Case data filling and enhancing TBM main parameter design method

Abstract

The invention discloses a case data filling and enhancing TBM (tunnel boring machine) main parameter design method, which comprises the following steps of: obtaining TBM main parameter data in domestic and overseas TBM construction cases, and constructing an original case data set; establishing a sub-database; randomly splitting complete data in each sub-database, and constructing a first training set and a test set; extracting the test set from the original case data set to obtain a second training set; performing multiple interpolation on missing data in the second training set, and finding an optimal multiple interpolation algorithm hyper-parameter through hyper-parameter optimization to obtain a filled and enhanced second training set; unit regression is carried out on the first training set and the filled and enhanced second training set, and two TBM main parameter design models under different cutterhead diameters are established; and performing prediction effect comparison on the two sets of models by using the test set, and verifying the effectiveness of the multi-interpolation method. The method has higher reliability and better universality, and efficient design of the main parameters of the TBM can be achieved.

Description

Technical Field

[0001] The invention relates to a case data filling and enhanced TBM main parameter design method, belonging to the field of hard rock TBM main parameter design. Background Art

[0002] As an efficient and environmentally friendly tunnel construction equipment, the main parameter design of the tunnel boring machine (TBM) is of great significance to the excavation performance. The main parameter design of TBM refers to the reasonable configuration of key parameters such as the cutter thrust, torque, speed, power, etc. of the TBM according to factors such as tunnel size, construction requirements, and geological conditions. The quality of the main parameter design of TBM is directly related to the excavation efficiency and the safety of the construction process. Reasonable main parameter design can enable the TBM to have good rock breaking and escape capabilities during the excavation process, maintain a stable excavation speed, and improve the construction progress. In addition, by optimizing the main parameters, it is also possible to reduce excavation energy consumption, reduce maintenance costs, and further reduce construction costs.

[0003] At present, there is no unified consensus in the industry on the design methods of main parameters. From the perspective of design basis, the mainstream methods can be divided into two types: the formula calculation method (or semi-empirical method) that focuses on theoretical analysis and the empirical design method based on the statistics of current engineering application cases. Based on these two methods, domestic and foreign scholars have conducted a lot of research. The basic principle of the formula calculation method is to study the physical meaning of the main parameters of different types of TBM installations, decompose the main parameters of TBM installations into different components, use theoretical formulas or empirical formulas to calculate the value of each component respectively, integrate the calculation results and determine the main parameters of different types of TBM installations. However, in the formulas given by the research institute, some coefficients are often difficult to give accurate values, and usually a recommended range is given, which has a certain impact on the accuracy of the design results. In addition, there is still a gap between the prediction results of the established theoretical formulas and the case conditions, and most of them are analyzed for individual engineering cases, making it difficult to judge the universality and reliability of the formulas. The empirical design method based on a large number of engineering practices avoids these problems and has long been of great concern to researchers. The basic principle of the empirical design method is to collect a large number of open, single-shield and double-shield TBM selection and design cases, build a selection and design database containing different types and diameters of TBMs, reveal the empirical relationship between the diameters (and other related parameters) of different types of TBMs and the main installation parameters, and build a related evaluation model, and then estimate the main installation parameters of different types of TBMs under different diameters (and other related parameters). Most of the related research is done abroad, and most of them consider the cases of earth pressure balance shield machines (EPB) and slurry balance shield machines. There are relatively few studies on open, single-shield and double-shield TBMs.

[0004] In recent years, the design, construction, and construction technology of TBMs have developed rapidly. The main parameter design formulas obtained based on the limited case analysis in the early years are difficult to meet the on-site requirements, and it is urgent to develop new and more reliable TBM main parameter design formulas. Summary of the Invention

[0005] Aiming at the problem that the current research on the design of TBM main parameters cannot accurately and efficiently provide the optimal reference for the design of TBM main parameters, the present invention provides a method for enhancing the design of TBM main parameters by filling and enhancing case data.

[0006] In order to effectively solve the above problems, the technical solutions provided by the present invention are described in detail:

[0007] In a first aspect, the present invention provides a method for enhancing the design of TBM main parameters by filling and enhancing case data, including the following steps:

[0008] Obtain the TBM main parameter data in domestic and foreign TBM construction cases, and construct an original case dataset; the original case dataset includes complete data and missing data;

[0009] Split the complete data in the original case dataset, and establish sub-databases for each main parameter data corresponding to the cutterhead diameter under different TBM types; randomly split the complete data in each sub-database to construct a first training set and a test set;

[0010] Extract the test set from the original case dataset to obtain a second training set; perform multiple imputations on the missing data in the second training set, and find the best hyperparameters of the multiple imputation algorithm through hyperparameter optimization to obtain a filled and enhanced second training set;

[0011] Perform unit regression on the first training set and the filled and enhanced second training set respectively to establish two sets of TBM main parameter design models under different cutterhead diameters;

[0012] Use the test set to compare the prediction effects of the two sets of models to verify the effectiveness of the multiple imputation method, so as to achieve efficient design of TBM main parameters.

[0013] In a possible implementation manner, the types of TBMs in the domestic and foreign TBM construction cases include three types: open TBM, single shield TBM, and double shield TBM.

[0014] In a possible implementation manner, the main parameters of the TBM main parameter data include: TBM type, cutterhead diameter, rated thrust, maximum thrust, rated torque, maximum breakout torque, maximum cutterhead speed, and cutterhead power.

[0015] In a possible implementation manner, the method for performing multiple imputations on the missing data in the second training set, finding the optimal hyperparameters of the multiple imputation algorithm through hyperparameter optimization, and obtaining the filled and enhanced second training set includes:

[0016] Perform initial imputation on the missing data in the second training set. The initial imputation method is mean imputation or median imputation to obtain an initial imputation set;

[0017] Use machine learning methods to establish multiple multiple imputation prediction models; preset hyperparameter arrays corresponding to the parameters required for each prediction model; perform hyperparameter combination optimization on the prediction models to generate multiple optimized prediction models;

[0018] Use the optimized prediction models to update the missing data in the initial imputation set variable by variable; assume that after the missing data of the nth column variable is updated, use the updated data to predict and fill the (n + 1)th column variable; iterate column by column until all the initially missing data is updated for the first time to achieve the first imputation;

[0019] Perform multiple iterative loop imputations on the training set after the first imputation until the number of imputations converges or reaches the maximum number of iterations to achieve multiple imputations and obtain a multiple imputation training set;

[0020] Perform correlation analysis of each variable with the diameter on the multiple imputation training sets obtained by each optimized prediction model, select the best-performing group, and use the optimized prediction model corresponding to its hyperparameters as the best prediction model, and the obtained multiple imputation training set as the best filled and enhanced second training set.

[0021] In a possible implementation manner, the method for performing simple regression on the first training set and the filled and enhanced second training set respectively includes:

[0022] Use the cutter head diameter as the independent variable and each main parameter as the dependent variable;

[0023] Perform simple regression on the first training set and the filled and enhanced second training set respectively; the simple regression includes linear regression, exponential regression, logarithmic regression, and polynomial regression.

[0024] In a possible implementation manner, the TBM main parameter design model is as follows:

[0025] The rated thrust is the dependent variable:

[0026] y1 = a1x 2 + a2x + a3, for open TBM;

[0027] y1 = b1 lnx + b2, for single shield TBM;

[0028] y1 = cx p , double - shield TBM;

[0029] Among them, y1 is the rated thrust, x is the cutter head diameter, and a1 - a3, b1, b2, c, and p are all coefficients;

[0030] The maximum thrust is the dependent variable:

[0031] y2 = d1x 2 + d2x + d3, open - type TBM;

[0032] y2 = e1x 2 + e2x + e3, single - shield TBM;

[0033] y2 = f1x 2 + f2x + f3, double - shield TBM;

[0034] Among them, y2 is the maximum thrust, x is the cutter head diameter, and d1 - d3, e1 - e3, and f1 - f3 are all coefficients;

[0035] The rated torque is the dependent variable:

[0036] open - type TBM;

[0037] single - shield TBM;

[0038] double - shield TBM;

[0039] Among them, y3 is the rated torque, x is the cutter head diameter, and g1 - g2, q1 - q3, h, and i are all coefficients;

[0040] The maximum breakout torque is the dependent variable:

[0041] open - type TBM;

[0042] single - shield TBM;

[0043] double - shield TBM;

[0044] Among them, y4 is the rated torque, x is the cutter head diameter, and j1 - j2, r1 - r3, k, and l are all coefficients;

[0045] The maximum cutter head speed is the dependent variable:

[0046] open - type TBM;

[0047] single - shield TBM;

[0048] Double-shield TBM;

[0049] Among them, y5 is the maximum cutterhead rotation speed, x is the cutterhead diameter, and m, n, u, and t1 to t3 are all coefficients;

[0050] The cutterhead power is the dependent variable:

[0051] Open TBM;

[0052] Single-shield TBM;

[0053] Double-shield TBM;

[0054] Among them, y6 is the cutterhead power, x is the cutterhead diameter, and α1 to α2, β, γ, and s1 to s3 are all coefficients.

[0055] In a second aspect, the present invention provides a TBM main parameter design device with enhanced case data filling, including:

[0056] A data acquisition module for obtaining TBM main parameter data in domestic and foreign TBM construction cases and constructing an original case dataset; the original case dataset includes complete data and missing data;

[0057] A data splitting module for splitting the complete data in the original case dataset, establishing sub-databases of the main parameter data corresponding to the cutterhead diameter under different TBM types; randomly splitting the complete data in each sub-database to construct a first training set and a test set;

[0058] A filling and enhancement module for extracting the test set from the original case dataset to obtain a second training set; performing multiple imputations on the missing data in the second training set, and finding the optimal hyperparameters of the multiple imputation algorithm through hyperparameter optimization to obtain a filled and enhanced second training set;

[0059] A model construction module for performing unit regression on the first training set and the filled and enhanced second training set respectively to establish two sets of TBM main parameter design models under different cutterhead diameters;

[0060] A verification module for comparing the prediction effects of the two sets of models using the test set to verify the effectiveness of the multiple imputation method, so as to achieve efficient design of TBM main parameters.

[0061] In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements the TBM main parameter design method with enhanced case data filling as described in the first aspect.

[0062] Fourthly, the present invention provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the method for designing the main parameters of a TBM with enhanced case data filling as described in the first aspect.

[0063] Fifthly, the present invention provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the method for designing the main parameters of a TBM with enhanced case data filling as described in the first aspect. Compared with the prior art, the present invention has the following beneficial effects:

[0064] Compared with the theoretical method, the method of the present invention has high decision-making efficiency and higher reliability;

[0065] Compared with the early experience method, in the modeling process of the method of the present invention, the case years corresponding to the supporting data are closer and there are more data cases. It is a data-driven and more objective method for the main parameters of a TBM, with higher reliability and better universality. Description of the Drawings

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0067] Figure 1 It is a flowchart of an efficient design method for the main parameters of a TBM with enhanced data filling;

[0068] Figure 2 It is the basic principle of the multiple imputation algorithm used in the present invention;

[0069] Figure 3 It is the design model of the main parameters of various types of TBMs in the present invention;

[0070] Figure 4 It is the comparison of the prediction effects on the test set before and after multiple imputation in the embodiment of the present invention;

[0071] Figure 5 It is a schematic structural diagram of a device for designing the main parameters of a TBM with enhanced case data filling;

[0072] Figure 6 It is a schematic structural diagram of the electronic device provided by the present invention. Detailed Embodiments

[0073] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0074] Reference Figure 1 , the method for enhancing the design of main parameters of a TBM by filling case data includes the following steps:

[0075] S100, Obtain the main parameter data of TBMs in domestic and foreign TBM construction cases, and construct an original case dataset; the original case dataset includes complete data and missing data.

[0076] In a possible and realistic manner, the types of TBMs in the domestic and foreign TBM construction cases include three types: open TBM, single-shield TBM, and double-shield TBM.

[0077] In a possible implementation manner, the main parameters of the TBM main parameter data include: TBM type, cutterhead diameter, rated thrust, maximum thrust, rated torque, maximum breakout torque, maximum cutterhead speed, and cutterhead power.

[0078] S200, Split the complete data in the original case dataset, and establish sub-databases for the cutterhead diameter corresponding to each main parameter data under different TBM types; randomly split the complete data in each sub-database to construct a first training set and a test set.

[0079] In a possible implementation manner, the complete data in each sub-database is randomly split according to a ratio of 8:2, where 80% is used to construct a first training set (training set 1) as the comparison data for the dataset (second training set, training set 2) after filling and enhancing the case data; 20% is named as the test set as the reliability verification dataset for the main parameter design model.

[0080] S300, Extract the test set from the original case dataset to obtain a second training set; perform multiple imputations on the missing data in the second training set, and find the optimal hyperparameters of the multiple imputation algorithm through hyperparameter optimization to obtain a filled and enhanced second training set.

[0081] In a possible implementation manner, the S300 includes the following sub-steps:

[0082] S310, Perform initial imputation on the missing data in the second training set. The initial imputation method is mean imputation or median imputation, etc., to obtain an initial imputation set for paving the way for subsequent multiple imputation iterations;

[0083] S320. Establish multiple multiple imputation prediction models using machine learning methods; preset hyperparameter arrays corresponding to the parameters required for each prediction model; perform hyperparameter combination optimization on the prediction models, that is, traverse the hyperparameter arrays to generate multiple combinations, thereby generating multiple optimized prediction models;

[0084] S330. Use the optimized prediction models to update the missing data in the initialized imputation set variable by variable; assume that after the missing data of the nth column variable is updated, use this updated data to predict and fill the (n + 1)th column variable; iterate column by column until all the initialized missing data is updated for the first time to achieve the first imputation, see Figure 2 ;

[0085] S340. Perform multiple iterative loop imputations on the training set after the first imputation until the number of imputations converges or reaches the maximum number of iterations to achieve multiple imputations and obtain a multiple imputation training set;

[0086] S350. Perform correlation analysis between each variable and the diameter on the multiple imputation training sets obtained by each optimized prediction model respectively, and select the group with the best model prediction effect. The optimized prediction model corresponding to its hyperparameters is used as the best prediction model, and the obtained multiple imputation training set is used as the second training set with the best filling enhancement.

[0087] Further, in the step S320, the machine learning includes random forest, AdaBoost, XGBoost, and Bayesian ridge regression.

[0088] Further, in the step S320, the required parameters include:

[0089] Random forest: the number of estimators n_estimators, criterion criterion, maximum depth max_depth, minimum samples for splitting min_samples_split, minimum samples for leaf nodes min_samples_leaf, maximum number of features max_features;

[0090] AdaBoost model: the number of estimators n_estimators, learning rate learning_rate, loss function loss;

[0091] XGBoost model: the number of estimators n_estimators, learning rate learning_rate, maximum depth max_depth, subsampling rate subsample, column sampling rate colsample_bytree;

[0092] Bayesian Ridge Regression Model: maximum number of iterations niter, tolerance tol, shape and scale parameters alpha1 and alpha2 of the prior for sample precision, and shape and scale parameters lambda1 and lambda2 of the prior for weight precision.

[0093] Exemplarily, in S320, taking the example of using the AdaBoost model to find the best filling to enhance the training set for the rated thrust of an open TBM, the hyperparameter array corresponding to the parameters required by this model is as follows:

[0094] 'n_estimators': [5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 105, 110, 115, 120, 125, 130, 135, 140, 145, 150, 155, 160, 165, 170, 175, 180, 185, 190, 195, 200]; # Number of estimators

[0095] 'learning_rate': [0.1, 0.3, 0.5, 0.7, 0.9, 1.1, 1.3, 1.5, 1.7, 1.9, 2.1, 2.3, 2.5, 2.7, 2.9]; # Learning rate

[0096] 'loss': ['linear','square', 'exponential'] # Loss function.

[0097] Through hyperparameter optimization, the obtained hyperparameter array is as follows:

[0098] {'n_estimators': 25, 'loss':'square', 'learning_rate': 1.5}.

[0099] Exemplarily, to construct an optimal enhanced dataset for the rated thrust of an open TBM, a multiple imputation strategy driven by multiple machine learning methods is adopted. Taking the AdaBoost algorithm as an example, one of the combination cases is selected through hyperparameter grid search, such as {'n_estimators': 70, 'loss':'square', 'learning_rate': 1.7} as the prediction model parameters, and multiple data imputations are performed on the training set 2 to generate the filled and enhanced training set 2, and then the correlation coefficient between the rated thrust and the cutterhead diameter is analyzed. Change the hyperparameter combination and loop this process.

[0100] On this basis, continue to iteratively execute the following process:

[0101] ① Model expansion: Select other machine learning model algorithms (such as random forest, XGBoost, Bayesian ridge regression) to build a prediction model;

[0102] ② Parameter traversal: Traverse all hyperparameter combinations as possible prediction model parameters;

[0103] ③ Data augmentation: Perform multiple imputations on Training Set 2 based on hyperparameter combinations to generate the augmented Training Set 2 after filling;

[0104] ④ Correlation analysis: Calculate the correlation coefficient between the rated thrust and the cutterhead diameter, and retain the test set data with the highest correlation and the corresponding parameter combinations.

[0105] So far, the test set data finally retained, where the rated thrust is the rated thrust data of the optimally enhanced open TBM, and the corresponding parameter combination used is the best hyperparameter array obtained by hyperparameter optimization.

[0106] For the main parameters (cutterhead torque, cutterhead speed, etc.) and even different TBM models (double shield, single shield, etc.), repeat the above full-process parameter optimization and data imputation. Finally, through the global optimization of TBM models, TBM main parameters, machine learning models, and hyperparameter combinations, integrate the data subsets with the best enhancement effects under each parameter dimension to form a highly confident augmented Training Set 2 after filling.

[0107] S400, perform simple regression on the first training set and the augmented second training set respectively to establish TBM main parameter design models under two different cutterhead diameters.

[0108] In a possible implementation, S400 includes the following sub-steps:

[0109] S410, with the cutterhead diameter as the independent variable and each main parameter as the dependent variable respectively;

[0110] S420, perform simple regression on the first training set and the augmented second training set respectively. Further, the simple regression includes linear regression, exponential regression, logarithmic regression, and polynomial regression. In a possible implementation, the TBM main parameter design model is as follows:

[0111] 1. When the rated thrust is the dependent variable:

[0112] y1 = a1x 2 + a2x + a3, for open TBM;

[0113] y1 = b1 lnx + b2, for single shield TBM;

[0114] y1 = cx p , for double shield TBM;

[0115] Among them, y1 is the rated thrust, x is the cutterhead diameter, and a1 to a3, b1, b2, c, and p are all coefficients;

[0116] 2. The maximum thrust is the dependent variable:

[0117] y2 = d1x 2 + d2x + d3, for open TBM;

[0118] y2 = e1x 2 + e2x + e3, for single shield TBM;

[0119] y2 = f1x 2 + f2x + f3, for double shield TBM;

[0120] Among them, y2 is the maximum thrust, x is the cutterhead diameter, and d1 to d3, e1 to e3, and f1 to f3 are all coefficients;

[0121] 3. The rated torque is the dependent variable:

[0122] For open TBM;

[0123] For single shield TBM;

[0124] For double shield TBM;

[0125] Among them, y3 is the rated torque, x is the cutterhead diameter, and g1 to g2, q1 to q3, h, and i are all coefficients;

[0126] 4. The maximum breakout torque is the dependent variable:

[0127] For open TBM;

[0128] For single shield TBM;

[0129] For double shield TBM;

[0130] Among them, y4 is the rated torque, x is the cutterhead diameter, and j1 to j2, r1 to r3, k, and l are all coefficients;

[0131] 5. The maximum cutterhead rotational speed is the dependent variable:

[0132] For open TBM;

[0133] For single shield TBM;

[0134] For double shield TBM;

[0135] Among them, y5 is the maximum cutter head speed, x is the cutter head diameter, m, n, u and t1~t3 are coefficients;

[0136] 6. Cutter power is the dependent variable:

[0137] Open TBM;

[0138] Single shield TBM;

[0139] Double shield TBM;

[0140] Among them, y6 is the cutter disc power, x is the cutter disc diameter, α1~α2, β, γ and s1~s3 are coefficients.

[0141] S500, using the test set to compare the prediction effects of the two models, verifying the effectiveness of the multiple interpolation method to achieve efficient design of TBM main parameters.

[0142] The following is a more specific example.

[0143] In this embodiment, a total of 42 main parameter data of various types of TBMs that have been put into offline use are collected to construct an original case data set, where the TBM types include open TBM, single shield TBM, and double shield TBM. The collected case data are specifically: TBM type, cutter head diameter, rated thrust, maximum thrust, rated torque, maximum escape torque, maximum cutter head speed, and cutter head power.

[0144] In this embodiment, during the multiple interpolation process, the best prediction model and corresponding hyperparameters are obtained by hyperparameter optimization as follows:

[0145] Open TBM:

[0146]

[0147] Single Shield TBM:

[0148]

[0149]

[0150] Double Shield TBM:

[0151]

[0152] Through the above method, unit regression is obtained as follows Figure 3 The main parameter design model of TBM is shown in Table 1.

[0153] Table 1 TBM main parameter design model

[0154]

[0155]

[0156] Based on the TBM main parameter design model described in Table 1, the model is verified using the validation set. The verification results are as Figure 4 shown. It can be seen from Figure 4 that for the models of the three types of TBMs, the R2 values of the filled and enhanced training set 2 are all greater than those of training set 1 and are also closer to 1, indicating that the models of the filled and enhanced training set 2 fit the data better and are more conducive to the efficient design of the main parameters of TBMs.

[0157] Next, the TBM main parameter design device with case data filling and enhancement provided by the present invention will be described. The TBM main parameter design device with case data filling and enhancement described below can be mutually corresponding and referred to the TBM main parameter design method with case data filling and enhancement described above.

[0158] Figure 5 FIG. is a schematic structural diagram of the TBM main parameter design device with case data filling and enhancement provided by an embodiment of the present invention. As Figure 5 shown, it includes: a data acquisition module 51, a data splitting module 52, a filling and enhancement module 53, a model construction module 54, and a verification module 55, where:

[0159] The data acquisition module 51 is configured to obtain the TBM main parameter data in domestic and foreign TBM construction cases and construct an original case data set; the original case data set includes complete data and missing data;

[0160] The data splitting module 52 is configured to split the complete data in the original case data set, establish a sub-database of the main parameter data corresponding to the cutter head diameter under different TBM types; randomly split the complete data in each sub-database to construct a first training set and a test set;

[0161] The filling and enhancement module 53 is configured to extract the test set from the original case data set to obtain a second training set; perform multiple imputations on the missing data in the second training set, and find the optimal hyperparameters of the multiple imputation algorithm through hyperparameter optimization to obtain a filled and enhanced second training set;

[0162] The model construction module 54 is configured to perform unit regression on the first training set and the filled and enhanced second training set respectively to establish two sets of TBM main parameter design models under different cutter head diameters;

[0163] The verification module 55 is configured to compare the prediction effects of the two sets of models using the test set to verify the effectiveness of the multiple imputation method, so as to achieve efficient design of the main parameters of TBMs.

[0164] Figure 6 Illustrates a schematic diagram of the physical structure of an electronic device, such as Figure 6 shown, the electronic device may include: a processor 610, a communications interface 620, a memory 630, and a communication bus 640. Among them, the processor 610, the communications interface 620, and the memory 630 complete mutual communication through the communication bus 640. The processor 610 can call the logical instructions in the memory 630 to execute the case data filling and enhancing TBM main parameter design method.

[0165] In addition, when the logical instructions in the above-mentioned memory 630 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0166] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the case data filling and enhancing TBM main parameter design method provided by the above-mentioned methods.

[0167] On yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the case data filling and enhancing TBM main parameter design method provided by the above-mentioned methods.

[0168] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0169] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A TBM main parameter design method enhanced by case data filling, characterized in that: The following steps are involved: Obtain TBM main parameter data from TBM construction cases at home and abroad, and construct an original case data set; the original case data set includes complete data and missing data; The complete data in the original case data set is split to establish a sub-database of the main parameter data corresponding to the cutterhead diameter under different TBM types; the complete data in each sub-database is randomly split to construct the first training set and test set; The test set is extracted from the original case data set to obtain a second training set; multiple interpolation is performed on the missing data in the second training set, and the optimal multiple interpolation algorithm hyperparameters are found through hyperparameter optimization to obtain a filled and enhanced second training set; Unit regression was performed on the first training set and the second training set after filling and enhancement, and two sets of TBM main parameter design models with different cutterhead diameters were established. The prediction effects of the two models were compared using the test set to verify the effectiveness of the multiple interpolation method and to achieve efficient design of the main parameters of the TBM.

2. The case data-filled and enhanced TBM main parameter design method according to claim 1 is characterized in that: The types of TBMs in the domestic and foreign TBM construction cases include open TBM, single shield TBM and double shield TBM.

3. The case data-filled and enhanced TBM main parameter design method according to claim 1 is characterized in that: The main parameters of the TBM main parameter data include: TBM type, cutter head diameter, rated thrust, maximum thrust, rated torque, maximum escape torque, maximum cutter head speed and cutter head power.

4. The case data-filled and enhanced TBM main parameter design method according to claim 1 is characterized in that: The method of performing multiple interpolation on the missing data in the second training set, finding the optimal multiple interpolation algorithm hyperparameters by hyperparameter optimization, and obtaining the filled and enhanced second training set includes: Initialize interpolation of missing data in the second training set, using mean interpolation or median interpolation to obtain an initial interpolation set; Use machine learning methods to establish multiple multiple interpolation prediction models; preset the hyperparameter array corresponding to the parameters required for each prediction model; optimize the hyperparameter combination of the prediction model to generate multiple optimized prediction models; The optimization prediction model is used to update the missing data in the initial interpolation set variable by variable. Assuming that the missing data of the nth column variable is updated, the updated data is used to predict and fill the n+1th column variable. The variables are iterated column by column until all the initialized missing data are updated for the first time, thus achieving the first interpolation. Perform multiple iterative cyclic interpolation on the training set after the first interpolation is completed until the interpolation number reaches convergence or reaches the maximum number of iterations, thus realizing multiple interpolation and obtaining a multiple interpolation training set; The correlation analysis between each variable and diameter was performed on the multiple interpolation training sets obtained from each optimal prediction model. The set with the best model prediction effect was selected, and the optimal prediction model corresponding to its hyperparameters was used as the best prediction model. The obtained multiple interpolation training set was used as the second training set for optimal filling enhancement.

5. The case data-filled and enhanced TBM main parameter design method according to claim 1 is characterized in that: The method of performing unit regression on the first training set and the second training set after filling and enhancement respectively comprises: The cutterhead diameter was taken as the independent variable, and each main parameter was taken as the dependent variable; Unit regression is performed on the first training set and the second training set after filling and enhancement respectively; the unit regression includes linear regression, exponential regression, logarithmic regression and polynomial regression.

6. The case data-filled and enhanced TBM main parameter design method according to claim 1 is characterized in that: The TBM main parameter design model is as follows: Rated thrust is the dependent variable: y1=a1x 2 +a2x+a3, open TBM; y1=b1 ln x+b2, single shield TBM; y1=cx p , double shield TBM; Among them, y1 is the rated thrust, x is the cutter head diameter, a1~a3, b1, b2, c and p are all coefficients; Maximum thrust is the dependent variable: y2=d1x 2 +d2x+d3, open TBM; y2=e1x 2 +e2x+e3, single shield TBM; y2=f1x 2 +f2x+f3, double shield TBM; Among them, y2 is the maximum thrust, x is the cutter head diameter, d1~d3, e1~e3 and f1~f3 are coefficients; Rated torque is the dependent variable: Open TBM; Single shield TBM; Double shield TBM; Among them, y3 is the rated torque, x is the cutter head diameter, g1~g2, q1~q3, h and i are all coefficients; The maximum breakaway torque is the dependent variable: Open TBM; Single shield TBM; Double shield TBM; Among them, y4 is the rated torque, x is the cutter head diameter, j1~j2, r1~r3, k and l are all coefficients; The maximum cutterhead speed is the dependent variable: Open TBM; Single shield TBM; Double shield TBM; Among them, y5 is the maximum cutter head speed, x is the cutter head diameter, m, n, u and t1~t3 are coefficients; The cutterhead power is the dependent variable: Open TBM; Single shield TBM; Double shield TBM; Among them, y6 is the cutter disc power, x is the cutter disc diameter, α1~α2, β, γ and s1~s3 are coefficients.

7. A case data-filled and enhanced TBM main parameter design device, characterized in that: include: The data acquisition module is used to obtain the main parameter data of TBM in TBM construction cases at home and abroad and construct the original case data set; the original case data set includes complete data and missing data; The data splitting module is used to split the complete data in the original case data set, establish a sub-database of the main parameter data corresponding to the cutter head diameter under different TBM types; randomly split the complete data in each sub-database to construct the first training set and test set; The filling and enhancement module is used to extract the test set from the original case data set to obtain the second training set; multiple interpolation is performed on the missing data in the second training set, and the optimal multiple interpolation algorithm hyperparameters are found through hyperparameter optimization to obtain the filled and enhanced second training set; The model building module is used to perform unit regression on the first training set and the second training set after filling and enhancement, and establish two sets of TBM main parameter design models under different cutterhead diameters; The verification module is used to compare the prediction effects of the two models using the test set and verify the effectiveness of the multiple interpolation method to achieve efficient design of the main parameters of the TBM.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the TBM main parameter design method enhanced by case data filling as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for designing TBM main parameters enhanced by case data filling is implemented as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for designing TBM main parameters enhanced by case data filling is implemented as described in any one of claims 1 to 6.

Images