Method, device and equipment for predicting tunneling parameters of TBM (Tunnel Boring Machine) and storage medium

Abstract

The embodiment of the invention provides a TBM tunneling parameter prediction method. The TBM tunneling parameter prediction method comprises the steps that TBM tunneling parameter data are collected and preprocessed; analyzing the correlation of the data, screening out an input characteristic quantity having high correlation with the output characteristic quantity, and combining the input characteristic quantity with the output characteristic quantity to form a TBM tunneling data set; dividing the TBM tunneling data set into a TBM tunneling training set and a TBM tunneling test set; constructing a TBM tunneling parameter prediction model, wherein the model comprises two layers; inputting the TBM tunneling training set and the TBM tunneling test set into the first-layer model for training and prediction, and adjusting parameters of the first-layer prediction model by using a particle swarm optimization algorithm to obtain a meta-training set and a meta-test set; respectively inputting the meta-training set and the meta-test set into the second-layer model for training and prediction to obtain a prediction result; and evaluating the performance of the TBM tunneling parameter model by using the evaluation index. According to the method, correlation analysis and stacking generalization are combined, and the prediction accuracy is improved.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the technical field of tunnel construction and related technical fields, and in particular, to a method, apparatus, device and storage medium suitable for predicting TBM excavation parameters. Background Art

[0002] Compared with traditional tunnel excavation methods, the full-section hard rock tunnel boring machine (TBM) has demonstrated multiple advantages such as fast excavation speed, strong environmental adaptability, excellent engineering quality, timely tunnel lining support, high level of mechanization, and economy and environmental protection.

[0003] However, in the existing technology, in actual tunneling operations, the adjustment of TBM excavation parameters mainly depends on the personal experience of the TBM driver. Since the actual surrounding rock conditions in front of the face are difficult to predict during tunneling, improper adjustment of excavation parameters often leads to problems such as TBM jamming and surrounding rock instability and collapse, which in turn poses a serious threat to the safety, progress and cost of construction.

[0004] Therefore, in order to further improve the accuracy and reliability of TBM excavation parameter prediction, a more advanced performance prediction method is urgently needed to guide parameter prediction and adjustment during TBM construction. Summary of the invention

[0005] The embodiments described herein provide a method, apparatus, device and storage medium for predicting TBM excavation parameters, which solve the problems existing in the prior art.

[0006] According to a first aspect of the present disclosure, a method for predicting TBM excavation parameters is provided, comprising:

[0007] Collecting and preprocessing tunnel boring machine (TBM) excavation parameter data to obtain preprocessed TBM excavation parameter data, and dividing the TBM excavation parameter data into input feature quantities and output feature quantities;

[0008] Performing correlation analysis on the TBM excavation parameter data using the Spearman correlation coefficient, screening out the input feature quantity having a high correlation with the output feature quantity, and combining the input feature quantity with the output feature quantity to form a TBM excavation data set;

[0009] By using cross-validation technology, the TBM excavation data set is divided into a TBM excavation training set and a TBM excavation test set;

[0010] Based on the stacking generalization principle, a TBM excavation parameter prediction model is constructed, wherein the TBM excavation parameter prediction model includes two layers: a first layer prediction model and a second layer prediction model;

[0011] Input the TBM excavation training set into the first-layer prediction model for training, use the TBM excavation test set for prediction, and use the particle swarm optimization algorithm to adjust the parameters of the first-layer prediction model to determine the optimal parameter combination, thereby obtaining a meta-training set and a meta-test set;

[0012] Inputting the meta-training set and the meta-test set into the second-layer prediction model for training and prediction respectively to obtain prediction results;

[0013] The performance of the TBM excavation parameter model is evaluated by comparing the error between the prediction result and the output characteristic quantity.

[0014] In some embodiments of the present disclosure, the TBM excavation parameter data set includes but is not limited to the following feature quantities:

[0015] Cutter disc speed, cutter disc set speed, tool torque, propulsion speed, thrust, clamping pressure, propulsion pump pressure, propulsion cylinder rod chamber pressure, propulsion cylinder rodless chamber pressure, clamping pump pressure, belt pump head pressure, belt motor pressure, cutter disc water spray pressure, cutter disc water spray flow, propulsion stroke, continuous belt operation, penetration and belt speed.

[0016] In some embodiments of the present disclosure, the step of preprocessing tunnel boring machine (TBM) excavation parameter data to obtain preprocessed TBM excavation parameter data includes:

[0017] The TBM excavation parameter data is subjected to data cleaning, missing value processing, outlier detection, normalization processing and standardization processing to obtain the pre-processed TBM excavation parameter data.

[0018] In some embodiments of the present disclosure, the step of performing correlation analysis on the input feature quantity using the Spearman correlation coefficient, screening out the input feature quantity having a high correlation with the output feature quantity, and combining it with the output feature quantity to form a TBM excavation data set includes:

[0019] At least two attributes are set in the pre-processed TBM excavation parameter data, and corresponding values ​​are assigned to each of them;

[0020] Sort the values ​​of the attributes in ascending order, and replace the original value of each value with its rank in the sorted sequence;

[0021] Based on the ranking information after ranking, calculate the level difference between the two attribute rankings at each moment;

[0022] Based on the level difference and the number of feature quantities, calculating the Spearman correlation coefficient between each of the input feature quantities and the output feature quantity;

[0023] The strength of the correlation between the two variables is determined based on the absolute value of the calculated Spearman correlation coefficient. If the absolute value of the Spearman correlation coefficient is high, it indicates that the correlation between the two variables is strong. The feature quantity with a strong correlation with the output feature quantity is screened out, and the feature quantity with a strong correlation with the output feature quantity is combined with the output feature quantity to form a TBM excavation data set.

[0024] In some embodiments of the present disclosure, the step of dividing the TBM excavation data set into a TBM excavation training set and a TBM excavation test set by using a cross-validation technique includes:

[0025] In each cross-validation, the TBM excavation data set is randomly divided into five parts, four of which are selected as the TBM excavation training set, and the other part is selected as the TBM excavation test set;

[0026] The above steps are repeated at least five times, and a different TBM excavation data set is selected each time.

[0027] In some embodiments of the present disclosure, the step of constructing a TBM excavation parameter prediction model based on the stacking generalization principle includes:

[0028] Using the Stacking fusion algorithm and the layered prediction strategy, a two-layer prediction model is constructed;

[0029] Among them, the first-layer prediction model is composed of multiple base learners, and the second-layer prediction model is composed of a meta-learner, the base learners include but are not limited to GRU, SVM and XGBoot, and the meta-learner is but not limited to LSTM neural network.

[0030] In some embodiments of the present disclosure, the evaluation indicators include: average relative error, relative mean square error and determination coefficient R 2 ;

[0031] In some embodiments of the present disclosure, after the step of evaluating the performance of the TBM excavation parameter model, the following steps are included:

[0032] According to the results obtained after the evaluation, the TBM excavation parameter prediction model is adjusted and optimized.

[0033] According to a second aspect of the present disclosure, a device for predicting TBM excavation parameters is provided, comprising:

[0034] A preprocessing module is used to collect and preprocess the tunnel boring machine (TBM) excavation parameter data, obtain the preprocessed TBM excavation parameter data, and divide the TBM excavation parameter data into an input feature quantity and an output feature quantity;

[0035] An analysis module is used to perform correlation analysis on the TBM excavation parameter data using the Spearman correlation coefficient, screen out input feature quantities that have a high correlation with the output feature quantities, and combine the input feature quantities with the output feature quantities to form a TBM excavation data set;

[0036] A partitioning module is used to partition the TBM excavation data set into a TBM excavation training set and a TBM excavation test set by using a cross-validation technology;

[0037] A construction module is used to construct a TBM excavation parameter prediction model based on the stacking generalization principle, wherein the TBM excavation parameter prediction model includes two layers: a first layer prediction model and a second layer prediction model;

[0038] A first processing module is used to input the TBM excavation training set into the first-layer prediction model for training, use the TBM excavation test set for prediction, and use a particle swarm optimization algorithm to adjust the parameters of the first-layer prediction model to determine the optimal parameter combination, thereby obtaining a meta-training set and a meta-test set;

[0039] A second processing module, used for inputting the meta-training set and the meta-test set into the second-layer prediction model for training and prediction, respectively, to obtain a prediction result;

[0040] The evaluation module is used to evaluate the performance of the TBM excavation parameter model using evaluation indicators.

[0041] In some embodiments of the present disclosure, the TBM excavation parameter data set includes but is not limited to the following feature quantities:

[0042] Cutter disc speed, cutter disc set speed, tool torque, propulsion speed, thrust, clamping pressure, propulsion pump pressure, propulsion cylinder rod chamber pressure, propulsion cylinder rodless chamber pressure, clamping pump pressure, belt pump head pressure, belt motor pressure, cutter disc water spray pressure, cutter disc water spray flow, propulsion stroke, continuous belt operation, penetration and belt speed.

[0043] In some embodiments of the present disclosure, the preprocessing module is specifically used to perform data cleaning, missing value processing, outlier detection, normalization and standardization on the TBM excavation parameter data to obtain the final preprocessed TBM excavation parameter data.

[0044] In some embodiments of the present disclosure, the analysis module is specifically used to set at least two attributes in the pre-processed TBM excavation parameter data and assign corresponding values ​​to each of them; sort the values ​​of the attributes in ascending order, and replace the original numerical value of each value with its rank in the sorted sequence; calculate the rank difference between the two attribute rankings at each moment based on the ranking information after ranking; calculate the Spearman correlation coefficient between each input feature quantity and the output feature quantity based on the rank difference and the number of feature quantities; judge the correlation strength between the two variables according to the absolute value of the calculated Spearman correlation coefficient, if the absolute value of the Spearman correlation coefficient is high, it indicates that the correlation between the two variables is strong, screen out the feature quantity with a strong correlation with the output feature quantity, and combine the feature quantity with a strong correlation with the output feature quantity with the output feature quantity to form a TBM excavation data set.

[0045] In some embodiments of the present disclosure, the partitioning module is specifically used to randomly divide the TBM excavation data set into five parts in each cross-validation, select four parts as the TBM excavation training sets, and the other part as the TBM excavation test set; repeat the above steps at least five times, and select a different TBM excavation data set each time.

[0046] In some embodiments of the present disclosure, the construction module is specifically used to construct a two-layer prediction model using a stacking fusion algorithm and a layered prediction strategy;

[0047] Among them, the first-layer prediction model is composed of multiple base learners, and the second-layer prediction model is composed of a meta-learner, the base learners include but are not limited to GRU, SVM and XGBoot, and the meta-learner is but not limited to LSTM neural network.

[0048] In some embodiments of the present disclosure, the evaluation indicators include: average relative error, relative mean square error and determination coefficient R 2 .

[0049] In some embodiments of the present disclosure, it further includes:

[0050] The optimization module is used to adjust and optimize the TBM excavation parameter prediction model according to the results obtained after the evaluation.

[0051] According to a third aspect of the present disclosure, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method in any one of the above embodiments are implemented.

[0052] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method in any one of the above embodiments are implemented.

[0053] The method for predicting TBM excavation parameters provided by the embodiment of the present disclosure includes: collecting and preprocessing TBM excavation parameter data; analyzing the correlation of data, screening out input feature quantities with high correlation with output feature quantities, and combining them with output feature quantities to form a TBM excavation data set; dividing the TBM excavation data set into a TBM excavation training set and a TBM excavation test set; constructing a TBM excavation parameter prediction model, which contains two layers; inputting the TBM excavation training set and the TBM excavation test set into the first layer model for training and prediction, and adjusting the parameters of the first layer prediction model by using a particle swarm optimization algorithm to obtain a meta-training set and a meta-test set; inputting the meta-training set and the meta-test set into the second layer model for training and prediction, respectively, to obtain prediction results; and using evaluation indicators to evaluate the performance of the TBM excavation parameter model. The embodiment of the present disclosure uses the Spearman coefficient for correlation analysis, uses variables with a strong nonlinear relationship with the prediction results as model input features, screens the model input features, and combines five-fold cross validation and PSO algorithm to optimize parameters to select optimal parameters, and combines the Stacking ensemble learning method to complete the training and prediction on the original data set. Finally, the training set and test set of the meta-learner are obtained. The meta-learner completes training on the training set of the meta-learner and makes predictions on the test set of the meta-learner. The PSO algorithm can effectively search the optimal parameter space by continuously adjusting and optimizing the parameters of each sub-model, so that the model can better fit the training data, thereby significantly improving the prediction accuracy of TBM performance. It effectively avoids model overfitting, improves the generalization ability of the model, and makes full use of the advantages of different learners. During the TBM construction process, various emergencies and abnormal data may be encountered. The new integrated learning model based on the PSO algorithm can better cope with these situations. It can quickly adapt to changes in data, adjust model parameters, maintain stable prediction performance, and provide reliable support for construction decisions.

[0054] The above description is only an overview of the technical solution of the embodiment of the present application. In order to more clearly understand the technical means of the embodiment of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments will be briefly described below. It should be noted that the drawings described below only relate to some embodiments of the present disclosure, but are not intended to limit the present disclosure, wherein:

[0056] Figure 1 It is a flow chart of a method for predicting TBM excavation parameters provided by an embodiment of the present disclosure;

[0057] Figure 2 It is a structural schematic diagram of a device for predicting TBM excavation parameters provided by an embodiment of the present disclosure;

[0058] Figure 3 It is a structural diagram of a computer device provided in an embodiment of the present disclosure.

[0059] In the drawings, reference numerals with the same last two digits correspond to the same elements. It should be noted that the elements in the drawings are schematic and not drawn to scale. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work also fall within the scope of protection of the present disclosure.

[0061] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase "embodiments" in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0062] The term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists, A and B exist at the same time, and B exists. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0063] Furthermore, in all embodiments of the present disclosure, terms such as “first” and “second” are used only to distinguish one component (or a part of a component) from another component (or another part of a component).

[0064] In the description of the present application, unless otherwise specified, "plurality" means more than two (including two), and similarly, "multiple groups" means more than two groups (including two).

[0065] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0066] The method for predicting TBM excavation parameters provided in the embodiment of the present disclosure is applied to a server.

[0067] Glossary:

[0068] Tunnel Boring Machine (TBM) refers to a large-scale underground construction equipment that integrates mechanical, electrical and hydraulic functions with a high degree of technology intensity. It is mainly used for tunnel excavation and construction of railways, highways, water conservancy and hydropower diversion tunnels, subways and underground engineering projects in rock geological structures.

[0069] Stacking, also known as stacked generalization or stacked integration, is an ensemble learning method. A model is trained to combine the prediction results of multiple other models to obtain a final prediction output. In this process, multiple different base models are first trained, and then a second-layer model is trained using the outputs of the previous models as input.

[0070] Particle Swarm Optimization (PSO), also known as particle swarm algorithm, is an evolutionary computing technology that comes from the simulation of a simplified social model.

[0071] Based on the problems existing in the existing technology, Figure 1 is a flow chart of a method for predicting TBM excavation parameters provided by an embodiment of the present disclosure, such as Figure 1 As shown in FIG. 1 , the specific process of the method for predicting TBM excavation parameters includes:

[0072] S110, collecting and preprocessing tunnel boring machine (TBM) excavation parameter data, obtaining preprocessed TBM excavation parameter data, and dividing the TBM excavation parameter data into input feature quantities and output feature quantities.

[0073] Specifically, the TBM excavation parameter data includes but is not limited to the following characteristic quantities:

[0074] Cutter disc speed, cutter disc set speed, tool torque, propulsion speed, thrust, clamping pressure, propulsion pump pressure, propulsion cylinder rod chamber pressure, propulsion cylinder rodless chamber pressure, clamping pump pressure, belt pump head pressure, belt motor pressure, cutter disc water spray pressure, cutter disc water spray flow, propulsion stroke, continuous belt operation, penetration and belt speed, etc.

[0075] In a specific implementation process, the step of preprocessing TBM excavation characteristic parameter data includes:

[0076] First, the TBM excavation parameter data is cleaned, missing values ​​are processed, and outliers are detected to ensure high standards of data quality;

[0077] Secondly, the cleaned data is normalized to eliminate the dimensional differences between different features, thereby improving the efficiency and accuracy of model training;

[0078] Finally, the normalized data are standardized to further adjust the data distribution and obtain the final preprocessed TBM excavation parameter data.

[0079] After the above steps, the accuracy and consistency of the data are ensured.

[0080] In addition, before implementing the above steps, it also includes:

[0081] S100, monitoring items during TBM excavation are monitored to obtain TBM excavation parameter data.

[0082] In the specific implementation process, during the TBM excavation operation, a series of key monitoring items are continuously and accurately monitored so as to obtain and record various excavation parameters during the TBM excavation process in real time.

[0083] S120, performing correlation analysis on the TBM excavation parameter data using the Spearman correlation coefficient, screening out the input feature quantities having a high correlation with the output feature quantities, and combining the input feature quantities with the output feature quantities to form a TBM excavation data set.

[0084] In the specific implementation process, in order to deeply understand the correlation between each feature parameter, the Spearman correlation coefficient is used to perform correlation analysis on the input feature quantity. The Spearman correlation coefficient is a non-parametric statistical method specifically used to evaluate the correlation between two ordered variables. The specific steps are as follows:

[0085] 1. Data preparation:

[0086] At least two attributes (features) are set in the pre-processed TBM excavation parameter data, and corresponding values ​​are assigned to each of them, for example: x1, x2, ..., xn and y1, y2, ..., yn, respectively, where n represents the number of different moments.

[0087] 2. Rank conversion:

[0088] Sort the values ​​of these two attributes by size, and use the ranking of each value in the sorted sequence to replace its original value. The purpose of this step is to eliminate the dimensional difference of the original value, making the subsequent analysis more fair and accurate. Suppose the ranking of x is The ranking of y is

[0089] 3. Grade difference calculation:

[0090] Assume that the ranking of xi is y i The ranking is Based on the ranking information after ranking, calculate the level difference d between the two attribute rankings at each moment (or data point) i ,Right now and The level difference reflects the relative position difference of two attributes at the same time. The level difference is formula (1):

[0091]

[0092] 4. Using the rank difference and the number of features, calculate the Spearman correlation coefficient between each feature and the output feature. The Spearman correlation coefficient is a statistical indicator that measures the rank correlation between two variables. It is not limited by the data distribution form, so it has significant advantages when analyzing data with non-normal distribution or nonlinear relationships.

[0093] The Spearman correlation coefficient is calculated using formula (2). Formula (2) is as follows:

[0094]

[0095] Among them, m represents the number of features, i is a positive integer, and d i is the rank difference at each moment, and ρ is the Spearman correlation coefficient. i To calculate the correlation between two variables.

[0096] 5. The strength of the correlation between the two variables is determined based on the absolute value of the calculated Spearman correlation coefficient;

[0097] Specifically, if the Spearman correlation coefficient is closer to 1 or -1, it means that the positive or negative correlation between the two variables is stronger.

[0098] When the correlation coefficient is 0, it indicates that there is no linear correlation between the two variables. For example, when the absolute value of the coefficient is greater than 0.4, there is a strong correlation between the two features.

[0099] The number 6 indicates a specific parameter.

[0100] If the absolute value of the Spearman correlation coefficient is high, it can be considered that the correlation between the two variables is strong; conversely, if the absolute value is low, it indicates that the correlation between them is weak. This judgment basis provides intuitive and quantitative information about the association between two attributes.

[0101] 6. Feature screening:

[0102] In order to reduce the number of input features, reduce the complexity of the model and avoid overfitting, the Spearman correlation coefficient between each feature and the output feature is calculated.

[0103] According to the size of the Spearman correlation coefficient, a feature quantity with a strong correlation with the output feature quantity is selected, and the feature quantity with a strong correlation with the output feature quantity is combined with the output feature quantity to form a TBM excavation data set;

[0104] Specifically, feature quantities with smaller correlation coefficients are regarded as weakly correlated features, which have less impact on the prediction results; conversely, feature quantities with larger correlation coefficients are regarded as strongly correlated features, which have a greater impact on the output feature quantities; through screening, input feature quantities with higher correlation with the output feature quantities are retained, and they are combined with the output feature quantities to form a TBM excavation data set as the input of subsequent models.

[0105] Through the above steps, the most influential features can be screened out as input features of subsequent models, which helps to reduce the complexity of the model and improve the prediction efficiency and accuracy of the model.

[0106] S130, dividing the TBM excavation data set into a TBM excavation training set and a TBM excavation test set by using a cross-validation technology;

[0107] In the specific implementation process, in order to make full use of the data set and evaluate the generalization ability of the model, a cross-validation method is used in each cross-validation. For example, a five-fold cross-validation method is used, which randomly divides the TBM data set into five roughly equal parts, and then selects four of them as TBM excavation training sets in turn, and the remaining part is used as the TBM excavation test set. That is to say, the TBM excavation data set is reasonably divided according to the ratio of 80% training set and 20% test set. This process is repeated at least five times, and a different part is selected as the test set each time to ensure that each sample in the data set has the opportunity to be used as a test. For another example, a four-fold cross-validation method is used, which randomly divides the TBM excavation data set into four roughly equal parts, and then selects three of them as TBM excavation training sets in turn, and the remaining part is used as the TBM excavation test set. That is to say, the data set is reasonably divided according to the ratio of 75% training set and 25% test set. This process is repeated at least four times, and a different part is selected as the test set each time to ensure that each sample in the data set has the opportunity to be used as a test.

[0108] S140, constructing a TBM excavation parameter prediction model based on the stacking generalization principle, wherein the TBM excavation parameter prediction model includes two layers: a first layer prediction model and a second layer prediction model;

[0109] In the specific implementation process, the TBM excavation parameter prediction model is a PSO-Stacking model. The PSO-Stacking model is an advanced integrated learning method that combines the particle swarm optimization (PSO) algorithm and the Stacking generalization principle, and aims to improve the prediction performance of the overall model by optimizing the parameter configuration of the base learner. In the process of building the prediction model, the Stacking fusion algorithm is used and a hierarchical prediction strategy is adopted, in which the first-layer prediction model is composed of multiple base learners, and the second-layer prediction model is a meta-learner. This hierarchical structure aims to improve the prediction performance of the overall model by combining the advantages of different learners.

[0110] Specifically, GRU (Gated Recurrent Unit), SVM (Support Vector Machine) and XGBoot are selected as base learners;

[0111] GRU (Gated Recurrent Unit): GRU is an improved recurrent neural network (RNN) structure that captures long-term dependencies in sequence data by introducing update gates and reset gates. In the PSO-Stacking model, GRU can be used as one of the base learners to process and analyze data sets with time series characteristics.

[0112] SVM (Support Vector Machine): SVM is a powerful classification and regression tool that separates data points of different categories by finding an optimal hyperplane. In the PSO-Stacking model, SVM can be used as one of the base learners to handle classification or regression tasks, especially when the data set has nonlinear relationships.

[0113] XGBoost (Extreme Gradient Boosting): XGBoost is an efficient gradient boosting algorithm that improves prediction performance by building multiple weak classifiers (such as decision trees) and combining them into a strong classifier. In the PSO-Stacking model, XGBoost can be used as one of the base learners to handle complex classification or regression tasks.

[0114] The LSTM neural network (Long Short-Term Memory Network) is selected as the meta-learner, which generates the final prediction results based on the prediction results of the base learner.

[0115] LSTM is a special RNN structure that captures long-term dependencies in sequence data by introducing mechanisms such as memory units and forget gates. In the PSO-Stacking model, the LSTM neural network can be used as one of the meta-learners, which is used to process and analyze prediction tasks with time series characteristics.

[0116] By combining different types of base learners and meta learners, the PSO-Stacking model can fully utilize the advantages of each and search and select the optimal parameter configuration through the particle swarm optimization (PSO) algorithm during the training process. This layered prediction and parameter optimization strategy aims to improve the predictive performance of the model, enabling it to more accurately process and analyze complex data sets.

[0117] S150. Determine optimal parameters using a particle swarm optimization algorithm, input the TBM excavation training set into the first-layer prediction model for training until completion, use the trained first-layer prediction model to predict the TBM excavation test set, and use the particle swarm optimization algorithm to adjust the parameters of the first-layer prediction model to determine the optimal parameter combination, thereby obtaining a meta-training set and a meta-test set.

[0118] In the specific implementation process, after data preprocessing and correlation analysis, a screened and optimized TBM excavation training set and TBM excavation test set were obtained, which were used as input for the training of the TBM excavation parameter prediction model (PSO-Stacking model).

[0119] At the same time, in order to find the optimal parameter combination and improve model performance, the parameter optimization theory of the particle swarm optimization (PSO) algorithm is used to adjust the parameters of the first-layer prediction model to determine the optimal parameter combination. The PSO algorithm is an optimization technology based on swarm intelligence, which simulates the movement and interaction of particles in the search space to find the global optimal solution. In this process, each particle represents a set of possible parameter configurations, and its position and velocity are updated by evaluating its corresponding model performance.

[0120] The TBM excavation training set is input into the first-layer prediction model of the PSO-Stacking model, i.e., the base learner, such as the base learners with excellent performance such as GRU, SVM and XGBoost. These base learners are trained on the TBM excavation training set until the model training is completed, and the trained first-layer prediction model is used to predict the TBM excavation test set, and the parameters of the first-layer prediction model are adjusted using the particle swarm optimization algorithm to determine the optimal parameter combination, and a meta-training set and a meta-test set are obtained, which are used as the input of the meta-learner.

[0121] During TBM construction, various emergencies and abnormal data may be encountered. The integrated learning model of the PSO algorithm, namely the PSO-Stacking model, can better cope with these situations. It can quickly adapt to data changes, adjust model parameters, and maintain stable prediction performance.

[0122] S160, inputting the meta-training set and the meta-test set into the second-layer prediction model for training and prediction, respectively, to obtain a prediction result;

[0123] Specifically, the meta-training set is input into the second-layer prediction model for training until completion, and the trained second-layer prediction model is used to predict the meta-test set to obtain the prediction result of the TBM excavation parameter prediction model;

[0124] In a specific implementation, the meta-training set is input into the second-layer prediction model, i.e., the meta-learner, LSTM (Long Short-Term Memory Network). As an advanced recurrent neural network, LSTM has a powerful ability to process sequence data and is very suitable for capturing long-term dependencies in time series data. During the model training phase, the LSTM meta-learner will make full use of the data in the meta-training set, and continuously adjust and optimize its internal parameters to learn how to extract useful information from the output of the base learner and generate accurate prediction results. The training process will continue until the preset training stop condition is reached, such as reaching a specified number of iterations, the model performance is no longer significantly improved, or the error on the validation set begins to increase. In the end, a trained LSTM meta-learner model will be obtained.

[0125] After the model training phase, the meta-test set is input into the trained second-layer prediction model, namely the LSTM meta-learner model. This model, after the previous data preprocessing, feature extraction and in-depth training of the LSTM meta-learner, already has strong prediction capabilities. When the data in the meta-test set flows through the model, the model will use its internally learned knowledge and rules to carefully analyze and process the data, and finally output the prediction results for each test sample. These prediction results not only reflect the model's understanding of the data, but also provide an important basis for evaluating the performance of the model. By comparing the error between the prediction results and the true value (output feature quantity) in the TBM excavation test set, the performance of the TBM excavation parameter model can be quantitatively evaluated, thereby further verifying the effectiveness and reliability of the model.

[0126] S170. Evaluate the performance of the TBM excavation parameter model using evaluation indicators.

[0127] In a specific implementation, in the process of developing and evaluating the prediction model, it is very important to select appropriate evaluation indicators, which can intuitively reflect the prediction performance of the model. In order to predict the TBM excavation parameters, the disclosed embodiment adopts three key prediction evaluation indicators: mean absolute percentage error (MAPE), relative mean square error (actually it should be the percentage form of root mean square error, but usually referred to as RMSE, which is generally understood as a percentage form or directly understood as root mean square error) and determination coefficient R 2 The following is a detailed description of these three evaluation indicators:

[0128] 1. Mean relative error (MAPE):

[0129] Definition: MAPE is a relative indicator to measure the deviation between the predicted value and the actual value, which is obtained by calculating the average of the ratio of the absolute value of all forecast errors to the actual value.

[0130] Calculation formula: MAPE = (1 / n) * Σ (|(predicted value - actual value) / actual value| * 100%), that is, formula (3):

[0131]

[0132] Where n is the number of prediction points.

[0133] Advantages: MAPE expresses the error in percentage form, which is convenient for intuitive understanding of prediction accuracy; it has good stability for large-scale data sets.

[0134] Application: In TBM excavation parameter prediction, MAPE can be used to evaluate the overall prediction performance of the model under different excavation conditions. A lower MAPE value means a higher prediction accuracy.

[0135] 2. Relative mean square error (RMSE, or understand it as a percentage):

[0136] Definition: RMSE is an absolute measure of the deviation between the predicted value and the actual value. It is calculated by taking the square root of the average of the sum of the squares of the prediction errors (if expressed as a percentage, multiply by 100%).

[0137] Calculation formula: RMSE = √[(1 / n)*Σ((predicted value - actual value)^2)] (If calculated in percentage form, multiply the result by 100%), that is, formula (4):

[0138]

[0139] Advantages: RMSE is highly sensitive to prediction errors and can reflect the fluctuations of the model during the prediction process.

[0140] Application: In TBM excavation parameter prediction, RMSE can be used to evaluate the stability of the prediction model in predicting specific excavation parameters. A lower RMSE value indicates that the prediction results of the model are more stable and reliable.

[0141] 3. Coefficient of determination R 2 :

[0142] Definition: Coefficient of determination R 2 , also known as goodness of fit, is a statistic that measures the fitting effect of a regression model. It indicates the degree to which the independent variables in the model explain the changes in the dependent variable.

[0143] Calculation formula: Determination coefficient R 2It is usually calculated by comparing the total sum of squares of the predicted values ​​of the regression model and the actual values ​​of the dependent variable with the total sum of squares of the actual values ​​of the dependent variable, as follows:

[0144]

[0145] Advantages: Coefficient of determination R 2 It can intuitively reflect the degree of fit of the regression model to the data. The closer its value is to 1, the better the model fit is and the stronger the explanatory power of the independent variable on the dependent variable is.

[0146] Application: In various regression analysis scenarios, such as the relationship analysis between TBM excavation parameters and influencing factors, the determination coefficient R 2 It can be used to evaluate the predictive accuracy and explanatory power of regression models. 2 A higher value means that the model is better able to predict and explain changes in the dependent variable.

[0147] Through the above steps, the mean relative error (MAPE), relative mean square error (RMSE, or understood as a percentage) and the determination coefficient R 2 These three evaluation indicators can comprehensively and objectively evaluate the performance of the TBM excavation parameter model.

[0148] In some embodiments, the performance of the TBM excavation parameter model can also be evaluated by comparing the error between the prediction result and the output feature quantity (true value) and the accuracy of the model.

[0149] In addition, after the above step S170 is implemented, the following steps are also included:

[0150] S180. According to the evaluation results, make necessary adjustments and optimizations to the TBM excavation parameter prediction model, such as adjusting the selection of base learners, the parameters of the PSO algorithm, the structure of the LSTM meta-learner, etc., to further improve its generalization ability and prediction accuracy.

[0151] The method for predicting TBM excavation parameters provided by the embodiment of the present disclosure includes: collecting and preprocessing TBM excavation parameter data; analyzing the correlation of data, screening out input feature quantities with high correlation with output feature quantities, and combining them with output feature quantities to form a TBM excavation data set; dividing the TBM excavation data set into a TBM excavation training set and a TBM excavation test set; constructing a TBM excavation parameter prediction model, which contains two layers; inputting the TBM excavation training set and the TBM excavation test set into the first layer model for training and prediction, and adjusting the parameters of the first layer prediction model by using a particle swarm optimization algorithm to obtain a meta-training set and a meta-test; inputting the meta-training set and the meta-test set into the second layer model for training and prediction, respectively, to obtain prediction results; and using evaluation indicators to evaluate the performance of the TBM excavation parameter model. The embodiment of the present disclosure uses the Spearman coefficient for correlation analysis, uses variables with a strong nonlinear relationship with the prediction results as model input features, screens the model input features, and combines five-fold cross validation and PSO algorithm to optimize parameters to select optimal parameters, and combines the Stacking ensemble learning method to complete the training and prediction on the original data set. Finally, the optimal TBM excavation parameter training set and test set are obtained. The meta-learner completes training on the optimal TBM excavation parameter training set and makes predictions on the optimal TBM excavation parameter test set. The PSO algorithm can effectively search the optimal parameter space by continuously adjusting and optimizing the parameters of each sub-model, so that the model can better fit the training data, thereby significantly improving the prediction accuracy of TBM performance. It effectively avoids model overfitting, improves the generalization ability of the model, and makes full use of the advantages of different learners. During the TBM construction process, various emergencies and abnormal data may be encountered. The new integrated learning model based on the PSO algorithm can better cope with these situations. It can quickly adapt to changes in data, adjust model parameters, maintain stable prediction performance, and provide reliable support for construction decisions.

[0152] Based on the above embodiments, the present disclosure also provides a device for predicting TBM excavation parameters, such as Figure 2 As shown, the device for predicting TBM excavation parameters includes:

[0153] The preprocessing module 201 is used to collect and preprocess the tunnel boring machine (TBM) excavation parameter data, obtain the preprocessed TBM excavation parameter data, and divide the TBM excavation parameter data into an input feature quantity and an output feature quantity;

[0154] An analysis module 202 is used to perform correlation analysis on the TBM excavation parameter data using the Spearman correlation coefficient, screen out input feature quantities that have a high correlation with the output feature quantities, and combine the input feature quantities with the output feature quantities to form a TBM excavation data set;

[0155] A division module 203 is used to divide the TBM excavation data set into a TBM excavation training set and a TBM excavation test set by using a cross-validation technology;

[0156] A construction module 204 is used to construct a TBM excavation parameter prediction model based on the stacking generalization principle, wherein the TBM excavation parameter prediction model includes two layers: a first layer prediction model and a second layer prediction model;

[0157] The first processing module 205 is used to input the TBM excavation training set into the first layer prediction model for training, use the TBM excavation test set for prediction, and use the particle swarm optimization algorithm to adjust the parameters of the first layer prediction model to determine the optimal parameter combination, so as to obtain a meta-training set and a meta-test set;

[0158] A second processing module 206, configured to input the meta-training set and the meta-testing set into the second-layer prediction model for training and prediction, respectively, to obtain a prediction result;

[0159] The evaluation module 207 is used to evaluate the performance of the TBM excavation parameter model using evaluation indicators.

[0160] In a specific implementation, the TBM excavation parameter data includes but is not limited to the following characteristic quantities:

[0161] Cutter disc speed, cutter disc set speed, tool torque, propulsion speed, thrust, clamping pressure, propulsion pump pressure, propulsion cylinder rod chamber pressure, propulsion cylinder rodless chamber pressure, clamping pump pressure, belt pump head pressure, belt motor pressure, cutter disc water spray pressure, cutter disc water spray flow, propulsion stroke, continuous belt operation, penetration and belt speed.

[0162] In a specific implementation, the preprocessing module 210 is specifically used to perform data cleaning, missing value processing, outlier detection, normalization processing and standardization processing on the TBM excavation parameter data to obtain preprocessed TBM excavation parameter data.

[0163] In a specific implementation, the analysis module 220 is specifically used to set at least two attributes in the pre-processed TBM excavation parameter data and assign corresponding values ​​to each of them; sort the values ​​of the attributes in ascending order, and replace the original numerical value of each value with its rank in the sorted sequence; calculate the rank difference between the two attribute rankings at each moment based on the ranking information after ranking; calculate the Spearman correlation coefficient between each input feature quantity and the output feature quantity based on the rank difference and the number of feature quantities; judge the correlation strength between the two variables according to the absolute value of the calculated Spearman correlation coefficient. If the absolute value of the Spearman correlation coefficient is high, it indicates that the correlation between the two variables is strong, and the feature quantity with a strong correlation with the output feature quantity is screened out, and the feature quantity with a strong correlation with the output feature quantity is combined with the output feature quantity to form a TBM excavation data set.

[0164] In a specific implementation, the division module 230 is specifically used to randomly divide the TBM excavation data set into five parts in each cross-validation, select four parts as the TBM excavation training sets, and the other part as the TBM excavation test set; repeat the above steps at least five times, and select a different TBM excavation data set each time.

[0165] In a specific implementation, the construction module 240 is specifically used to use the Stacking fusion algorithm and adopt a hierarchical prediction strategy to construct a two-layer prediction model;

[0166] Among them, the first-layer prediction model is composed of multiple base learners, and the second-layer prediction model is composed of a meta-learner, the base learners include but are not limited to GRU, SVM and XGBoot, and the meta-learner is but not limited to LSTM neural network.

[0167] In a specific implementation, the first processing module 250 is specifically used to randomly divide the TBM excavation input characteristic parameters into five parts in each cross-validation, select four parts as TBM excavation input characteristic parameter training sets, and the remaining part as TBM excavation input characteristic parameter test set; repeat the above steps at least five times, and select different TBM excavation input characteristic parameters each time.

[0168] In a specific embodiment, the evaluation indicators include: average relative error, relative mean square error and determination coefficient R 2 .

[0169] In a specific embodiment, it also includes:

[0170] The optimization module 280 is used to adjust and optimize the TBM excavation parameter prediction model according to the results obtained after the evaluation.

[0171] The device for predicting TBM excavation parameters provided by the embodiment of the present disclosure adopts the Spearman coefficient for correlation analysis, takes variables with strong nonlinear relationship with the prediction results as model input features, screens the model input features, and combines five-fold cross validation and PSO algorithm to optimize parameters to select the optimal parameters, and combines the Stacking ensemble learning method to complete the training and prediction on the original data set. Finally, the training set and test set of the meta-learner are obtained. The meta-learner completes the training on the training set of the meta-learner and predicts on the test set of the meta-learner. The PSO algorithm can effectively search the optimal parameter space by continuously adjusting and optimizing the parameters of each sub-model, so that the model can better fit the training data, thereby significantly improving the prediction accuracy of TBM performance. It effectively avoids model overfitting, improves the generalization ability of the model, and makes full use of the advantages of different learners. In the process of TBM construction, various emergencies and abnormal data may be encountered. The new ensemble learning model based on the PSO algorithm can better cope with these situations. It can quickly adapt to changes in data, adjust model parameters, maintain stable prediction performance, and provide reliable support for construction decisions.

[0172] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of the present invention. Ordinary technicians in this field can understand and implement it without paying creative work.

[0173] The following is a detailed description of the method for predicting TBM excavation parameters using a specific example.

[0174] 1. Collect TBM excavation parameters, including the following characteristic quantities: cutter head speed (X1), cutter head set speed (X2), tool torque (X3), propulsion speed (X4), thrust (X5), clamping pressure (X6), propulsion pump pressure (X7), propulsion cylinder rod chamber pressure (X8), propulsion cylinder rodless chamber pressure (X9), clamping pump pressure (X10), belt pump head pressure (X11), belt motor pressure (X12), cutter head water spray pressure (X13), cutter head water spray flow (X14), propulsion stroke (X15), continuous belt operation (X16), penetration (X17), belt speed (X18). These characteristic quantities are preprocessed through data cleaning, normalization, and standardization to obtain the preprocessed data. The TBM excavation parameter data is divided into input characteristic quantities and output characteristic quantities. In this embodiment, the input quantity is the cutter head speed (X1), the cutter head set speed (X2), the tool torque (X3), the propulsion speed (X4), the clamping pressure (X6), the propulsion pump pressure (X7), the propulsion cylinder rod chamber pressure (X8), the propulsion cylinder rodless chamber pressure (X9), the clamping pump pressure (X10), the belt pump head pressure (X11), the belt motor pressure (X12), the cutter head water spray pressure (X13), the cutter head water spray flow (X14), the propulsion stroke (X15), the continuous belt operation (X16), the penetration (X17), and the belt speed (X18). The output quantity characteristic is the thrust (X5).

[0175] 2. The Spearman correlation coefficient is used to perform correlation analysis on each characteristic quantity in the pre-processed TBM excavation parameter data to measure the monotonic relationship between each characteristic quantity.

[0176] First, since thrust (X5) is the predicted output, the Spearman correlation coefficient between each input feature value and thrust (X5) is calculated, and the following results are obtained:

[0177] |Feature quantity|Spearman correlation coefficient|

[0178] |---|---|

[0179] |Cutter speed (X1)|0.35|

[0180] |Setting speed of cutter head (X2)|0.52|

[0181] |Tool torque(X3)|0.29|

[0182] |Advance speed (X4)|0.68|

[0183] ||Tightening pressure (X6)|0.30|

[0184] |Thrust pump pressure (X7)|0.26|

[0185] |Thrust cylinder rod chamber pressure (X8)|0.19|

[0186] |Thrust cylinder rodless chamber pressure (X9)|0.22|

[0187] |Tightening pump pressure (X10)|0.38|

[0188] |Belt pump head pressure (X11)|0.11|

[0189] |Belt motor pressure (X12)|0.21|

[0190] |Water spray pressure of cutter head (X13)|0.18|

[0191] |Disc water spray flow rate (X14)|0.22|

[0192] |Advance stroke (X15)|0.24|

[0193] |Continuous belt operation (X16)|0.37|

[0194] |Penetration(X17)|0.21|

[0195] |Belt speed (X18)|0.25|

[0196] Through the Spearman coefficient correlation analysis, the input feature quantity with strong correlation with thrust (X5) is selected as the input feature of the TBM excavation parameter prediction model. In this example, the absolute values ​​of the Spearman correlation coefficients between the cutter head setting speed (X2) and the propulsion speed (X4) and the thrust (X5) are large (both greater than 0.40), so it can be considered that the cutter head setting speed (X2) and the propulsion speed (X4) are strongly correlated with the thrust (X5), and the cutter head setting speed (X2) and the propulsion speed (X4) are selected as the input feature quantities with strong correlation, while the absolute values ​​of the Spearman correlation coefficients between other input feature quantities and the thrust (X5) are small (all less than 0.40), so it can be considered that the cutter head setting speed (X2) and the propulsion speed (X4) are weakly correlated with the thrust (X5). Therefore, the TBM excavation input feature parameters include the following feature quantities: cutter head setting speed (X2) and propulsion speed (X4). There is a strong monotonic relationship between these characteristic quantities and the predicted characteristic quantity thrust (X5), which can provide key input for the subsequent TBM excavation parameter prediction model construction. The cutterhead set speed (X2), propulsion speed (X4) and thrust (X5) are combined to form a TBM excavation data set as the input of the subsequent TBM excavation parameter prediction model.

[0197] 3. Through cross-validation technology, the TBM excavation data set is divided into a TBM excavation training set and a TBM excavation test set; in this example, the TBM excavation training set contains 1,000 samples, which are divided into five subsets (each subset contains 200 samples) using a five-fold cross-validation method. In each cross-validation, four subsets (800 samples) are used as TBM excavation training sets for training the first-level prediction model, and the remaining subset (200 samples) is used as the TBM excavation test set.

[0198] 4. Building a stacked generalization model

[0199] Using the Stacking fusion algorithm and the layered prediction strategy, a two-layer prediction model is constructed;

[0200] (1) First-level prediction model

[0201] Three different machine learning algorithms are selected as the first-layer prediction models, such as support vector machine (SVM), random forest (RF), and gradient boosted decision tree (GBDT).

[0202] (2) Second-layer prediction model

[0203] The LSTM neural network is selected as the second-layer prediction model, also known as the meta-learner.

[0204] 5. The TBM excavation training set including the cutterhead set speed (X2), propulsion speed (X4) and thrust (X5) is input into the base learner to train the base learner, and the trained base learner is used to predict the TBM excavation test set. The PSO algorithm is used to adjust the parameters of the base learner to determine the optimal parameter combination, and a meta-training set and a meta-test set are obtained as the output of the base learner.

[0205] 6. Use the meta-training set as the input feature of the meta-learner and train the meta-learner until the model training is completed. Use the trained meta-learner to predict the meta-test set to obtain the prediction result of the TBM excavation parameter prediction model, that is, the predicted value of thrust.

[0206] 7. Use average relative error, relative mean square error and determination coefficient R 2 As evaluation indicators, the performance of the TBM excavation parameter prediction model is comprehensively evaluated to assess the prediction accuracy and generalization ability of the TBM excavation parameter model.

[0207] Through this process, a TBM excavation parameter prediction model is obtained, which can take advantage of the advantages of multiple machine learning algorithms to improve prediction accuracy and generalization ability. At the same time, through cross-validation and PSO parameter optimization, the stability and reliability of the model on different data sets can be ensured.

[0208] The present application also provides a computer device. Figure 3 , Figure 3 This is a basic structural block diagram of the computer device in this embodiment.

[0209] The computer device includes a memory 310 and a processor 320 that are connected to each other through a system bus. It should be noted that the figure only shows a computer device with components 310-320, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0210] Computer devices can be computing devices such as desktop computers, notebooks, PDAs, and cloud servers. Computer devices can interact with users through keyboards, mice, remote controls, touch pads, or voice control devices.

[0211] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes a non-volatile memory or a volatile memory, for example, a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc., and the RAM may include a static RAM or a dynamic RAM. In some embodiments, the memory 310 may be an internal storage unit of a computer device, for example, a hard disk or a memory of the computer device. In other embodiments, the memory 310 may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, or a flash card (FlashCard) equipped on the computer device. Of course, the memory 310 may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory 310 is generally used to store the operating system and various application software installed on the computer device, such as the program code of the above method. In addition, the memory 310 may also be used to temporarily store various data that have been output or are to be output.

[0212] The processor 320 is generally used to perform the overall operation of the computer device. In this embodiment, the memory 310 is used to store program codes or instructions, the program code includes computer operation instructions, and the processor 320 is used to execute the program codes or instructions stored in the memory 310 or process data, such as running the program code of the above method.

[0213] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus system can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0214] Another embodiment of the present application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads a computer-readable program code stored in the computer-readable medium, so that the processor can execute the functional actions specified in each step or a combination of steps in the above method; and generate a device for implementing the functional actions specified in each block or a combination of blocks in the block diagram.

[0215] Computer-readable media include but are not limited to electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any appropriate combination of the foregoing, the memory is used to store program codes or instructions, the program codes include computer operating instructions, and the processor is used to execute the program codes or instructions of the above methods stored in the memory.

[0216] For the definitions of memory and processor, please refer to the description of the aforementioned computer device embodiment and will not be repeated here.

[0217] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0218] Each functional unit or module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0219] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program code.

[0220] Unless the context clearly indicates otherwise, the singular form of the words used herein and in the appended claims includes the plural and vice versa. Thus, when referring to the singular, the plural form of the corresponding term is generally included. Similarly, the words "comprise" and "include" are to be interpreted as inclusive rather than exclusive. Likewise, the terms "include" and "or" should be interpreted as inclusive unless such interpretation is expressly prohibited herein. Where the term "example" is used herein, particularly when it is located after a group of terms, the "example" is merely exemplary and illustrative and should not be considered exclusive or comprehensive.

[0221] Further aspects and scopes of adaptability become apparent from the description provided herein. It should be understood that various aspects of the present application can be implemented individually or in combination with one or more other aspects. It should also be understood that the description and specific embodiments herein are intended for purposes of illustration only and are not intended to limit the scope of the present application.

[0222] Several embodiments of the present disclosure are described in detail above, but it is obvious that those skilled in the art can make various modifications and variations to the embodiments of the present disclosure without departing from the spirit and scope of the present disclosure. The protection scope of the present disclosure is defined by the attached claims.

Claims

1. A method for predicting TBM excavation parameters, characterized in that: include: Collecting and preprocessing tunnel boring machine (TBM) excavation parameter data to obtain preprocessed TBM excavation parameter data, and dividing the TBM excavation parameter data into input feature quantities and output feature quantities; Performing correlation analysis on the TBM excavation parameter data using the Spearman correlation coefficient, screening out the input feature quantity having a high correlation with the output feature quantity, and combining the input feature quantity with the output feature quantity to form a TBM excavation data set; By using cross-validation technology, the TBM excavation data set is divided into a TBM excavation training set and a TBM excavation test set; Based on the stacking generalization principle, a TBM excavation parameter prediction model is constructed, wherein the TBM excavation parameter prediction model includes two layers: a first layer prediction model and a second layer prediction model; Input the TBM excavation training set into the first-layer prediction model for training, use the TBM excavation test set for prediction, and use the particle swarm optimization algorithm to adjust the parameters of the first-layer prediction model to determine the optimal parameter combination, thereby obtaining a meta-training set and a meta-test set; Inputting the meta-training set and the meta-test set into the second-layer prediction model for training and prediction respectively to obtain prediction results; The performance of the TBM excavation parameter model is evaluated using evaluation indicators.

2. The method according to claim 1, characterized in that The TBM excavation parameter data set includes but is not limited to the following feature quantities: Cutter disc speed, cutter disc set speed, tool torque, propulsion speed, thrust, clamping pressure, propulsion pump pressure, propulsion cylinder rod chamber pressure, propulsion cylinder rodless chamber pressure, clamping pump pressure, belt pump head pressure, belt motor pressure, cutter disc water spray pressure, cutter disc water spray flow, propulsion stroke, continuous belt operation, penetration and belt speed.

3. The method according to claim 1, characterized in that The step of preprocessing the tunnel boring machine (TBM) excavation parameter data to obtain the preprocessed TBM excavation parameter data comprises: The TBM excavation parameter data is subjected to data cleaning, missing value processing, outlier detection, normalization processing and standardization processing to obtain the pre-processed TBM excavation parameter data.

4. The method according to claim 1, characterized in that: The step of performing correlation analysis on the input feature quantity by using the Spearman correlation coefficient, screening out the input feature quantity having a high correlation with the output feature quantity, and combining the input feature quantity with the output feature quantity to form a TBM excavation data set includes: At least two attributes are set in the pre-processed TBM excavation parameter data, and corresponding values ​​are assigned to each of them; Sort the values ​​of the attributes in ascending order, and replace the original value of each value with its rank in the sorted sequence; Based on the ranking information after ranking, calculate the level difference between the two attribute rankings at each moment; Based on the level difference and the number of feature quantities, calculating the Spearman correlation coefficient between each of the input feature quantities and the output feature quantity; The strength of the correlation between the two variables is determined based on the absolute value of the calculated Spearman correlation coefficient. If the absolute value of the Spearman correlation coefficient is high, it indicates that the correlation between the two variables is strong. The feature quantity with a strong correlation with the output feature quantity is screened out, and the feature quantity with a strong correlation with the output feature quantity is combined with the output feature quantity to form a TBM excavation data set.

5. The method according to claim 1, characterized in that The step of dividing the TBM excavation data set into a TBM excavation training set and a TBM excavation test set by using a cross-validation technology comprises: In each cross-validation, the TBM excavation data set is randomly divided into five parts, four of which are selected as the TBM excavation training set, and the other part is selected as the TBM excavation test set; The above steps are repeated at least five times, and a different TBM excavation data set is selected each time.

6. The method according to claim 1, characterized in that The steps of constructing a TBM excavation parameter prediction model based on the stacking generalization principle include: Using the Stacking fusion algorithm and the layered prediction strategy, a two-layer prediction model is constructed; Among them, the first-layer prediction model is composed of multiple base learners, and the second-layer prediction model is composed of a meta-learner, the base learners include but are not limited to GRU, SVM and XGBoot, and the meta-learner is but not limited to LSTM neural network.

7. The method according to claim 1, characterized in that The evaluation indicators include: average relative error, relative mean square error and determination coefficient R 2 ; After the step of evaluating the performance of the TBM excavation parameter model, the method further comprises: According to the results obtained after the evaluation, the TBM excavation parameter prediction model is adjusted and optimized.

8. A device for predicting TBM excavation parameters, characterized in that: include: A preprocessing module is used to collect and preprocess the tunnel boring machine (TBM) excavation parameter data, obtain the preprocessed TBM excavation parameter data, and divide the TBM excavation parameter data into an input feature quantity and an output feature quantity; An analysis module is used to perform correlation analysis on the TBM excavation parameter data using the Spearman correlation coefficient, screen out input feature quantities that have a high correlation with the output feature quantities, and combine the input feature quantities with the output feature quantities to form a TBM excavation data set; A partitioning module is used to partition the TBM excavation data set into a TBM excavation training set and a TBM excavation test set by using a cross-validation technology; A construction module is used to construct a TBM excavation parameter prediction model based on the stacking generalization principle, wherein the TBM excavation parameter prediction model includes two layers: a first layer prediction model and a second layer prediction model; A first processing module is used to input the TBM excavation training set into the first-layer prediction model for training, use the TBM excavation test set for prediction, and use a particle swarm optimization algorithm to adjust the parameters of the first-layer prediction model to determine the optimal parameter combination, thereby obtaining a meta-training set and a meta-test set; A second processing module, used for inputting the meta-training set and the meta-test set into the second-layer prediction model for training and prediction, respectively, to obtain a prediction result; The evaluation module is used to evaluate the performance of the TBM excavation parameter model using evaluation indicators.

9. A computer device, characterized in that: include: one or more processors; a storage device 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 method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images