter values, i.e., the product of the two, is calculated, as shown in Table 1. The calculation results show that the optimal penetration depth is 7.0 mm / r, the optimal cutterhead speed is 5.3 rpm, and the corresponding tunneling speed is 37.1 mm / min.

[0073] Table 1. Sample tunneling speed values ​​within the feasible region.

[0074] Joint frequency (m-1) Rock slag particle size (mm) Penetration (mm / r) Cutter head rotation speed (rpm) Tunneling speed (mm / min) 2.1 26 6.7 5.4 36.1 2.1 28 7.0 5.3 37.1 2.3 26 6,7 5.4 35.9 2.3 28 7.0 5.2 36.8 2.5 26 6.7 5.3 35.7 2.5 28 7.1 5.1 36.6 2.7 26 6.8 5.2 35.5

[0075] Reference Figure 3 A schematic diagram of an asymptotic step-by-step decision-making system for TBM master parameters is presented. Figure 3 In this context, the TBM master control parameter progressive step-by-step decision-making system includes:

[0076] (1) Optimal penetration range determination module, which is used to calculate the breaking energy under different rock mass strengths and penetrations based on rock mass test data, with the optimal breaking energy as the first decision objective, and obtain the optimal penetration range under different rock mass strengths according to the optimal breaking energy constraint range.

[0077] Specifically, in the optimal penetration range determination module, the process for determining the optimal penetration range under different rock mass strengths is as follows:

[0078] Based on rock mass test data, the actual rock breaking physical laws of the cutting tool and the rock mass are obtained. Then, through multi-parameter regression analysis, the formula for the change of the breaking energy of the two variables of rock mass strength and penetration is obtained.

[0079] Based on the definition of the optimal fracture energy range, the optimal penetration range under different rock mass strengths is obtained.

[0080] (2) The feasible domain determination module for master control parameters is used to obtain the feasible domain of TBM master control parameters based on the partition-weighted TBM rock-machine relationship model and the optimal penetration range under different rock mass strengths; wherein, the partition-weighted TBM rock-machine relationship model is composed of several machine learning models with adaptive weights connected in series; the TBM master control parameters include the penetration of TBM and the cutterhead rotation speed.

[0081] Specifically, in the module for determining the feasible region of the master control parameters, the process for determining the adaptive weights of the partition-weighted TBM rock-machine relationship model is as follows:

[0082] Based on the distribution characteristics of rock mass parameters, rock mass parameters are partitioned, and their performance on samples in different partitions is weighted differently.

[0083] It should be noted here that in the feasible region determination module of the master control parameters, the machine learning model includes support vector regression, random forest and decision tree algorithms.

[0084] In the main control parameter feasible domain determination module, the training sample data of the partition-weighted TBM rock-machine relationship model are actual data of TBM tunneling sites under different geological conditions; the actual data of TBM tunneling sites include rock mass strength, TBM penetration depth and cutterhead rotation speed.

[0085] (3) The main control parameter decision determination module is used to determine the optimal tunneling speed as the second decision objective within the feasible domain of the TBM main control parameters. Then, based on the known relationship between the penetration depth of the TBM, the cutterhead rotation speed and the tunneling speed, the final decision value of the main control parameters is obtained, thereby realizing the progressive step-by-step decision of the TBM main control parameters.

[0086] It should also be noted that each module in the TBM master parameter progressive step-by-step decision system corresponds one-to-one with each step in the TBM master parameter progressive step-by-step decision method, and their specific implementation processes are the same, so they will not be described in detail here.

[0087] In one or more embodiments, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the TBM master control parameter progressive step-by-step decision-making method as described above.

[0088] In one or more embodiments, an electronic device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the TBM master control parameter progressive step-by-step decision method as described above.

[0089] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A progressive step-by-step decision-making method for TBM master control parameters, characterized in that, include: Based on rock mass test data, the specific energy of breakage under different rock mass strengths and penetration depths is calculated. Taking the optimal specific energy of breakage as the first decision objective, the optimal penetration depth range under different rock mass strengths is obtained according to the constraint range of the optimal specific energy of breakage. Based on the partitioned weighted TBM rock-machine relationship model and the optimal penetration range under different rock mass strengths, the feasible region of the TBM master control parameters is obtained; the TBM master control parameters include the TBM penetration and the cutterhead rotation speed. Within the feasible region of the TBM master control parameters, the optimal tunneling speed is taken as the second decision objective. Then, based on the known relationship between the TBM penetration depth, cutterhead rotation speed and tunneling speed, the final master control parameter decision value is obtained, thereby realizing the incremental step-by-step decision of the TBM master control parameters. The training sample data for the partition-weighted TBM rock-machine relationship model are actual data from TBM tunneling sites under different geological conditions. The actual data from the TBM tunneling site includes rock mass strength, TBM penetration depth, and cutterhead rotation speed. The partitioned weighted TBM rock-machine relationship model is composed of several machine learning models with adaptive weights connected in series. The machine learning models select three basic algorithms: support vector regression, random forest, and decision tree. The TBM master control parameters are used as outputs to test the models of the above three basic algorithms. Based on the test results, the training sample data of the above three basic algorithms are partitioned. Based on the partitioning results, the output results of the above three basic algorithms are weighted and summed to obtain the feasible region of the TBM master control parameters. The process of determining the adaptive weights of the partitioned weighted TBM rock-machine relationship model is as follows: rock mass parameters are partitioned according to the distribution characteristics of rock mass parameters, and differentiated weights are assigned based on the performance of samples in different partitions.

2. The progressive step-by-step decision-making method for TBM master control parameters as described in claim 1, characterized in that, The process for determining the optimal penetration range under different rock mass strengths is as follows: Based on rock mass test data, the actual rock breaking physical laws of the cutting tool and the rock mass are obtained. Then, through multi-parameter regression analysis, the formula for the change of the breaking energy of the two variables of rock mass strength and penetration is obtained. Based on the definition of the optimal fracture energy range, the optimal penetration range under different rock mass strengths is obtained.

3. The progressive step-by-step decision-making method for TBM master control parameters as described in claim 1, characterized in that, The machine learning models include support vector regression, random forest, and decision tree algorithms.

4. A progressive step-by-step decision-making system for TBM master control parameters, characterized in that, include: The optimal penetration range determination module is used to calculate the fracture energy under different rock mass strengths and penetrations based on rock mass test data. Taking the optimal fracture energy as the first decision objective, the optimal penetration range under different rock mass strengths is obtained according to the optimal fracture energy constraint range. The feasible region determination module for master control parameters is used to obtain the feasible region of TBM master control parameters based on the partition-weighted TBM rock-machine relationship model and the optimal penetration range under different rock mass strengths; the TBM master control parameters include the penetration of TBM and the cutterhead rotation speed. The main control parameter decision determination module is used to determine the optimal tunneling speed as the second decision objective within the feasible domain of the TBM main control parameters. Then, based on the known relationship between the TBM penetration depth, cutterhead rotation speed and tunneling speed, the final decision value of the main control parameters is obtained, thereby realizing the progressive step-by-step decision of the TBM main control parameters. The training sample data for the partition-weighted TBM rock-machine relationship model are actual data from TBM tunneling sites under different geological conditions. The actual data from the TBM tunneling site includes rock mass strength, TBM penetration depth, and cutterhead rotation speed. The partitioned weighted TBM rock-machine relationship model is composed of several machine learning models with adaptive weights connected in series. The machine learning models select three basic algorithms: support vector regression, random forest, and decision tree. The TBM master control parameters are used as outputs to test the models of the above three basic algorithms. Based on the test results, the training sample data of the above three basic algorithms are partitioned. Based on the partitioning results, the output results of the above three basic algorithms are weighted and summed to obtain the feasible region of the TBM master control parameters. The process of determining the adaptive weights of the partitioned weighted TBM rock-machine relationship model is as follows: rock mass parameters are partitioned according to the distribution characteristics of rock mass parameters, and differentiated weights are assigned based on the performance of samples in different partitions.

5. The TBM master control parameter progressive step-by-step decision-making system as described in claim 4, characterized in that, In the main control parameter feasible region determination module, the process for determining the adaptive weights of the partition-weighted TBM rock-machine relationship model is as follows: Based on the distribution characteristics of rock mass parameters, rock mass parameters are partitioned, and their performance on samples in different partitions is weighted differently.

6. The TBM master control parameter progressive step-by-step decision-making system as described in claim 4, characterized in that, In the optimal penetration range determination module, the process for determining the optimal penetration range under different rock mass strengths is as follows: Based on rock mass test data, the actual rock breaking physical laws of the cutting tool and the rock mass are obtained. Then, through multi-parameter regression analysis, the formula for the change of the breaking energy of the two variables of rock mass strength and penetration is obtained. Based on the definition of the optimal fracture energy range, the optimal penetration range under different rock mass strengths is obtained.

7. The TBM master control parameter progressive step-by-step decision-making system as described in claim 4, characterized in that, In the main control parameter feasible region determination module, the machine learning model includes support vector regression, random forest and decision tree algorithms; Alternatively, in the main control parameter feasible domain determination module, the training sample data of the partition-weighted TBM rock-machine relationship model is the actual data of TBM tunneling sites under different geological conditions; The actual data from the TBM tunneling site includes rock mass strength, TBM penetration depth, and cutterhead rotation speed.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the progressive step-by-step decision-making method for TBM master control parameters as described in any one of claims 1-3.

Images