Tunnel construction mud cake blockage prediction method and system based on knowledge assistance, terminal and storage medium

Abstract

The invention discloses a tunnel construction mud cake blockage prediction method and system based on knowledge assistance, a terminal and a storage medium. The method comprises the steps that knowledge feature data and corresponding mud cake blockage risk category labels are generated through a shield mud cake formation risk map; extracting target feature distribution from actual engineering data, simulating the target feature distribution by adopting a kernel density estimation algorithm, and generating supplementary feature data; processing the knowledge feature data, the mud cake blockage risk category labels and the supplementary feature data to obtain a synthetic data set, constructing a deep classification neural network model, and pre-training the deep classification neural network model by using the data set to obtain a preliminary classification model; and acquiring actual data of a shield tunneling site, finely adjusting the preliminary classification model by adopting a transfer learning technology, inputting the actual data into the model for classification prediction, and outputting the mud cake blockage condition of the shield tunneling site. According to the invention, real-time and accurate blocking risk early warning is provided for the construction site.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a knowledge-assisted prediction method, system, terminal and computer-readable storage medium for predicting mud cake blockage in tunnel construction. Background Art

[0002] Shield tunneling is a common tunnel construction technology, and its construction efficiency and safety have an important impact on the progress of the project. However, the occurrence of mud cake blockage during the tunneling process may cause construction stagnation and damage to equipment, so predicting the risk of mud cake blockage is one of the key technologies.

[0003] There are two main knowledge application paradigms for mud cake blockage risk prediction in the engineering field: on the one hand, there are knowledge paradigms such as analytical formulas and risk charts formed by summarizing expert experience. Although these methods contain rich engineering experience, the empirical information is often in the form of unstructured charts or texts, which are difficult to accurately express quantitatively, and their prediction results are often inaccurate; on the other hand, data-driven machine learning models construct prediction models through a large amount of actual field data. Although they have certain self-learning capabilities, due to the high cost of obtaining high-quality data, limited sample size, and complex and changeable engineering environment, the model lacks generalization and robustness, making it difficult to play a stable role in actual engineering.

[0004] In other words, most of the current application methods are limited to a single paradigm, and lack methods that can organically integrate expert experience and data-driven, thus failing to fully utilize the complementary advantages of the two.

[0005] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0006] The main purpose of the present invention is to provide a knowledge-assisted tunnel construction mud cake blockage prediction method, system, terminal and computer-readable storage medium, aiming to solve the problem that the risk prediction of mud cake blockage in the prior art is limited to a single paradigm, resulting in the inability to accurately predict the mud cake blockage situation at the shield excavation site.

[0007] To achieve the above object, the present invention provides a knowledge-assisted prediction method for tunnel construction mud cake blockage, the knowledge-assisted prediction method for tunnel construction mud cake blockage comprising the following steps: The boundary conditions in the shield mud cake risk map are converted into mathematical expressions through mathematical modeling, knowledge feature data is generated based on the mathematical expressions, and corresponding mud cake blockage risk category labels are annotated for the knowledge feature data according to the shield mud cake risk map; Extract target feature distribution from actual engineering data, simulate the target feature distribution using a kernel density estimation algorithm, and generate supplementary feature data; The knowledge feature data, the mud cake blockage risk category label and the supplementary feature data are subjected to data standardization and splicing processing to obtain a synthetic data set, and the synthetic data set is preprocessed to obtain a target data set; Constructing a deep classification neural network model, and using the target data set to pre-train the deep classification neural network model to obtain a preliminary classification model; Actual data from the shield tunneling site is collected and preprocessed to obtain an actual data set. Based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology to obtain a target classification model. The actual data is input into the target classification model for classification prediction, and the mud cake blockage situation at the shield tunneling site is output.

[0008] Optionally, in the knowledge-assisted tunnel construction mud cake blockage prediction method, the shield mud cake risk map includes: plastic limit water content, liquid limit water content and viscosity index; The viscosity index is represented by the plastic limit water content and the liquid limit water content.

[0009] Optionally, the knowledge-assisted tunnel construction mud cake blockage prediction method, wherein the knowledge feature data is generated based on the mathematical expression, and the corresponding mud cake blockage risk category label is labeled for the knowledge feature data according to the shield mud cake risk map, specifically includes: Within the boundary range of the mathematical expression, a plurality of knowledge feature data describing key indicators are generated by random sampling; According to the shield mud cake risk map, corresponding mud cake blockage risk category labels are annotated for the plurality of knowledge feature data.

[0010] Optionally, the knowledge-assisted tunnel construction mud cake blockage prediction method, wherein the use of a kernel density estimation algorithm to simulate the target feature distribution to generate supplementary feature data specifically includes: A Gaussian kernel function is selected as the kernel function of the kernel density estimation algorithm, an optimal bandwidth parameter of the kernel density estimation algorithm is determined by a grid search method, and a target kernel density estimation algorithm is constructed according to the kernel function and the optimal bandwidth parameter; The target kernel density estimation algorithm is used to simulate the target feature distribution to generate supplementary feature data, wherein the supplementary feature data includes shield cutter head rotation speed, cutter head torque, average tunneling speed, penetration, total thrust and porosity.

[0011] Optionally, the knowledge-assisted tunnel construction mud cake blockage prediction method, wherein the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data are subjected to data standardization and splicing processing to obtain a synthetic data set, specifically includes: The knowledge feature data, the mud cake blockage risk category label, the shield cutter head rotation speed, the cutter head torque, the average excavation speed, the penetration, the total thrust and the porosity are standardized to obtain standard knowledge feature data, standard mud cake blockage risk category label, standard shield cutter head rotation speed, standard cutter head torque, standard average excavation speed, standard penetration, standard total thrust and standard porosity; The standard knowledge feature data, the standard mud cake blockage risk category label, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average tunneling speed, the standard penetration, the standard total thrust and the standard porosity are spliced ​​to obtain a synthetic data set; Among them, the input features of the synthetic data set are the standard knowledge feature data, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average excavation speed, the standard penetration, the standard total thrust and the standard porosity, and the output feature of the synthetic data set is the standard mud cake blockage risk category label.

[0012] Optionally, the knowledge-assisted tunnel construction mud cake blockage prediction method, wherein the actual data is preprocessed to obtain an actual data set, and based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology to obtain a target classification model, specifically comprising: Performing normalization, standardization and outlier removal processing on the actual data to obtain pre-processed actual data, and constructing an actual data set based on the pre-processed actual data; Based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology, the overall network structure of the preliminary classification model is kept unchanged, all parameters of the preliminary classification model are frozen, and the weights of the last two layers of the preliminary classification model are unfrozen to obtain the target classification model.

[0013] Optionally, the knowledge-assisted tunnel construction mud cake blockage prediction method, wherein the actual data is input into the target classification model for classification prediction, and the mud cake blockage situation at the shield tunneling site is output, specifically includes: The actual data is input into the target classification model for classification prediction, and when the classification result output by the target classification model is non-blocking, it is determined that there is no risk of mud cake blockage at the shield tunneling site; When the classification result output by the target classification model is blockage, it is determined that there is a risk of mud cake blockage at the shield tunneling site, and the classification result is reported as a basis for on-site early warning.

[0014] In addition, to achieve the above-mentioned purpose, the present invention also provides a knowledge-assisted tunnel construction mud cake blockage prediction system, wherein the knowledge-assisted tunnel construction mud cake blockage prediction system comprises: A knowledge feature data generation module is used to convert the boundary conditions in the shield mud cake risk map into mathematical expressions through mathematical modeling, generate knowledge feature data based on the mathematical expressions, and annotate the knowledge feature data with corresponding mud cake blockage risk category labels according to the shield mud cake risk map; A supplementary feature data generation module is used to extract target feature distribution from actual engineering data, simulate the target feature distribution using a kernel density estimation algorithm, and generate supplementary feature data; A data set construction module is used to perform data standardization and splicing processing on the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set, and preprocess the synthetic data set to obtain a target data set; A model pre-training module is used to construct a deep classification neural network model, and pre-train the deep classification neural network model using the target data set to obtain a preliminary classification model; The blockage prediction module is used to collect actual data from the shield tunneling site, preprocess the actual data to obtain an actual data set, use transfer learning technology to fine-tune the preliminary classification model to obtain a target classification model, input the actual data into the target classification model for classification prediction, and output the mud cake blockage situation at the shield tunneling site.

[0015] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a knowledge-assisted tunnel construction mud cake blockage prediction program stored in the memory and executable on the processor, wherein the knowledge-assisted tunnel construction mud cake blockage prediction program implements the steps of the knowledge-assisted tunnel construction mud cake blockage prediction method as described above when executed by the processor.

[0016] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a knowledge-assisted tunnel construction mud cake blockage prediction program, and when the knowledge-assisted tunnel construction mud cake blockage prediction program is executed by a processor, the steps of the knowledge-assisted tunnel construction mud cake blockage prediction method as described above are implemented.

[0017] In the present invention, knowledge feature data and corresponding mud cake blockage risk category labels are generated through the shield mud cake risk map; target feature distribution is extracted from actual engineering data, and the target feature distribution is simulated by using the kernel density estimation algorithm to generate supplementary feature data; the knowledge feature data, mud cake blockage risk category labels and supplementary feature data are processed to obtain a synthetic data set, a deep classification neural network model is constructed, and the deep classification neural network model is pre-trained using the data set to obtain a preliminary classification model; actual data from the shield tunneling site is collected, and after the preliminary classification model is fine-tuned using the transfer learning technology, the actual data is input into the model for classification prediction, and the mud cake blockage situation at the shield tunneling site is output. The present invention provides a real-time and accurate blockage risk warning for the shield construction site. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of a preferred embodiment of the knowledge-assisted tunnel construction mud cake blockage prediction method of the present invention; Figure 2 It is a principle framework diagram of the knowledge-assisted tunnel construction mud cake blockage prediction method of the present invention; Figure 3 It is a shield mud cake risk determination diagram based on implicit knowledge in the knowledge-assisted tunnel construction mud cake blockage prediction method of the present invention; Figure 4 It is a data scatter plot matrix based on actual engineering in the knowledge-assisted tunnel construction mud cake blockage prediction method of the present invention; Figure 5 It is a scatter plot matrix of constructed data based on KDE in the knowledge-assisted tunnel construction mud cake blockage prediction method of the present invention; Figure 6 It is a structural diagram of a preferred embodiment of the knowledge-assisted tunnel construction mud cake blockage prediction system of the present invention; Figure 7 It is a structural diagram of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION

[0019] The present application provides a knowledge-assisted prediction method, system, terminal and storage medium for predicting mud cake blockage in tunnel construction. In order to make the purpose, technical solution and effect of the present application clearer and more specific, the present application is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with those in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0021] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0022] The knowledge-assisted tunnel construction mud cake blockage prediction method described in the preferred embodiment of the present invention is as follows: Figure 1 and Figure 2 As shown, the knowledge-assisted tunnel construction mud cake blockage prediction method includes the following steps: Step S10: converting the boundary conditions in the shield mud cake risk map into mathematical expressions through mathematical modeling, generating knowledge feature data based on the mathematical expressions, and labeling the knowledge feature data with corresponding mud cake blockage risk category labels according to the shield mud cake risk map.

[0023] Specifically, the shield mud cake risk map (such as Figure 3 As shown in the figure, mathematical modeling is performed to convert each boundary condition in the shield mud cake risk diagram into a mathematical expression. In this embodiment, the shield mud cake risk diagram includes: plastic limit water content (WP-Wn), liquid limit water content (WL-Wn), plasticity index (Ip) and viscosity index (Ic), wherein the plastic limit water content is the horizontal coordinate x, and the value range of the plastic limit water content is [-160%, 50%], the liquid limit water content is the vertical coordinate y, and the value range of the liquid limit water content is [-20%, 200%], and the viscosity index is jointly represented by the plastic limit water content and the liquid limit water content.

[0024] Furthermore, the ordinate (WL-Wn) of all sample points in the shield mud cake risk map must be greater than the abscissa (WP-Wn). At the same time, the risk is divided according to the viscosity index, and the viscosity index is converted into a relationship that can be represented by the plastic limit water content and the liquid limit water content. For example: Ic=0, which can be represented by y=0% in the shield mud cake risk map; Ic=0.5, which can be represented by y=-x in the shield mud cake risk map; Ic=0.75, which can be represented by y=-3x in the shield mud cake risk map; Ic=1, which can be represented by x=0% in the shield mud cake risk map.

[0025] Furthermore, the generating of knowledge feature data based on the mathematical expression and labeling of corresponding mud cake blockage risk category labels for the knowledge feature data according to the shield mud cake risk map specifically includes: Within the boundary range of the mathematical expression, a plurality of knowledge feature data describing key indicators are generated by random sampling; According to the shield mud cake risk map, corresponding mud cake blockage risk category labels are annotated for the plurality of knowledge feature data.

[0026] In this embodiment, random sampling is used to generate knowledge feature data describing key indicators such as plastic limit moisture content and liquid limit moisture content within the above boundary range, and the corresponding mud cake blockage risk category is marked for each sample according to the risk map.

[0027] Table 1: Sample point label assignment

[0028] Step S20: extract target feature distribution from actual engineering data, simulate the target feature distribution using a kernel density estimation algorithm, and generate supplementary feature data.

[0029] Specifically, a Gaussian kernel function is selected as the kernel function of the kernel density estimation algorithm, a grid search method is used to determine the optimal bandwidth parameter of the kernel density estimation algorithm, and a target kernel density estimation algorithm is constructed according to the kernel function and the optimal bandwidth parameter.

[0030] It is understandable that this application is to supplement the shield tunneling related parameters (such as Figure 4 As shown in Figure 2, the target feature distribution is extracted from the actual engineering data, and the kernel density estimation (KDE) method is used to generate supplementary feature data similar to the real data distribution (such as Figure 5 shown), where Figure 4 and Figure 5 a, b, c, d, and e represent the shield cutter head rotation speed, cutter head torque, average excavation speed, penetration, total thrust, and porosity, respectively.

[0031] Furthermore, the target kernel density estimation algorithm is used to simulate the target feature distribution to generate supplementary feature data, wherein the supplementary feature data includes shield cutter head rotation speed, cutter head torque, average excavation speed, penetration, total thrust and porosity.

[0032] In this embodiment, the extracted engineering parameters include shield cutter head speed, cutter head torque, average excavation speed, penetration, total thrust and porosity. The specific parameters of the kernel density estimation algorithm are set as follows: the kernel function uses the Gaussian kernel function, the bandwidth parameter is determined by the grid search method, the bandwidth parameter range is set to [0.1, 10], and the logarithmic scale is uniformly distributed. A total of 20 candidate values ​​are selected, and the optimal bandwidth is evaluated to be 0.6951. Because the risk of shield mud cake only covers the two types of data, liquid limit water content and plastic limit water content, if you want to generate other characteristic data, you can find other actual engineering data, and use the kernel density estimation method to imitate the characteristic data of these actual projects, and then generate other characteristics. Then the final data set can be expressed as: [liquid limit water content, plastic limit water content, shield cutter head speed, cutter head torque, ... porosity].

[0033] Step S30: performing data standardization and splicing processing on the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set, and preprocessing the synthetic data set to obtain a target data set.

[0034] The data standardization and splicing of the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set specifically includes: The knowledge feature data, the mud cake blockage risk category label, the shield cutter head rotation speed, the cutter head torque, the average excavation speed, the penetration, the total thrust and the porosity are standardized to obtain standard knowledge feature data, standard mud cake blockage risk category label, standard shield cutter head rotation speed, standard cutter head torque, standard average excavation speed, standard penetration, standard total thrust and standard porosity; The standard knowledge feature data, the standard mud cake blockage risk category label, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average tunneling speed, the standard penetration, the standard total thrust and the standard porosity are spliced ​​to obtain a synthetic data set; Among them, the input features of the synthetic data set are the standard knowledge feature data, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average excavation speed, the standard penetration, the standard total thrust and the standard porosity, and the output feature of the synthetic data set is the standard mud cake blockage risk category label.

[0035] It can be understood that the present application unifies the dimensions of the knowledge feature data and the generated supplementary feature data and then splices them to form a synthetic data set containing all features and risk labels, organically integrating expert experience and data-driven, thereby making full use of the complementary advantages of the two.

[0036] Furthermore, the synthetic data set is preprocessed, including normalization, standardization and outlier removal, to obtain a target data set, and the target data set is used to pretrain a deep classification neural network model so that it has a preliminary classification capability for mud cake blockage risks.

[0037] Step S40: construct a deep classification neural network model, and use the target data set to pre-train the deep classification neural network model to obtain a preliminary classification model.

[0038] It can be understood that a deep classification neural network model (8×64×128×64×1) is pre-trained using the constructed target data set. The target data set is divided into a training set, a test set, and a validation set according to a preset ratio, the training set is used to train the deep classification neural network model, and the test set is used to evaluate the deep classification neural network model in each round of training to obtain a trained model, and the validation set is used to evaluate the trained model to obtain a preliminary classification model.

[0039] Furthermore, in the pre-training process of the model, the training parameters are set, and the optimizer is set to Adam (the main function of the Adam optimizer is to update the neural network parameters according to the gradient information, thereby minimizing the loss function), the loss function is the cross entropy loss, the learning rate is 0.001, and the number of iterations is 100. After the pre-training is completed, a preliminary classification model is obtained, and the preliminary classification model can preliminarily realize the classification prediction of the mud cake blockage risk.

[0040] Step S50, collecting actual data from the shield tunneling site, preprocessing the actual data to obtain an actual data set, based on the actual data set, using transfer learning technology to fine-tune the preliminary classification model to obtain a target classification model, inputting the actual data into the target classification model for classification prediction, and outputting the mud cake blockage situation at the shield tunneling site.

[0041] Specifically, the actual data is preprocessed, including normalization, standardization and outlier removal, to obtain preprocessed actual data, and an actual data set is constructed based on the preprocessed actual data.

[0042] Normalization is a way to simplify calculations, that is, to transform a dimensional expression into a dimensionless expression, which becomes a scalar. Normalization usually converts data into a distribution with the same standard deviation and mean. Normalization can make data easier to compare and analyze. Eliminating outliers refers to the method of processing or excluding extreme data that deviates from the normal range in data analysis.

[0043] Furthermore, based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology, the overall network structure of the preliminary classification model is kept unchanged, all parameters of the preliminary classification model are frozen, and the weights of the last two layers of the preliminary classification model are unfrozen to obtain the target classification model.

[0044] It is understandable that transfer learning is a machine learning method, which means that a pre-trained model is reused in another task. In this embodiment, based on the actual data set, the transfer learning technology is used to fine-tune the pre-trained preliminary classification model. In this process, the overall network structure of the model is kept unchanged, some model parameters are frozen, and only the weights of specific layers are adjusted, so that the model can quickly adapt to the actual working conditions and output classification results that directly reflect the risk of on-site congestion. For example, first freeze all the parameters of the pre-trained model, and only unfreeze the weights of the last two layers (i.e., 128×64×1), so that the model can quickly adapt to the actual working conditions on site, thereby further improving the classification performance.

[0045] Furthermore, the actual data is input into the target classification model for classification prediction. When the classification result output by the target classification model is non-blockage, it is determined that there is no risk of mud cake blockage at the shield tunneling site; when the classification result output by the target classification model is blockage, it is determined that there is a risk of mud cake blockage at the shield tunneling site, and the classification result is reported as a basis for on-site early warning.

[0046] It can be seen that this application constructs a synthetic data set with domain knowledge by converting expert experience in the shield risk map into a mathematical model to generate knowledge feature data, and using kernel density estimation to generate supplementary feature data. Then, the data set is used to pre-train a deep classification neural network, and the pre-trained model is fine-tuned through transfer learning, realizing an effective combination of expert experience knowledge and data-driven methods. This method not only reduces the cost of data collection, but also greatly improves the generalization ability and prediction accuracy of the model under actual engineering conditions, providing real-time and accurate blockage risk warnings for shield construction sites, and has broad prospects for engineering applications.

[0047] The beneficial effects of the present invention are as follows: (1) Using risk graphs to generate synthetic data for pre-training, we can explicitly inject domain knowledge and reduce data collection and annotation costs. (2) Using deep transfer learning technology, the pre-trained model can quickly adapt to the actual working conditions on site, directly output the classification results of blockage risks, and simplify the model structure; (3) This method innovatively realizes the effective connection between the expert experience knowledge paradigm and the data-driven machine learning paradigm. It not only overcomes the problem of insufficient quantification and precision of the traditional expert experience method, but also avoids the defects of the pure data-driven method in the case of insufficient generalization and robustness when there are insufficient samples, thus significantly improving the prediction accuracy and engineering adaptability; (4) It provides real-time and accurate blockage risk warning for shield construction sites and has broad engineering application prospects.

[0048] Furthermore, if Figure 6 As shown, based on the above-mentioned knowledge-assisted tunnel construction mud cake blockage prediction method, the present invention also provides a knowledge-assisted tunnel construction mud cake blockage prediction system, wherein the knowledge-assisted tunnel construction mud cake blockage prediction system includes: The knowledge feature data generating module 51 is used to convert the boundary conditions in the shield mud cake risk map into mathematical expressions through mathematical modeling, generate knowledge feature data based on the mathematical expressions, and annotate the knowledge feature data with corresponding mud cake blockage risk category labels according to the shield mud cake risk map; A supplementary feature data generation module 52 is used to extract target feature distribution from actual engineering data, simulate the target feature distribution using a kernel density estimation algorithm, and generate supplementary feature data; A data set construction module 53 is used to perform data standardization and splicing processing on the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set, and preprocess the synthetic data set to obtain a target data set; A model pre-training module 54 is used to construct a deep classification neural network model, and pre-train the deep classification neural network model using the target data set to obtain a preliminary classification model; The blockage prediction module 55 is used to collect actual data from the shield tunneling site, pre-process the actual data to obtain an actual data set, use transfer learning technology to fine-tune the preliminary classification model to obtain a target classification model, input the actual data into the target classification model for classification prediction, and output the mud cake blockage situation at the shield tunneling site.

[0049] Furthermore, if Figure 7 As shown, based on the above-mentioned knowledge-assisted tunnel construction mud cake blockage prediction method and system, the present invention also provides a terminal accordingly, and the terminal includes a processor 10, a memory 20 and a display 30. Figure 7 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0050] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, a knowledge-assisted tunnel construction mud cake blockage prediction program 40 is stored on the memory 20, and the knowledge-assisted tunnel construction mud cake blockage prediction program 40 can be executed by the processor 10, thereby realizing the knowledge-assisted tunnel construction mud cake blockage prediction method in the present application.

[0051] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the knowledge-assisted tunnel construction mud cake blockage prediction method.

[0052] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0053] In one embodiment, when the processor 10 executes the knowledge-assisted tunnel construction mud cake blockage prediction program 40 in the memory 20, the following steps are implemented: The boundary conditions in the shield mud cake risk map are converted into mathematical expressions through mathematical modeling, knowledge feature data is generated based on the mathematical expressions, and corresponding mud cake blockage risk category labels are annotated for the knowledge feature data according to the shield mud cake risk map; Extracting target feature distribution from actual engineering data, simulating the target feature distribution using a kernel density estimation algorithm, and generating supplementary feature data; The knowledge feature data, the mud cake blockage risk category label and the supplementary feature data are subjected to data standardization and splicing processing to obtain a synthetic data set, and the synthetic data set is preprocessed to obtain a target data set; Constructing a deep classification neural network model, and using the target data set to pre-train the deep classification neural network model to obtain a preliminary classification model; Actual data from the shield tunneling site is collected and preprocessed to obtain an actual data set. Based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology to obtain a target classification model. The actual data is input into the target classification model for classification prediction, and the mud cake blockage situation at the shield tunneling site is output.

[0054] The shield mud cake risk map includes: plastic limit water content, liquid limit water content and viscosity index; The viscosity index is represented by the plastic limit water content and the liquid limit water content.

[0055] The method of generating knowledge feature data based on the mathematical expression and labeling the knowledge feature data with corresponding mud cake blockage risk category labels according to the shield mud cake risk map specifically includes: Within the boundary range of the mathematical expression, a plurality of knowledge feature data describing key indicators are generated by random sampling; According to the shield mud cake risk map, corresponding mud cake blockage risk category labels are annotated for the plurality of knowledge feature data.

[0056] The method of simulating the target feature distribution by using a kernel density estimation algorithm to generate supplementary feature data specifically includes: A Gaussian kernel function is selected as the kernel function of the kernel density estimation algorithm, an optimal bandwidth parameter of the kernel density estimation algorithm is determined by a grid search method, and a target kernel density estimation algorithm is constructed according to the kernel function and the optimal bandwidth parameter; The target kernel density estimation algorithm is used to simulate the target feature distribution to generate supplementary feature data, wherein the supplementary feature data includes shield cutter head rotation speed, cutter head torque, average tunneling speed, penetration, total thrust and porosity.

[0057] The process of normalizing and concatenating the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set specifically includes: The knowledge feature data, the mud cake blockage risk category label, the shield cutter head rotation speed, the cutter head torque, the average excavation speed, the penetration, the total thrust and the porosity are standardized to obtain standard knowledge feature data, standard mud cake blockage risk category label, standard shield cutter head rotation speed, standard cutter head torque, standard average excavation speed, standard penetration, standard total thrust and standard porosity; The standard knowledge feature data, the standard mud cake blockage risk category label, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average tunneling speed, the standard penetration, the standard total thrust and the standard porosity are spliced ​​to obtain a synthetic data set; Among them, the input features of the synthetic data set are the standard knowledge feature data, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average excavation speed, the standard penetration, the standard total thrust and the standard porosity, and the output feature of the synthetic data set is the standard mud cake blockage risk category label.

[0058] The preprocessing of the actual data to obtain an actual data set, and fine-tuning the preliminary classification model using transfer learning technology based on the actual data set to obtain a target classification model specifically includes: Performing normalization, standardization and outlier removal processing on the actual data to obtain pre-processed actual data, and constructing an actual data set based on the pre-processed actual data; Based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology, the overall network structure of the preliminary classification model is kept unchanged, all parameters of the preliminary classification model are frozen, and the weights of the last two layers of the preliminary classification model are unfrozen to obtain the target classification model.

[0059] The step of inputting the actual data into the target classification model for classification prediction and outputting the mud cake blockage situation at the shield tunneling site specifically includes: The actual data is input into the target classification model for classification prediction, and when the classification result output by the target classification model is non-blocking, it is determined that there is no risk of mud cake blockage at the shield tunneling site; When the classification result output by the target classification model is blockage, it is determined that there is a risk of mud cake blockage at the shield tunneling site, and the classification result is reported as a basis for on-site early warning.

[0060] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a knowledge-assisted tunnel construction mud cake blockage prediction program, and when the knowledge-assisted tunnel construction mud cake blockage prediction program is executed by a processor, the steps of the knowledge-assisted tunnel construction mud cake blockage prediction method as described above are implemented.

[0061] In summary, the present invention proposes a knowledge-assisted prediction method, system, terminal and storage medium for mud cake blockage in tunnel construction. The method includes: generating knowledge feature data and corresponding mud cake blockage risk category labels through a shield mud cake risk map; extracting target feature distribution from actual engineering data, simulating target feature distribution using a kernel density estimation algorithm, and generating supplementary feature data; processing knowledge feature data, mud cake blockage risk category labels and supplementary feature data to obtain a synthetic data set, constructing a deep classification neural network model, and using the data set to pre-train the deep classification neural network model to obtain a preliminary classification model; collecting actual data from the shield tunneling site, using transfer learning technology to fine-tune the preliminary classification model, and then inputting the actual data into the model for classification prediction, and outputting the mud cake blockage situation at the shield tunneling site. The present invention extracts professional knowledge from the shield risk map to generate synthetic data to pre-train the deep network, and then uses the actual data on site for migration and fine-tuning, so that the model directly outputs the blockage risk classification result. The method effectively reduces the cost of data acquisition, makes up for the lack of domain knowledge injection in traditional transfer learning, and significantly improves the recognition accuracy and robustness of the model for low-probability blockage events.

[0062] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or terminal including the element.

[0063] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0064] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A knowledge-assisted tunnel construction mud cake blockage prediction method, characterized in that: The knowledge-assisted tunnel construction mud cake blockage prediction method includes: The boundary conditions in the shield mud cake risk map are converted into mathematical expressions through mathematical modeling, knowledge feature data is generated based on the mathematical expressions, and corresponding mud cake blockage risk category labels are annotated for the knowledge feature data according to the shield mud cake risk map; Extracting target feature distribution from actual engineering data, simulating the target feature distribution using a kernel density estimation algorithm, and generating supplementary feature data; The knowledge feature data, the mud cake blockage risk category label and the supplementary feature data are subjected to data standardization and splicing processing to obtain a synthetic data set, and the synthetic data set is preprocessed to obtain a target data set; Constructing a deep classification neural network model, and using the target data set to pre-train the deep classification neural network model to obtain a preliminary classification model; Actual data from the shield tunneling site is collected and preprocessed to obtain an actual data set. Based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology to obtain a target classification model. The actual data is input into the target classification model for classification prediction, and the mud cake blockage situation at the shield tunneling site is output.

2. The knowledge-assisted tunnel construction mud cake blockage prediction method according to claim 1 is characterized in that: The shield mud cake risk map includes: plastic limit water content, liquid limit water content and viscosity index; The viscosity index is represented by the plastic limit water content and the liquid limit water content.

3. The knowledge-assisted tunnel construction mud cake blockage prediction method according to claim 1 is characterized in that: The generating of knowledge feature data based on the mathematical expression and labeling of corresponding mud cake blockage risk category labels for the knowledge feature data according to the shield mud cake risk map specifically includes: Within the boundary range of the mathematical expression, a plurality of knowledge feature data describing key indicators are generated by random sampling; According to the shield mud cake risk map, corresponding mud cake blockage risk category labels are annotated for the plurality of knowledge feature data.

4. The knowledge-assisted tunnel construction mud cake blockage prediction method according to claim 1 is characterized in that: The use of a kernel density estimation algorithm to simulate the target feature distribution to generate supplementary feature data specifically includes: A Gaussian kernel function is selected as the kernel function of the kernel density estimation algorithm, an optimal bandwidth parameter of the kernel density estimation algorithm is determined by a grid search method, and a target kernel density estimation algorithm is constructed according to the kernel function and the optimal bandwidth parameter; The target kernel density estimation algorithm is used to simulate the target feature distribution to generate supplementary feature data, wherein the supplementary feature data includes shield cutter head rotation speed, cutter head torque, average tunneling speed, penetration, total thrust and porosity.

5. The knowledge-assisted tunnel construction mud cake blockage prediction method according to claim 4 is characterized in that: The data standardization and splicing of the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set specifically includes: The knowledge feature data, the mud cake blockage risk category label, the shield cutter head rotation speed, the cutter head torque, the average excavation speed, the penetration, the total thrust and the porosity are standardized to obtain standard knowledge feature data, standard mud cake blockage risk category label, standard shield cutter head rotation speed, standard cutter head torque, standard average excavation speed, standard penetration, standard total thrust and standard porosity; The standard knowledge feature data, the standard mud cake blockage risk category label, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average tunneling speed, the standard penetration, the standard total thrust and the standard porosity are spliced ​​to obtain a synthetic data set; Among them, the input features of the synthetic data set are the standard knowledge feature data, the standard shield cutter head rotation speed, the standard cutter head torque, the standard average excavation speed, the standard penetration, the standard total thrust and the standard porosity, and the output feature of the synthetic data set is the standard mud cake blockage risk category label.

6. The knowledge-assisted tunnel construction mud cake blockage prediction method according to claim 5 is characterized in that: The preprocessing of the actual data to obtain an actual data set, and fine-tuning the preliminary classification model using a transfer learning technique based on the actual data set to obtain a target classification model specifically includes: Performing normalization, standardization and outlier removal processing on the actual data to obtain pre-processed actual data, and constructing an actual data set based on the pre-processed actual data; Based on the actual data set, the preliminary classification model is fine-tuned using transfer learning technology, the overall network structure of the preliminary classification model is kept unchanged, all parameters of the preliminary classification model are frozen, and the weights of the last two layers of the preliminary classification model are unfrozen to obtain the target classification model.

7. The knowledge-assisted tunnel construction mud cake blockage prediction method according to claim 1 is characterized in that: The actual data is input into the target classification model for classification prediction, and the mud cake blockage situation at the shield tunneling site is output, specifically including: The actual data is input into the target classification model for classification prediction, and when the classification result output by the target classification model is non-blocking, it is determined that there is no risk of mud cake blockage at the shield tunneling site; When the classification result output by the target classification model is blockage, it is determined that there is a risk of mud cake blockage at the shield tunneling site, and the classification result is reported as a basis for on-site early warning.

8. A knowledge-assisted tunnel construction mud cake blockage prediction system, characterized in that: The knowledge-assisted tunnel construction mud cake blockage prediction system includes: A knowledge feature data generation module is used to convert the boundary conditions in the shield mud cake risk map into mathematical expressions through mathematical modeling, generate knowledge feature data based on the mathematical expressions, and annotate the knowledge feature data with corresponding mud cake blockage risk category labels according to the shield mud cake risk map; A supplementary feature data generation module is used to extract target feature distribution from actual engineering data, simulate the target feature distribution using a kernel density estimation algorithm, and generate supplementary feature data; A data set construction module is used to perform data standardization and splicing processing on the knowledge feature data, the mud cake blockage risk category label and the supplementary feature data to obtain a synthetic data set, and preprocess the synthetic data set to obtain a target data set; A model pre-training module is used to construct a deep classification neural network model, and pre-train the deep classification neural network model using the target data set to obtain a preliminary classification model; The blockage prediction module is used to collect actual data from the shield tunneling site, preprocess the actual data to obtain an actual data set, use transfer learning technology to fine-tune the preliminary classification model to obtain a target classification model, input the actual data into the target classification model for classification prediction, and output the mud cake blockage situation at the shield tunneling site.

9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a knowledge-assisted tunnel construction mud cake blockage prediction program stored in the memory and executable on the processor. When the knowledge-assisted tunnel construction mud cake blockage prediction program is executed by the processor, the steps of the knowledge-assisted tunnel construction mud cake blockage prediction method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a knowledge-assisted tunnel construction mud cake blockage prediction program, which, when executed by a processor, implements the steps of the knowledge-assisted tunnel construction mud cake blockage prediction method as described in any one of claims 1 to 7.

Images