Intelligent determining method for tunnel lining type

Abstract

The invention discloses an intelligent determination method for a tunnel lining type. The method comprises the following steps: acquiring a data set of a designed tunnel; according to the designed tunnel data set, constructing a lining type sub-data set; according to the lining type sub data set, a GWO-NGBoost lining type prediction model is constructed; for a to-be-predicted tunnel, determining section division of surrounding rock grades based on geological survey data; and for each section of the to-be-predicted tunnel, respectively extracting features and inputting the features into the lining type prediction model of the corresponding surrounding rock grade to realize prediction of the lining type. According to the method, the GWO-NGBoost model is established in each tunnel lining type sub-data set, lining type prediction is realized, the whole process is standard and unified, and automatic and efficient completion can be realized. According to the method, the corresponding lining type prediction model is constructed in each tunnel lining type sub-data set, and a plurality of lining type prediction models are finally constructed, so that the lining type prediction precision can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel design, and particularly relates to an intelligent determination method for tunnel lining types. Background Art

[0002] At present, the selection of tunnel lining types depends on manual experience for determination. The main process is as follows: (1) Determine the surrounding rock grades based on geological exploration data, and divide the tunnel into sections with different surrounding rock grades; (2) For sections with the same surrounding rock grade, determine the lining type according to geological conditions such as burial depth, bias pressure, and groundwater development; (3) Gradually determine the lining types of each section of surrounding rock grade until all sections are determined. Therefore, the existing design method standards are too vague and highly dependent on the experience of designers. In addition, the design efficiency is low, and it is necessary to manually determine the lining types of each section step by step. Therefore, there is an urgent need for an intelligent calculation method to improve the determination efficiency and accuracy of tunnel lining types. Summary of the Invention

[0003] In view of the above problems, the present invention is proposed to provide an intelligent determination method for tunnel lining types that overcomes the above problems or at least partially solves the above problems.

[0004] To solve the above technical problems, the embodiments of the present application disclose the following technical solutions:

[0005] In a first aspect, an embodiment of the present invention discloses an intelligent determination method for tunnel lining types, including:

[0006] S100. Obtain a data set of designed tunnels;

[0007] S200. Construct a sub-data set of lining types according to the data set of the designed tunnels;

[0008] S300. Construct a GWO-NGBoost lining type prediction model according to the sub-data set of lining types;

[0009] S400. For the tunnel to be predicted, determine the section division of the surrounding rock grades based on geological exploration data;

[0010] S500. For each section of the tunnel to be predicted, extract features and input them into the lining type prediction model of the corresponding surrounding rock grade to achieve the prediction of the lining type.

[0011] Further, in S100, the data set of the designed tunnels includes at least the longitude and latitude of the tunnel, the tunnel length, the starting and ending mileage of the surrounding rock section, the burial depth at multiple mileage in the tunnel, and the corresponding surrounding rock grade and lining type.

[0012] Further, in S200, according to the designed tunnel dataset, a lining type sub-dataset is constructed. The specific method includes: extracting and transforming the features of the obtained dataset, dividing the data with the same surrounding rock grade into a sub-dataset, resampling the sub-dataset to form a lining type sub-dataset, and finally constructing multiple lining type sub-datasets corresponding to multiple surrounding rock grades.

[0013] Further, the method for extracting and transforming the features of the obtained dataset specifically includes: extracting the tunnel longitude and latitude, tunnel length, surrounding rock section length, and buried depth features from the dataset, and performing plane coordinate transformation on the tunnel longitude and latitude features to convert the geographical coordinates such as longitude and latitude into plane coordinates. The coordinate transformation formula is as shown in formulas (1) to (4):

[0014]

[0015] In the formula, x log and x lat are the longitude and latitude values of the tunnel respectively, x log_sin and x log_cos are the coordinate values after longitude conversion, and x lat_sin and x lat_cos are the coordinate values after latitude conversion.

[0016] Further, the method for resampling the sub-dataset to form a lining type sub-dataset specifically includes:

[0017] Downsampling the sub-dataset. In the sub-dataset, for samples with too many of a certain lining type, randomly select one sample without repetition each time until the specified number of samples is selected; Upsampling the sub-dataset. In the sub-dataset, for samples with too few of a certain lining type, generate a new specified number of samples based on the Smote algorithm. The specific method of the Smote algorithm includes: randomly selecting a sample with too few of a certain lining type, calculating the k nearest neighbors of this sample, randomly selecting 1 sample from the k neighbors, and randomly selecting 1 point between this neighbor sample and the sample point, and taking it as the newly generated sample. Repeat the above steps until the specified number of samples is generated.

[0018] Further, the method for finally constructing multiple lining type sub-datasets corresponding to multiple surrounding rock grades specifically includes: repeating the resampling step for the sub-datasets corresponding to each surrounding rock grade until all sub-datasets corresponding to all surrounding rock grades are covered, that is, constructing the lining type sub-datasets corresponding to all surrounding rock grades.

[0019] Further, in S300, according to the lining type sub-dataset, a GWO-NGBoost lining type prediction model is constructed. The specific method includes:

[0020] S301. Dataset division: In the same lining type sub-dataset, divide the dataset into a training set and a test set. The training set is used to build the model and find the optimal hyperparameters, and the test set is used to evaluate the model.

[0021] S302. Use the GWO algorithm to find the optimal hyperparameters of the NGBoost model.

[0022] S303. Build a lining type prediction model. For each lining type sub-dataset, establish an NGBoost lining type prediction model on the entire training set based on the obtained optimal hyperparameters, and finally obtain multiple NGBoost lining type prediction models corresponding to multiple lining type sub-datasets.

[0023] Furthermore, in S302, the specific method for using the GWO algorithm to find the optimal hyperparameters of the NGBoost model includes:

[0024] S3021. Set the hyperparameter range: Set the value ranges of the parameters n_estimators, learning_rate, minibatch_frac, and col_sample in the NGBoost model.

[0025] S3022. Initialize the positions of the wolf pack: Randomly select a number within the value range for each hyperparameter, and generate a list consisting of M groups of hyperparameters, which is the initialized position of the wolf pack.

[0026] S3023. Calculate the fitness values of the wolf pack individuals and save the best three wolves. Respectively set the position parameters of each wolf as the hyperparameter values of the NGBoost model, and use the negative value of the average precision rate based on K-fold cross-validation as the fitness value of this wolf individual. The calculation formula of the precision rate is shown in Formula 5. The precision rate represents the proportion of samples that are truly positive among the samples predicted as positive by the model.

[0027]

[0028] In the formula, Precision is the precision rate, TP is the sample predicted as positive and actually positive, and FP is the sample predicted as positive and actually negative. After obtaining the fitness values of the wolf pack individuals, select the three wolves with the smallest fitness and save them as α, β, and δ wolves.

[0029] S3024. Calculate the correlation coefficients. The three coefficients of the convergence factor a, parameter C, and parameter A are calculated according to Formulas 6 - 8:

[0030]

[0031] C i =2×r i (7)

[0032] A i = 2 × a × r i -a (8)

[0033] Where t is the current iteration number, t o is the total number of iterations, i takes values 1, 2, 3, r i is a random number between 0 and 1;

[0034] S3025. Update the positions of the wolf pack. Update the positions of the wolf pack according to the positions of the α, β, and δ wolves. The update formula is shown in Equations 9 - 11:

[0035]

[0036] Where X represents the positions of the wolf pack, X α , X β , X δ represent the positions of the α, β, and δ wolves, and X(t + 1) represents the updated positions of the wolf pack;

[0037] S3206. Obtain the optimal hyperparameters. Repeat S3203 - S3205 until the number of iterations reaches the maximum number of iterations. Then, the position of the α wolf is the obtained optimal hyperparameters.

[0038] Furthermore, in S500, for each section of the tunnel to be predicted, features are extracted and input into the lining type prediction model corresponding to the surrounding rock grade to achieve the prediction of the lining type. The specific method includes:

[0039] S501. Feature extraction and transformation. On the section of the surrounding rock grade, extract the features of buried depth, tunnel longitude and latitude, tunnel length, surrounding rock section length, and perform transformation;

[0040] S502. Obtain the lining type of each section. For each section of the surrounding rock grade, find the lining type prediction model corresponding to the surrounding rock grade, and input the features extracted in S501 into the lining type prediction model to obtain the lining type of the corresponding section;

[0041] S503. Determine the lining type. Detect each adjacent section in S502. If the lining types of adjacent sections are the same, they are merged into one section. The merged section and the corresponding lining type are the finally determined lining types.

[0042] In a second aspect, an embodiment of the present invention discloses an electronic device, including:

[0043] One or more processors;

[0044] A memory for storing one or more programs;

[0045] When the one or more programs are executed by the one or more processors, the one or more processors implement an intelligent determination method for the tunnel lining type.

[0046] The beneficial effects of the above technical solutions provided by the embodiments of the present invention at least include:

[0047] The present invention discloses an intelligent determination method for tunnel lining types, including: obtaining a data set of designed tunnels; constructing a lining type sub-data set according to the designed tunnel data set; constructing a GWO-NGBoost lining type prediction model according to the lining type sub-data set; for a tunnel to be predicted, determining the section division of the surrounding rock grade based on geological exploration data; for each section of the tunnel to be predicted, extracting features and inputting them into the lining type prediction model corresponding to the surrounding rock grade to achieve the prediction of the lining type. The present invention realizes the prediction of the lining type by establishing a GWO-NGBoost model in each tunnel lining type sub-data set, and the whole process is standard and unified, and can be completed efficiently and automatically. The present invention constructs corresponding lining type prediction models in each tunnel lining type sub-data set, and finally constructs multiple lining type prediction models, which can effectively improve the accuracy of lining type prediction.

[0048] The technical solutions of the present invention will be further described in detail below with reference to the drawings and embodiments. Description of the Drawings

[0049] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0050] Figure 1 is a flowchart of an intelligent determination method for a tunnel lining type in Embodiment 1 of the present invention;

[0051] Figure 2 is a flowchart of constructing a GWO-NGBoost lining type model in Embodiment 1 of the present invention;

[0052] Figure 3 is a process diagram of lining type prediction in Embodiment 1 of the present invention;

[0053] Figure 4 is a schematic structural diagram of an electronic device in Embodiment 2 of the present invention. Detailed Embodiments

[0054] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be completely conveyed to those skilled in the art.

[0055] To solve the problems existing in the prior art, an embodiment of the present invention provides an intelligent determination method for tunnel lining types.

[0056] Embodiment 1

[0057] The present invention discloses an intelligent determination method for tunnel lining types, as Figure 1 , including:

[0058] S100. Obtain a dataset of designed tunnels; wherein, the dataset of designed tunnels at least includes tunnel longitude and latitude, tunnel length, starting and ending mileage of surrounding rock sections, depths of multiple mileage in the tunnel, and corresponding surrounding rock grades and lining types.

[0059] Specifically, the specific data of the dataset of designed tunnels is shown in Table 1, and the data of a tunnel to be designed, including tunnel longitude and latitude, tunnel length, starting and ending mileage of surrounding rock sections, depths of multiple mileage in the tunnel, and corresponding surrounding rock grades, is specifically shown in Table 2.

[0060] Table 1 Dataset of a certain designed tunnel

[0061]

[0062] Table 2 Dataset table of a certain tunnel to be designed

[0063]

[0064] S200. Construct a sub-dataset of lining types according to the dataset of designed tunnels; in S200 of this embodiment, constructing a sub-dataset of lining types according to the dataset of designed tunnels, the specific method includes:

[0065] S201. Feature extraction and transformation, extract tunnel longitude and latitude, tunnel length, surrounding rock section length, depth features from the dataset, and perform plane coordinate transformation on the tunnel longitude and latitude features, converting the geographical coordinates such as longitude and latitude into plane coordinates, and the coordinate transformation formula is as shown in formulas (1) to (4):

[0066]

[0067] In the formula, x log , x latThey are the longitude and latitude values of the tunnel, x log_sin 、x log_cos are the coordinate values after longitude conversion, x lat_sin 、x lat_cos are the coordinate values after latitude conversion.

[0068] S202. Split the sub-datasets. Split the data in the dataset with the same surrounding rock grade after feature extraction into the same sub-dataset, so as to split the dataset into multiple sub-datasets. Among them, each sub-dataset corresponds to a surrounding rock grade.

[0069] S203. Perform resampling. In the same sub-dataset, there are multiple lining types, and the corresponding quantities of the lining types are not the same. And sample imbalance will cause the subsequent model to tend to predict the samples with a large quantity of a certain lining type, and the prediction performance for the samples with a small quantity of a certain lining type is poor. Therefore, it is necessary to perform resampling on the sub-dataset. The specific steps of resampling are as follows:

[0070] S2031. Downsample the sub-dataset. In the sub-dataset, for the samples with too many of a certain lining type, randomly and non-repeatedly select one of them each time until the specified number of sample quantities is selected;

[0071] S2032. Upsample the sub-dataset. In the sub-dataset, for the samples with too few of a certain lining type, generate a new specified number of sample quantities based on the Smote algorithm. The specific method of the Smote algorithm includes: randomly select a sample with too few of a certain lining type, calculate the k nearest neighbors of this sample, randomly select 1 sample from the k neighbors, and randomly select 1 point between this neighbor sample and the sample point, and use it as the newly generated sample. Repeat the above steps until the specified number of samples is generated.

[0072] S204. Construct the lining type sub-datasets. Repeat S203 for each sub-dataset corresponding to a surrounding rock grade until all sub-datasets corresponding to all surrounding rock grades are covered, that is, construct the lining type sub-datasets corresponding to all surrounding rock grades.

[0073] For example, according to the data in Table 2, the feature extraction and conversion results of the designed tunnel are shown in Table 3. Split the data in the dataset with the same surrounding rock grade after feature extraction into a sub-dataset. Then, Sub-dataset 1 is: {Sample 1, Sample 2}, and the corresponding surrounding rock grade is Grade V. Sub-dataset 2 is: {Sample 3, Sample 5, Sample 7, Sample 9, Sample 11, Sample 13}, and the corresponding surrounding rock grade is Grade IV. Sub-dataset 3 is: {Sample 4, Sample 6, Sample 8, Sample 10, Sample 12, Sample 14}, and the corresponding surrounding rock grade is Grade III. Sub-dataset 4 is: {Sample 15}, and the corresponding surrounding rock grade is Grade II.

[0074] Table 3 Feature extraction and conversion results of a certain tunnel

[0075]

[0076] Among all the collected tunnel datasets, the sub-dataset I corresponding to the surrounding rock grade of V contains 3,075 lining samples of type Vb, 290 lining samples of type Va, and 72 lining samples of open cut tunnel lining. Resampling is performed on it, and the corresponding sub-dataset I of lining types is obtained, containing 200 lining samples of type Vb, 200 lining samples of type Va, and 200 lining samples of open cut tunnel lining.

[0077] The sub-dataset II corresponding to the surrounding rock grade of IV contains 1,711 lining samples of type IVa and 1,717 lining samples of type IVb. Since the number of samples of type IVa and type IVb lining is not much different, no resampling is required, and this sub-dataset II is the corresponding sub-dataset II of lining types, containing 1,711 lining samples of type IVa and 1,717 lining samples of type IVb.

[0078] The sub-dataset III corresponding to the surrounding rock grade of III contains 1,418 lining samples of type IIIa and 831 lining samples of type IIIb. Resampling is performed on it, and the corresponding sub-dataset III of lining types is obtained, containing 831 lining samples of type IIIa and 831 lining samples of type IIIb.

[0079] The sub-dataset IV corresponding to the surrounding rock grade of II contains 558 lining samples of type IIa and 45 lining samples of type IIb. Resampling is performed on it, and the corresponding sub-dataset IV of lining types is obtained, containing 150 lining samples of type IIIa and 150 lining samples of type IIIb.

[0080] S300. According to the sub-dataset of lining types, construct a GWO-NGBoost lining type prediction model; in S300 of this embodiment, according to the sub-dataset of lining types, construct a GWO-NGBoost lining type prediction model, as Figure 2 , the specific method includes:

[0081] S301. Dataset division. In the same sub-dataset of lining types, divide the dataset into a training set and a test set, training set: test set = 80%: 20%. The training set is used to construct the model and find the optimal hyperparameters, and the test set is used to evaluate the model;

[0082] S302. Use the GWO algorithm to find the optimal hyperparameters of the NGBoost model; among them, in S302, use the GWO algorithm to find the optimal hyperparameters of the NGBoost model, and the specific method includes:

[0083] S3021. Set the hyperparameter range, and set the value ranges of the parameters n_estimators, learning_rate, minibatch_frac, and col_sample in the NGBoost model;

[0084] S3022. Initialize the positions of the wolf pack. Randomly select a number within the value range for each hyperparameter, and generate a list consisting of M sets of hyperparameters, which is the initialized position of the wolf pack;

[0085] S3023. Calculate the fitness values of the individual wolves in the wolf pack and save the best three wolves. Set the position parameters of each wolf as the hyperparameter values of the NGBoost model, and use the negative value of the average precision rate based on K-fold cross-validation as the fitness value of this wolf individual. The calculation formula for precision is shown in Formula 5. Precision represents the proportion of samples that are truly positive among the samples predicted as positive by the model;

[0086]

[0087] In the formula, Precision is the precision rate, TP is the sample predicted as positive and actually positive, and FP is the sample predicted as positive and actually negative; after obtaining the fitness values of the individual wolves in the wolf pack, select the three wolves with the smallest fitness values and save them as the α, β, and δ wolves;

[0088] S3024. Calculate the correlation coefficients. The three coefficients of the convergence factor a, parameter C, and parameter A are calculated according to Formulas 6 - 8:

[0089]

[0090] C i =2×r i (7)

[0091] A i =2×a×r i -a (8)

[0092] In the formula, t is the current iteration number, t o is the total number of iterations, i takes values of 1, 2, 3, and r i is a random number between 0 and 1;

[0093] S3025. Update the positions of the wolf pack. Update the positions of the wolf pack according to the α, β, and δ wolves, and its update formula is shown in Formulas 9 - 11:

[0094]

[0095] In the formula, X represents the position of the wolf pack, X α 、X β 、X δRepresent the positions of α, β, and δ wolves, and X(t + 1) represents the updated positions of the wolf pack;

[0096] S3206. Obtain the optimal hyperparameters, and repeat S3203 - S3205 until the number of iterations reaches the maximum number of iterations. Then, the position of the α wolf is the obtained optimal hyperparameter.

[0097] S303. Construct a lining type prediction model. For each lining type sub - dataset, based on the obtained optimal hyperparameters, establish an NGBoost lining type prediction model on the entire training set, and finally obtain multiple NGBoost lining type prediction models corresponding to multiple lining type sub - datasets.

[0098] According to the features and conversion results extracted in Table 3, based on the GWO - NGBoost model, find the optimal hyperparameters in the training sets of the lining type sub - datasets respectively, and establish corresponding prediction models. The prediction accuracy evaluations of the models on the training set and the test set are shown in Table 4. It can be seen that the GWO - NGBoost prediction models established on each lining type sub - dataset have an accuracy rate of over 80% in both the training set and the test set, which indicates that the established prediction models have good performance and can accurately realize the prediction of the lining type.

[0099] Table 4 Prediction result table of GWO - NGBoost model

[0100]

[0101] S400. For the tunnel to be predicted, determine the section division of the surrounding rock grade based on the geological exploration data; the section division of the tunnel surrounding rock grade determined based on the geological exploration data for the tunnel to be designed is shown in Table 2.

[0102] S500. For each section of the tunnel to be predicted, extract features and input them into the lining type prediction models corresponding to the surrounding rock grades respectively to realize the prediction of the lining type.

[0103] In S500 of this embodiment, for each section of the tunnel to be predicted, extract features and input them into the lining type prediction models corresponding to the surrounding rock grades respectively to realize the prediction of the lining type, as Figure 3 , and the specific method includes:

[0104] S501. Feature extraction and conversion. On the section of the surrounding rock grade, extract features such as buried depth, tunnel longitude and latitude, tunnel length, and surrounding rock section length, and perform conversions;

[0105] S502. Obtain the lining type of each section. For each section of the surrounding rock grade, find the lining type prediction model corresponding to the surrounding rock grade, input the features extracted in S501 into the lining type prediction model, and obtain the lining type of the corresponding section;

[0106] S503. Determine the lining type, detect each adjacent section in S502. If the lining types of adjacent sections are the same, they are merged into one section. The merged section and the corresponding lining type are the finally determined lining types.

[0107] For the tunnel to be designed in Table 2, extract the corresponding tunnel length, surrounding rock section length, buried depth, and characteristics after longitude and latitude conversion. Input the extracted characteristics into the lining type prediction model for the corresponding surrounding rock grade, and the lining type prediction results for each mileage can be obtained as shown in Table 5. Since there are no identical lining type prediction results for adjacent surrounding rock grade sections, there is no need to merge the sections, and this prediction result is the finally determined lining type.

[0108] Table 5 Characteristic values and prediction results of evenly spaced mileage of the tunnel to be designed

[0109]

[0110] This embodiment discloses an intelligent determination method for tunnel lining types, including: obtaining a dataset of designed tunnels; constructing a lining type sub-dataset according to the designed tunnel dataset; constructing a GWO-NGBoost lining type prediction model according to the lining type sub-dataset; for the tunnel to be predicted, determining the section division of the surrounding rock grade based on geological exploration data; for each section of the tunnel to be predicted, extracting features and inputting them into the lining type prediction model for the corresponding surrounding rock grade to achieve the prediction of the lining type. The present invention realizes the prediction of the lining type by establishing a GWO-NGBoost model in each tunnel lining type sub-dataset, and the whole process is standard and unified, and can be completed efficiently and automatically. The present invention constructs corresponding lining type prediction models in each tunnel lining type sub-dataset, and finally constructs multiple lining type prediction models, which can effectively improve the accuracy of lining type prediction.

[0111] Embodiment 2

[0112] Based on the same inventive concept, the present disclosure embodiment also provides an electronic device. Fig. 4 is a schematic structural diagram of an electronic device according to the present disclosure embodiment. As Figure 4 shown, the present disclosure embodiment provides an electronic device including: one or more processors 101, a memory 102, and one or more I / O interfaces 103. One or more programs are stored on the memory 102. When the one or more programs are executed by the one or more processors, the one or more processors implement any optimization method in the above embodiments; one or more I / O interfaces 103 are connected between the processor and the memory and are configured to implement information interaction between the processor and the memory.

[0113] Among them, the processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU), etc.; the memory 102 is a device with data storage capabilities, including but not limited to a random access memory (RAM, more specifically such as SDRAM, DDR, etc.), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize the information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus), etc.

[0114] In some embodiments, the processor 101, the memory 102, and the I / O interface 103 are interconnected through a bus 104, and then connected to other components of the computing device.

[0115] In some embodiments, the one or more processors 101 include a field programmable gate array.

[0116] According to an embodiment of the present disclosure, there is also provided a computer-readable medium. A computer program is stored on the computer-readable medium, wherein when the program is executed by a processor, the steps in any of the optimization methods in the above embodiments are implemented.

[0117] It should be understood that the specific order or hierarchy of the steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of the steps in the process can be rearranged without departing from the protection scope of the present disclosure. The appended method claims present the elements of the various steps in an exemplary order and are not intended to be limited to the specific order or hierarchy.

[0118] In the above detailed description, various features are combined in a single embodiment to simplify the present disclosure. This method of disclosure should not be construed as reflecting an intention that the embodiments of the claimed subject matter require more features than are clearly recited in each claim. On the contrary, as reflected by the appended claims, the present invention lies in a state less than all the features of the disclosed single embodiment. Therefore, the appended claims are hereby clearly incorporated into the detailed description, where each claim stands alone as a separate preferred embodiment of the present invention.

[0119] Those skilled in the art should also understand that all the illustrative logical blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments herein can be implemented as electronic hardware, computer software, or a combination thereof. To clearly illustrate the interchangeability between hardware and software, the above various illustrative components, blocks, modules, circuits, and steps have been generally described in terms of their functions. Whether such a function is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Skilled technicians can implement the described functions in a flexible manner for each specific application. However, such implementation decisions should not be construed as departing from the scope of protection of this disclosure.

[0120] The steps of the methods or algorithms described in conjunction with the embodiments herein can be directly embodied as hardware, software modules executed by a processor, or a combination thereof. The software modules can be located in a RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium well-known in the art. An exemplary storage medium is connected to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and the storage medium can be located in an ASIC. The ASIC can be located in a user terminal. Of course, the processor and the storage medium can also exist as discrete components in the user terminal.

[0121] For software implementation, the techniques described in this application can be implemented using modules (e.g., procedures, functions, etc.) that execute the functions described in this application. These software codes can be stored in a memory unit and executed by a processor. The memory unit can be implemented inside the processor or outside the processor. In the latter case, it is communicatively coupled to the processor via various means, which are well-known in the art.

[0122] The above description includes examples of one or more embodiments. Of course, it is impossible to describe all possible combinations of components or methods for the purpose of describing the above embodiments. However, those of ordinary skill in the art should recognize that the various embodiments can be further combined and arranged. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of protection of the appended claims. In addition, with respect to the term "comprising" used in the specification or claims, this term is covered in a manner similar to the term "including," as interpreted when "including" is used as a transitional word in the claims. In addition, any term "or" used in the claims or the specification is intended to mean "non-exclusive or."

Claims

1. An intelligent method for determining the type of tunnel lining, characterized in that: include: S100. Obtaining a data set of designed tunnels; S200. Constructing a lining type sub-dataset according to the designed tunnel data set; S300. Constructing a GWO-NGBoost lining type prediction model according to the lining type sub-dataset; S400. For the tunnel to be predicted, determine the segment division of the surrounding rock grade based on the geological survey data; S500. For each section of the tunnel to be predicted, extract features respectively and input them into the lining type prediction model corresponding to the surrounding rock grade to realize the prediction of the lining type.

2. The intelligent method for determining the tunnel lining type according to claim 1, characterized in that: In S100, the designed tunnel data set includes at least the longitude and latitude of the tunnel, the length of the tunnel, the starting and ending mileages of the surrounding rock section, the burial depths of multiple mileages in the tunnel, and the corresponding surrounding rock grades and lining types.

3. The intelligent method for determining the type of tunnel lining according to claim 1, characterized in that: In S200, a lining type sub-dataset is constructed based on the designed tunnel data set. The specific method includes: extracting and converting features of the acquired data set, dividing data with the same surrounding rock grade into a sub-dataset, resampling the sub-dataset to form a lining type sub-dataset, and finally constructing multiple lining type sub-datasets corresponding to multiple surrounding rock grades.

4. A method for intelligently determining the type of tunnel lining according to claim 3, characterized in that: The acquired data set is subjected to feature extraction and conversion. The specific method includes: extracting tunnel longitude and latitude, tunnel length, surrounding rock section length, and buried depth features from the data set, and performing plane coordinate conversion on the tunnel longitude and latitude features, converting the longitude and latitude geographic coordinates into plane coordinates. The coordinate conversion formula is as shown in formulas (1) to (4): In the formula, x log 、x lat are the longitude and latitude values ​​of the tunnel, x log_sin 、x log_cos is the coordinate value after longitude conversion, x lat_sin 、x lat_cos The coordinate value after latitude conversion.

5. The intelligent method for determining the type of tunnel lining according to claim 3, characterized in that: The sub-datasets are resampled to form lining type sub-datasets. The specific methods include: Down-sampling is performed on the sub-dataset. For samples with too many lining types in the sub-dataset, one of the samples is randomly and non-repeatedly selected each time until a specified number of samples is selected; up-sampling is performed on the sub-dataset. For samples with too few lining types in the sub-dataset, a new specified number of samples is generated based on the Smote algorithm. The specific method of the Smote algorithm includes: randomly selecting a sample with too few lining types, calculating the k neighbors closest to the sample, randomly selecting a sample from the k neighbors, and randomly selecting a point between the neighbor sample and the sample point as the newly generated sample, and repeating the steps until the specified number of samples is generated.

6. The intelligent method for determining the type of tunnel lining according to claim 3, characterized in that: Finally, multiple lining type sub-datasets corresponding to multiple surrounding rock grades are constructed. The specific method includes: repeating the resampling step for each sub-dataset corresponding to each surrounding rock grade until all sub-datasets corresponding to all surrounding rock grades are covered, that is, constructing the lining type sub-datasets corresponding to all surrounding rock grades.

7. The intelligent method for determining the type of tunnel lining according to claim 1, characterized in that: In S300, a GWO-NGBoost lining type prediction model is constructed according to the lining type sub-dataset, and the specific method includes: S301. Dataset division: In the same lining type sub-dataset, the data set is divided into a training set and a test set. The training set is used to build the model and find the optimal hyperparameters, and the test set is used to evaluate the model. S302. Use the GWO algorithm to find the optimal hyperparameters of the NGBoost model; S303. Construct a lining type prediction model. For each lining type sub-dataset, establish an NGBoost lining type prediction model on the entire training set based on the obtained optimal hyperparameters, and finally obtain multiple NGBoost lining type prediction models corresponding to multiple lining type sub-datasets.

8. A method for intelligently determining the type of tunnel lining according to claim 7, characterized in that: In S302, the GWO algorithm is used to find the optimal hyperparameters of the NGBoost model. The specific method includes: S3021. Set the hyperparameter range, set the value range of the parameters n_estimators, learning_rate, minibatch_frac, col_sample in the NGBoost model; S3022. Initialize the position of the wolf pack, randomly select a number within the value range for each hyperparameter, and generate a list consisting of M groups of hyperparameters as the initialized position of the wolf pack; S3023. Calculate the fitness values ​​of individual wolves and save the best three wolves. Set the position parameters of each wolf as the hyperparameter value of the NGBoost model, and use the negative value of the average precision rate based on K-fold cross validation as the fitness value of the individual wolf. The precision rate is calculated as shown in Formula 5. The precision rate represents the proportion of samples predicted by the model to be positive that are actually positive. Where Precision is the accuracy rate, TP is the sample predicted to be positive and actually positive, and FP is the sample predicted to be positive and actually negative. After obtaining the fitness values ​​of the wolf pack individuals, the three wolves with the smallest fitness are selected and saved as α, β, and δ wolves. S3024. Calculate the correlation coefficient. The three coefficients of convergence factor a, parameter C, and parameter A are calculated according to formulas 6 to 8: C i ＝2×r i (7) A i =2×a×r i -a (8) In the formula, t is the current iteration number, t o is the total number of iterations, i takes values ​​of 1, 2, 3, r i A random number between 0 and 1; S3025. Update the position of the wolf pack. Update the position of the wolf pack based on α, β, and δ wolves. The update formula is shown in formulas 9 to 11: In the formula, X represents the location of the wolf pack, X α , X β , X δ represents the positions of α, β, and δ wolves, and X(t+1) represents the updated position of the wolf pack; S3206. Obtain the optimal hyperparameters, repeat S3203-S3205 until the number of iterations reaches the maximum number of iterations, then the position of α wolf is the optimal hyperparameter obtained.

9. The intelligent method for determining the type of tunnel lining according to claim 1, characterized in that: In S500, for each section of the tunnel to be predicted, features are extracted and input into the lining type prediction model corresponding to the surrounding rock grade to achieve the prediction of the lining type. The specific method includes: S501. Extract features and convert them. In the surrounding rock grade section, extract the buried depth, tunnel longitude and latitude, tunnel length, and surrounding rock section length features and convert them; S502. Obtain the lining type of each section. For each surrounding rock grade section, find the lining type prediction model corresponding to the surrounding rock grade, input the features extracted in S501 into the lining type prediction model, and obtain the lining type of the corresponding section; S503. Determine the lining type, detect each adjacent section in S502, and if the lining types of adjacent sections are the same, merge them into one section. The merged section and the corresponding lining type are the finally determined lining type.

10. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the intelligent determination method of the tunnel lining type in any one of claims 1-9.

Images