Tunnel end effect intelligent calculation method, device and equipment and readable storage medium

Abstract

The invention discloses a tunnel end effect intelligent calculation method, device and equipment and a readable storage medium, and relates to the technical field of tunnel engineering. Comprising the following steps: acquiring input parameters required by tunnel end effect calculation, and determining an evaluation object and a quantitative index according to tunnel end dynamic response characteristics; on the basis of the evaluation object and the quantitative index, tunnel end dynamic response is calculated or historical data is obtained under different working conditions, a sample data set containing input parameters and corresponding quantitative index results is constructed, and a machine learning method is used for training to obtain a tunnel end effect intelligent calculation model; and inputting input parameters of a tunnel to be evaluated into the trained intelligent calculation model, obtaining a dynamic response quantification result of the evaluation object under the quantification index, and applying the result to engineering analysis and optimization. The method provides a quantitative basis for engineering decision.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering technology, and more specifically to a method, apparatus, device, and readable storage medium for intelligent calculation of tunnel end effects. Background Technology

[0002] During construction and operation, tunnels may be subjected to various dynamic forces, including seismic activity, blasting vibrations, train dynamic loads, landslide impacts, and gas explosions. Compared to the middle section of a long tunnel, the tunnel ends, due to abrupt changes in geometry and boundary conditions, are prone to significant end effects. These end effects manifest as the reflection, superposition, and energy accumulation of dynamic waves in the tunnel end region, leading to a significant amplification of displacement, acceleration, stress-strain, and damage indicators. This makes this area a critical target in tunnel dynamic safety analysis. Therefore, efficient and accurate quantitative analysis of the dynamic response at tunnel ends is urgently needed in engineering design, construction organization, and safety assessment to support relevant engineering judgments and decisions.

[0003] In existing technologies, the analysis of dynamic response at tunnel ends mainly relies on numerical calculation methods such as finite element method, finite difference method, and discrete element method. High-fidelity models are constructed to perform multi-condition simulations to obtain response results. However, these methods are typically highly complex computational processes, suffering from numerous modeling parameters, long calculation cycles, and high costs associated with repetitive calculations. Especially under conditions of multiple parameters and multiple working conditions, efficient data processing and rapid calculation are difficult to achieve, limiting their application in real-time engineering assessments.

[0004] With the development of computer technology and data-driven methods, some studies have begun to introduce machine learning models to predict tunnel dynamic responses. By training these models on historical simulation or monitoring data, a mapping relationship between input parameters and response results is established, thereby improving computational efficiency. However, existing methods still have shortcomings in data processing and model building: on the one hand, there is a lack of data feature organization methods and dedicated computational models for tunnel end effects, making it difficult to accurately characterize key influencing factors related to end effects; on the other hand, the data processing process often focuses on single-result prediction, lacking unified modeling and comprehensive computational capabilities for multi-dimensional response data.

[0005] Furthermore, at the engineering application level, existing technologies typically output calculation results as independent numerical values, lacking further data processing and structured analysis based on the calculation results. This makes it difficult to form a unified evaluation index system and directly support decision-making processes such as engineering design optimization, construction parameter adjustment, and risk control, resulting in a disconnect between calculation results and engineering decisions.

[0006] Therefore, how to construct a data processing model for tunnel end effects under multiple working conditions and multiple parameters, realize rapid calculation of dynamic response and comprehensive analysis of multiple indicators, and further transform the calculation results into structured information that can be used for engineering decision-making has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] The purpose of this invention is to provide an intelligent calculation method, device, equipment, and readable storage medium for tunnel end effects. By constructing a quantitative index system for tunnel end effects, it unifies and structures the multi-dimensional data such as the amplification degree, influence range, and dynamic response at key locations of end effects, realizing the transformation from single numerical output to comprehensive calculation of tunnel end dynamic response. This ensures that the calculation results under different working conditions and parameter conditions have good comparability and analyzability, thereby improving the standardization of data processing and the application value of calculation results. Simultaneously, this invention establishes a mapping model between input parameters and dynamic response based on machine learning methods. By training and calculating on multi-source data, in practical applications, only relevant parameters need to be input to quickly obtain prediction results, avoiding the repetitive calculation process of traditional high-fidelity numerical simulation. This significantly improves computational efficiency and data processing capabilities, reduces computational resource consumption, and enhances the applicability and stability of the model under various working conditions. Furthermore, based on the obtained dynamic response calculation results, the present invention performs further data processing and comprehensive analysis on the results to form evaluation index outputs for engineering applications. Based on the indicators, engineering scheme comparison, construction parameter optimization and risk control strategy generation are realized, so that the calculation results can directly serve the engineering decision-making process, improve the decision support capability of tunnel end effect analysis, and realize the effective connection between data calculation and engineering application.

[0008] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides an intelligent calculation method for tunnel end effects, the method comprising: Obtain the input parameters required for calculating tunnel end effects, and determine the evaluation object and quantitative index based on the dynamic response characteristics of the tunnel end. The quantitative index is used to characterize the dynamic response level of the tunnel end region. Based on the evaluation object and quantitative indicators, the dynamic response of the tunnel end is calculated or historical data is obtained under different working conditions. A sample dataset containing input parameters and corresponding quantitative indicator results is constructed, and a machine learning method is used to train an intelligent calculation model of tunnel end effect. The input parameters of the tunnel to be evaluated are input into the trained intelligent computing model to obtain the dynamic response quantification results of the evaluation object under the quantitative index, and the results are used for engineering analysis and optimization.

[0009] In some embodiments, the input parameters required for calculating tunnel end effects are obtained, and the evaluation object and quantitative indicators are determined based on the dynamic response characteristics of the tunnel end, including: Obtain the engineering parameters and dynamic parameters of the tunnel to be analyzed. The engineering parameters include at least the tunnel geometry, portal or end location, surrounding rock and support structure parameters, portal topographic parameters, damping parameters, and end boundary conditions. The end boundary condition parameters include the portal free surface conditions, portal constraint form, and end transition section structural abrupt change parameters. The dynamic parameters include at least one of the following: seismic motion parameters, blasting vibration parameters, or train dynamic load parameters. The evaluation scope and key locations of the tunnel end area are determined as the evaluation objects, and dynamic response quantification indicators are selected to characterize the end effect. The indicators include the dynamic response of the key end locations, as well as the amplification degree or influence range indicators that reflect the strength of the end effect.

[0010] In some embodiments, based on the evaluation object and quantitative indicators, the dynamic response at the tunnel end is calculated or historical data is obtained under different working conditions to construct a sample dataset containing input parameters and corresponding quantitative indicator results. A machine learning method is then used to train an intelligent calculation model for tunnel end effects, including: Based on different tunnel end conditions, structural parameters, and dynamic input conditions, the dynamic response of the tunnel end is calculated or historical data is obtained according to quantitative indicators for the evaluation object, forming a sample dataset containing input parameters and corresponding quantitative indicator results. The sample dataset comes from at least one of high-fidelity numerical simulation samples or measured monitoring samples. The construction of the sample dataset includes combined sampling of different surrounding rock parameters, burial depth, abrupt changes in structural parameters, dynamic input characteristics, and end boundary condition categories. The sample dataset is preprocessed to establish a mapping relationship between input parameters and quantitative index results. The mapping relationship is then trained based on machine learning methods to obtain an intelligent computing model for tunnel end effects. The preprocessing includes at least one of training data partitioning, feature scaling and normalization, categorical feature encoding, and time history data feature extraction. The machine learning methods include at least one of regression models or deep learning models.

[0011] In some embodiments, the input parameters of the tunnel to be evaluated are input into a trained intelligent computing model to obtain the quantified results of the dynamic response of the evaluation object under quantitative indicators, and the results are used for engineering analysis and optimization, including: The end parameters and dynamic parameters of the tunnel to be evaluated are input into the trained intelligent calculation model of tunnel end effect to obtain the quantitative results of dynamic response for the evaluation object. The results include the end effect amplification degree, end influence range, and dynamic response index of key end positions under the quantitative index. The amplification degree index is the end effect amplification coefficient, which is defined as the ratio of the response statistic at a set distance from the tunnel entrance in the end influence area to the corresponding statistic in the reference area. The influence range index is the end influence length, which is defined as the length of the longest continuous segment that satisfies the amplification coefficient greater than or equal to the set threshold. The obtained end effect calculation results are output and organized for the seismic and vibration resistance design evaluation of the tunnel end area, or for the comparative analysis and optimization of different end structural parameters and reinforcement schemes.

[0012] In some embodiments, obtaining the dynamic response quantization result includes preprocessing the input parameters in the same way as in the training phase, inputting them into the model for inference, and performing a reasonableness check and confidence assessment on the output result; the reasonableness check includes at least one of the following: response non-negativity constraint check, distance decay trend check, or extreme value threshold check; the confidence assessment uses the ensemble model variance or quantile regression method to give the prediction interval.

[0013] In some embodiments, the method further includes evaluating the trained intelligent calculation model of tunnel end effect, with evaluation indicators including at least one of the coefficient of determination, root mean square error, or mean absolute percentage error, and determining whether the model meets the engineering requirements based on a preset qualification criterion.

[0014] Secondly, the present invention also provides a smart computing device for tunnel end effects, the device comprising: The index determination module obtains the input parameters required for calculating the tunnel end effect, and determines the evaluation object and quantitative index based on the dynamic response characteristics of the tunnel end. The quantitative index is used to characterize the dynamic response level of the tunnel end area. The model building module is used to calculate the dynamic response of the tunnel end under different working conditions or obtain historical data based on the evaluation object and quantitative indicators. It constructs a sample dataset containing input parameters and corresponding quantitative indicator results, and uses machine learning methods to train an intelligent calculation model of tunnel end effect. The model calculation module is used to input the input parameters of the tunnel to be evaluated into the trained intelligent calculation model, obtain the dynamic response quantification results of the evaluation object under the quantitative index, and use the results for engineering analysis and optimization.

[0015] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the intelligent calculation method for tunnel end effect provided in the first aspect.

[0016] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the intelligent calculation method for tunnel end effect provided in the first aspect.

[0017] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the intelligent calculation method for tunnel end effects provided in the first aspect.

[0018] The beneficial effects of this invention are as follows: By constructing a quantitative index system for tunnel end effects, this invention unifies and structures the modeling of multi-dimensional data such as the amplification degree, influence range, and dynamic response at key locations of end effects. This transforms the dynamic response at tunnel ends from a single numerical output to a comprehensive calculation of multiple indices, ensuring good comparability and analyzability of calculation results under different working conditions and parameter conditions, thereby improving the standardization of data processing and the application value of calculation results. Simultaneously, this invention establishes a mapping model between input parameters and dynamic response based on machine learning methods. By training and calculating on multi-source data, in practical applications, only relevant parameters need to be input to quickly obtain prediction results, avoiding the repetitive calculation process of traditional high-fidelity numerical simulations. This significantly improves computational efficiency and data processing capabilities, reduces computational resource consumption, and enhances the applicability and stability of the model under multiple working conditions. Furthermore, based on the obtained dynamic response calculation results, the present invention performs further data processing and comprehensive analysis on the results to form evaluation index outputs for engineering applications. Based on the indicators, engineering scheme comparison, construction parameter optimization and risk control strategy generation are realized, so that the calculation results can directly serve the engineering decision-making process, improve the decision support capability of tunnel end effect analysis, and realize the effective connection between data calculation and engineering application.

[0019] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating an intelligent calculation method for tunnel end effects according to an embodiment of the present invention; Figure 2This is a schematic diagram of the structure of an intelligent computing device for tunnel end effects according to an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device structure provided in an embodiment of this application. Detailed Implementation

[0021] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] It should be noted that references to "an embodiment," "embodiment," "example embodiment," etc., in this specification refer to the described embodiment including specific features, structures, or characteristics; however, not every embodiment must include these specific features, structures, or characteristics. Furthermore, such expressions do not refer to the same embodiment. Moreover, when describing specific features, structures, or characteristics in conjunction with embodiments, whether or not explicitly described, it is indicated that incorporating such features, structures, or characteristics into other embodiments is within the knowledge of those skilled in the art.

[0023] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0024] In some embodiments, such as Figure 1 As shown, Figure 1 This is a flowchart illustrating an intelligent calculation method for tunnel end effects. The specific method includes: S101, obtain the input parameters required for calculating the tunnel end effect, and determine the evaluation object and quantitative index based on the dynamic response characteristics of the tunnel end.

[0025] Among them, quantitative indicators are used to characterize the dynamic response level of the tunnel end area.

[0026] Optionally, the input parameters required for calculating tunnel end effects are obtained, and the evaluation objects and quantitative indicators are determined based on the dynamic response characteristics of the tunnel ends. This includes: obtaining the engineering parameters and dynamic action parameters of the tunnel to be analyzed. The engineering parameters include at least the tunnel geometry, portal or end location, surrounding rock and support structure parameters, portal topographic parameters, damping parameters, and end boundary conditions. The end boundary condition parameters include the portal free surface conditions, portal constraint form, and end transition section structural abrupt change parameters. The dynamic action parameters include at least one of the following: seismic motion parameters, blasting vibration parameters, or train dynamic load parameters. The evaluation range and key locations of the tunnel end area are determined as the evaluation objects, and dynamic response quantitative indicators for characterizing the end effects are selected. The indicators include the dynamic response quantity at the key end locations, and the amplification degree or influence range indicators reflecting the strength of the end effects.

[0027] For example, input parameters include engineering parameters and dynamic parameters. Engineering parameters at least cover tunnel geometry, portal or end location, surrounding rock and support structure parameters, portal section topographic parameters, damping parameters, and end boundary conditions. Specifically: tunnel geometry includes tunnel diameter or cross-sectional shape (e.g., circular, horseshoe-shaped), excavation width and height, and the inner and outer contours and thickness of the lining. Portal or end location refers to information such as tunnel portal mileage coordinates and burial depth. Surrounding rock and support structure parameters include surrounding rock grade or equivalent rock mass parameters (e.g., density, elastic modulus, Poisson's ratio, cohesion, internal friction angle, tensile strength), surrounding rock stratification information (layer thickness, layer interface), material parameters of initial support and secondary lining (elastic modulus, Poisson's ratio, density, strength), equivalent parameters of reinforcing steel or steel arches, and the thickness and constitutive parameters of the damping layer or vibration isolation layer. Portal section topographic parameters include the topographic slope and elevation of the portal section. Damping parameters, such as the Rayleigh damping coefficient, are used to describe the energy dissipation characteristics of the surrounding rock and structure. End boundary condition parameters are a set of parameters that describe the key control factors of end effects. Specifically, they include: free surface conditions at the tunnel entrance: the existence and geometry of the free surface at the tunnel entrance; constraint forms at the tunnel entrance: such as whether a tunnel portal, open tunnel, or slope constraint is set; and structural abrupt change parameters in the end transition section: such as abrupt changes in lining thickness or surrounding rock grade.

[0028] Dynamic parameters include at least one of seismic motion parameters, blasting vibration parameters, or train dynamic load parameters, and can be provided in time history form or characteristic quantity form. Seismic motion parameters may include acceleration / velocity / pressure time histories, peak ground acceleration (PGA), peak ground velocity (PGV), peak ground displacement (PGD), dominant frequency, response spectrum characteristic values, energy index, duration, etc. Blasting vibration parameters may include charge amount, minimum resistance line, detonation distance, delay mode, etc. Train dynamic load parameters may include axle load, train speed, train formation, track irregularity index, etc.

[0029] All parameters must be standardized in units (e.g., meters, seconds, Pascals) and coordinate systems (usually x for the tunnel axis, r for the radial direction, and θ for the circumferential direction). Outliers or missing values ​​should be handled according to preset rules (e.g., removal, interpolation, or backfilling with default values). Discrete category parameters (e.g., surrounding rock grade, support type) must be numerically encoded (e.g., unique thermal encoding or sequence encoding). End boundary conditions can be encoded using enumerated values ​​(e.g., free surface = 1, non-free surface = 0).

[0030] Based on the dynamic response characteristics at the tunnel end, the evaluation object and quantitative indicators are determined. The evaluation object refers to the specific evaluation range and key locations of the tunnel end area, while the quantitative indicators refer to measurable parameters used to characterize the dynamic response level at the end.

[0031] End evaluation range: Starting from the end of the tunnel entrance, a certain length is taken along the tunnel axis as the end influence area. This length is usually a multiple of the tunnel diameter or excavation width (e.g., 2 to 5 times), or it can be set according to engineering experience or specification requirements. At the same time, a middle section far from the end, where the boundary conditions tend to be stable, is selected as a reference area for comparative calculation.

[0032] Key locations: Within the influence area at the ends, select typical cross-sections (such as the tunnel entrance cross-section, cross-sections at 0.5 times the tunnel diameter, 1 time the tunnel diameter, 2 times the tunnel diameter, etc.) and typical parts (such as the arch crown, arch waist, sidewalls, and invert) as response evaluation points. If surrounding rock damage is considered, a surrounding rock ring domain or a set of grid cells can also be defined. The above evaluation scope and key locations together constitute the evaluation object.

[0033] Quantitative indicators include two categories: Basic dynamic response quantities: such as peak values ​​or statistics (e.g., root mean square, 95th percentile) of displacement, velocity, acceleration, stress, internal forces, etc., at key locations. End effect specific indicators: End effect amplification factor: reflects the degree of amplification of the dynamic response at the end, defined as the ratio of the response statistics at a certain distance from the tunnel entrance within the end's influence area to the corresponding statistics in the reference area. End influence range: reflects the length of the end effect along the tunnel axis, defined as the longest continuous segment length that satisfies an amplification factor greater than or equal to a set threshold (e.g., 1.05, 1.10), or the position where the value first returns to below the threshold.

[0034] The evaluation objects and quantitative indicators identified above will serve as the core basis for subsequent sample construction and model output.

[0035] S102, based on the evaluation object and quantitative indicators, calculates or obtains historical data on the dynamic response of the tunnel end under different working conditions, constructs a sample dataset containing input parameters and corresponding quantitative indicator results, and uses machine learning methods to train an intelligent calculation model of tunnel end effect.

[0036] Optionally, based on the evaluation object and quantitative indicators, the dynamic response at the tunnel end is calculated or historical data is obtained under different working conditions to construct a sample dataset containing input parameters and corresponding quantitative indicator results. A machine learning method is then used to train an intelligent calculation model for tunnel end effects. This includes: calculating the dynamic response at the tunnel end according to quantitative indicators or obtaining historical data based on different tunnel end conditions, structural parameters, and dynamic input working conditions, forming a sample dataset containing input parameters and corresponding quantitative indicator results; the sample dataset originates from at least one of high-fidelity numerical simulation samples or measured monitoring samples; the construction of the sample dataset includes combined sampling of different surrounding rock parameters, burial depth, structural parameter mutations, dynamic input characteristics, and end boundary condition categories; preprocessing the sample dataset to establish a mapping relationship between input parameters and quantitative indicator results, and training this mapping relationship based on machine learning methods to obtain an intelligent calculation model for tunnel end effects; preprocessing includes at least one of training data partitioning, feature scaling and normalization, category feature encoding, and time history data feature extraction; the machine learning method includes at least one of regression models or deep learning models.

[0037] For example, sample data can be derived from high-fidelity numerical simulations, field monitoring, or historical engineering data. During the construction process, data collection or calculation should be performed on the evaluation object (i.e., key locations within the end influence area) according to quantitative indicators (such as peak displacement, amplification factor, influence length, etc.).

[0038] High-fidelity numerical simulation: Dynamic time history analysis is performed using the finite element method, finite difference method, or discrete element method. A high-precision model is established that includes stratigraphic stratification, structure-surrounding rock contact, nonlinear constitutive model, boundary conditions, and dynamic input. For each working condition, the time history and statistical indicators of key points in the end region are output, and special indicators such as end effect amplification factor and influence length are calculated.

[0039] Actual monitoring: Sensors (such as accelerometers, strain gauges, and displacement gauges) are deployed at the tunnel entrance section. Combined with seismic records, blasting parameters, or train operation parameters, actual response data is obtained and processed to obtain corresponding quantitative indicators.

[0040] Historical engineering data: Collect calculation reports or monitoring data from existing projects, and incorporate them into the sample set after standardization.

[0041] Sample construction needs to be combined with sampling around the key factors of end effects, that is, to cover a variety of working conditions through experimental design: changes in surrounding rock parameters (different surrounding rock grades, mechanical parameters); changes in burial depth; changes in portal topography (different slopes, slope heights); abrupt changes in structural parameters (changes in lining thickness, whether or not a damping layer is installed); changes in dynamic input characteristics (seismic intensity and spectrum, blasting parameters, train speed / axle load); and types of end boundary conditions (free surface, open tunnel, portal constraint, etc.).

[0042] For each operating condition, tagged data (output results) is generated and stored, including the basic response at key locations, the end effect amplification factor curve, the end effect length, and possible damage indicators, plastic zone area, etc. All outputs correspond to the quantitative indicators defined in the first stage.

[0043] The sample dataset is preprocessed to establish a mapping relationship between input parameters and quantitative index results, and an intelligent computing model is trained based on machine learning methods.

[0044] Data preprocessing includes: dividing the samples into training, validation, and test sets (e.g., 7:2:1) to ensure that different tunnels or different seismic waves do not all fall into the same subset, thus avoiding information leakage; feature scaling and normalization (e.g., min-max normalization or Z-score standardization); categorical feature encoding (e.g., one-hot encoding or embedding encoding); and time history data feature extraction: if the input contains time histories, its statistical features (e.g., PGA, dominant frequency, energy) can be extracted as model input.

[0045] Machine learning methods can employ at least one of regression models or deep learning models: regression models include support vector regression (SVR), random forest, gradient boosting tree (GBDT / XGBoost / LightGBM); deep learning models include multilayer perceptron (MLP), convolutional neural network (CNN, suitable for processing spectral or image features), long short-term memory network (LSTM), or transformer (suitable for time series data).

[0046] A multi-output regression architecture can be adopted to simultaneously output the response and end-effect indices of multiple key points.

[0047] The training process requires defining a loss function (such as mean squared error (MSE) or mean absolute error (MAE)) and optimizing hyperparameters using grid search, Bayesian optimization, or genetic algorithms. Early stopping is employed to prevent overfitting. The final result is an intelligent computational model for tunnel end effects, which can directly output corresponding quantitative index results based on the input parameters.

[0048] S103: Input the input parameters of the tunnel to be evaluated into the trained intelligent computing model to obtain the dynamic response quantification results of the evaluation object under the quantitative index, and use the results for engineering analysis and optimization.

[0049] Optionally, the input parameters of the tunnel to be evaluated are input into the trained intelligent computing model to obtain the dynamic response quantification results of the evaluation object under the quantitative indicators, and the results are used for engineering analysis and optimization. This includes: inputting the end parameters and dynamic action parameters of the tunnel to be evaluated into the trained intelligent computing model of tunnel end effects to obtain the dynamic response quantification results for the evaluation object. The results include the end effect amplification degree, end influence range, and dynamic response indicators at key end locations under the quantitative indicators; the amplification degree indicator is the end effect amplification coefficient, defined as the ratio of the response statistics at a set distance from the tunnel entrance within the end influence area to the corresponding statistics in the reference area; the influence range indicator is the end influence length, defined as the length of the longest continuous segment that satisfies the amplification coefficient greater than or equal to a set threshold; the obtained end effect calculation results are output and organized for the seismic and vibration resistance design evaluation of the tunnel end area, or for the comparative analysis and optimization selection of different end structural parameters and reinforcement schemes.

[0050] For example, firstly, the input parameters of the tunnel to be evaluated undergo the same preprocessing as in the training phase (unit unification, encoding, normalization) to ensure consistent scaling. If some parameters are missing, they can be supplemented according to preset rules (such as the average of similar projects) or an error can be reported.

[0051] The preprocessed input vector is fed into the trained model, which quickly outputs the following quantitative metrics for the evaluation object: The basic dynamic response indicators at key end locations (such as peak values ​​or statistics of displacement, acceleration, stress, and internal forces); the degree of amplification of end effects (i.e., the distribution of amplification coefficient along the tunnel axis); and the range of influence of the end (the length of influence).

[0052] Perform a rationality check and confidence assessment on the model output: The rationality check includes: non-negative constraint of response quantity (such as displacement and stress should not be negative); check for attenuation trend with distance (the amplification factor should tend to 1 as the distance from the opening increases); extreme value threshold check (to determine whether it exceeds the physical possibility range).

[0053] Confidence assessment can use integrated model variance (such as the standard deviation of multiple model outputs) or quantile regression methods to give prediction intervals, which facilitates engineering risk assessment.

[0054] The obtained end-effect calculation results are output and processed for the following engineering analysis and optimization: Seismic and vibration resistance design assessment: Based on the end amplification factor and the range of influence, determine whether the opening section needs to be reinforced with seismic measures, or assess whether the existing design meets safety requirements.

[0055] Scheme comparison and optimization: For different end structure parameters (such as lining thickness, damping layer setting) or reinforcement schemes (such as anchor reinforcement, steel arch reinforcement), input the corresponding parameters for batch calculation, quickly obtain the quantitative index of end effect under each scheme, and select the optimal scheme through comparative analysis to provide a basis for engineering design.

[0056] Optionally, the above method also includes evaluating the trained intelligent calculation model of tunnel end effect, with evaluation indicators including at least one of the coefficient of determination, root mean square error, or mean absolute percentage error, and judging whether the model meets the engineering requirements according to the preset qualification criteria.

[0057] Specifically, after model training is completed, the intelligent calculation model for tunnel end effects should be evaluated to ensure it meets engineering application requirements. The evaluation process includes assessing the model's prediction accuracy on a test set using metrics such as the coefficient of determination (R²), root mean square error (RMSE), or mean absolute percentage error (MAPE). For end effect-specific metrics (such as amplification factor and influence length), their relative or interval errors can be calculated. The model's usability is determined based on preset acceptance criteria, such as key response errors not exceeding the engineering allowable range (e.g., 10%) and end influence length deviations less than a set threshold (e.g., 0.5 times the tunnel diameter). If the model fails to meet the criteria, the samples or model parameters need to be adjusted and the model retrained.

[0058] This embodiment uses "evaluation object and quantitative indicators" as the core thread running through the entire process of data preparation, sample construction, model training, and engineering application, forming a rigorous logical closed loop and ensuring the goal orientation and consistency of end-effect calculation. This method specifically defines a quantitative indicator system for the tunnel end region, including the end-effect amplification factor, influence range, and dynamic response at key locations. This achieves a substantial shift from qualitative description to quantitative calculation of end-effects, making the calculation results directly comparable and engineering interpretable. Simultaneously, by constructing a machine learning-based intelligent computing model, the calculation time required for traditional numerical simulations (which can take hours or even days) is compressed to seconds while maintaining computational accuracy. This significantly improves the efficiency of multi-condition batch evaluation and greatly reduces computational costs. Furthermore, the quantitative end-effect results output by this method can directly serve seismic and vibration-resistant design evaluation and reinforcement scheme optimization and selection, providing specific and reliable quantitative basis for engineering decisions, demonstrating outstanding technical effects and broad engineering application prospects.

[0059] Based on the same inventive concept, this application also provides a tunnel end effect intelligent computing device for implementing the tunnel end effect intelligent computing method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more tunnel end effect intelligent computing device embodiments provided below can be found in the limitations of the tunnel end effect intelligent computing method described above, and will not be repeated here.

[0060] In one embodiment, such as Figure 2 As shown, a smart computing device for tunnel end effects is provided, the device comprising: The index determination module 30 obtains the input parameters required for calculating the tunnel end effect, and determines the evaluation object and quantitative index based on the dynamic response characteristics of the tunnel end. The quantitative index is used to characterize the dynamic response level of the tunnel end area. The model building module 31 is used to calculate or obtain historical data on the dynamic response of the tunnel end under different working conditions based on the evaluation object and quantitative indicators, build a sample dataset containing input parameters and corresponding quantitative indicator results, and use machine learning methods to train an intelligent calculation model of tunnel end effect. The model calculation module 32 is used to input the input parameters of the tunnel to be evaluated into the trained intelligent calculation model, obtain the dynamic response quantification results of the evaluation object under the quantitative index, and use the results for engineering analysis and optimization.

[0061] This application also provides an electronic device, in some embodiments, referring to... Figure 3 As shown, the electronic device 700 includes an input unit 710, a memory 720, a processor 730, and an output unit 740. The memory 720 stores program instructions that can be executed on the processor 730. The processor 730 can execute the tunnel end effect intelligent calculation method and / or technical solution based on the aforementioned embodiments by calling the program instructions. The electronic device 700 can be a mobile terminal device such as a mobile phone or a computer.

[0062] Furthermore, embodiments of this application also provide a computer-readable storage medium for storing a computer program that performs the intelligent calculation method for tunnel end effects. For example, computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to this application through the operation of the computer. The program instructions that invoke the methods of this application may be stored in a fixed or removable storage medium, and / or transmitted via data streams in broadcast or other signal carrying media, and / or stored in a storage medium that operates according to the program instructions.

[0063] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0064] The technical features of the above embodiments can be arbitrarily integrated. For the sake of brevity, not all possible integrations of the technical features in the above embodiments are described. However, as long as the integration of these technical features does not contradict each other, they should be considered to be within the scope of this specification.

[0065] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A smart calculation method for tunnel end effects, characterized in that, The method includes: The input parameters required for calculating tunnel end effects are obtained, and the evaluation object and quantitative index are determined based on the dynamic response characteristics of the tunnel end; the quantitative index is used to characterize the dynamic response level of the tunnel end region. The process of obtaining the input parameters required for calculating tunnel end effects and determining the evaluation object and quantitative indicators based on the dynamic response characteristics of the tunnel end includes: obtaining the engineering parameters and dynamic action parameters of the tunnel to be analyzed; the engineering parameters include at least the tunnel geometry, portal or end location, surrounding rock and support structure parameters, portal topographic parameters, damping parameters, and end boundary conditions; the end boundary condition parameters include portal free surface conditions, portal constraint form, and end transition section structural abrupt change parameters; the dynamic action parameters include at least one of seismic motion parameters, blasting vibration parameters, or train dynamic load parameters; determining the evaluation range and key locations of the tunnel end area as the evaluation object, and selecting dynamic response quantitative indicators to characterize the end effects; the indicators include the dynamic response quantity at the key end locations, and indicators reflecting the amplification degree or influence range of the end effects. Based on the evaluation object and quantitative indicators, the dynamic response of the tunnel end is calculated or historical data is obtained under different working conditions. A sample dataset containing input parameters and corresponding quantitative indicator results is constructed, and a machine learning method is used to train an intelligent calculation model of tunnel end effect. The input parameters of the tunnel to be evaluated are input into the trained intelligent computing model to obtain the dynamic response quantification results of the evaluation object under the quantification index, and the results are used for engineering analysis and optimization.

2. The intelligent calculation method for tunnel end effects as described in claim 1, characterized in that, Based on the evaluation object and quantitative indicators, the dynamic response at the tunnel end is calculated or historical data is obtained under different working conditions. A sample dataset containing input parameters and corresponding quantitative indicator results is constructed, and a machine learning method is used to train an intelligent calculation model for tunnel end effects, including: Based on different tunnel end conditions, structural parameters, and dynamic input conditions, the dynamic response of the tunnel end is calculated or historical data is obtained according to the quantitative indicators for the evaluation object, forming a sample dataset containing input parameters and corresponding quantitative indicator results; the sample dataset comes from at least one of high-fidelity numerical simulation samples or measured monitoring samples; the construction of the sample dataset includes combined sampling of different surrounding rock parameters, burial depth, abrupt changes in structural parameters, dynamic input characteristics, and end boundary condition categories. The sample dataset is preprocessed to establish a mapping relationship between the input parameters and the quantitative index results, and the mapping relationship is trained based on a machine learning method to obtain an intelligent computing model for tunnel end effect; the preprocessing includes at least one of training data partitioning, feature scaling and normalization, categorical feature encoding, and time history data feature extraction; the machine learning method includes at least one of regression model or deep learning model.

3. The intelligent calculation method for tunnel end effects as described in claim 2, characterized in that, The input parameters of the tunnel to be evaluated are input into the trained intelligent computing model to obtain the dynamic response quantification results of the evaluation object under the quantification index, and the results are used for engineering analysis and optimization, including: The end parameters and dynamic parameters of the tunnel to be evaluated are input into the trained intelligent calculation model of tunnel end effect to obtain the dynamic response quantification results for the evaluation object. The results include the end effect amplification degree, end influence range, and dynamic response indicators of key end positions under the quantification index. The amplification degree index is the end effect amplification coefficient, defined as the ratio of the response statistic at a set distance from the tunnel entrance within the end influence area to the corresponding statistic in the reference area. The influence range index is the end influence length, defined as the length of the longest continuous segment that satisfies the amplification coefficient greater than or equal to a set threshold. The obtained end effect calculation results are output and organized for the seismic and vibration resistance design evaluation of the tunnel end area, or for the comparative analysis and optimization of different end structural parameters and reinforcement schemes.

4. The intelligent calculation method for tunnel end effects as described in claim 3, characterized in that, Obtaining the dynamic response quantification result for the evaluation object includes: preprocessing the input parameters in the same way as in the training phase, inputting them into the model for inference, and performing a reasonableness check and confidence assessment on the output result; the reasonableness check includes at least one of the following: response non-negativity constraint check, distance decay trend check, or extreme value threshold check; the confidence assessment uses the ensemble model variance or quantile regression method to give the prediction interval.

5. The intelligent calculation method for tunnel end effects as described in any one of claims 1-4, characterized in that, The method also includes evaluating the trained intelligent calculation model of tunnel end effect, with evaluation indicators including at least one of the coefficient of determination, root mean square error, or mean absolute percentage error, and judging whether the model meets the engineering requirements based on preset qualification criteria.

6. A smart computing device for tunnel end effects, characterized in that, The device includes: The index determination module acquires the input parameters required for calculating tunnel end effects and determines the evaluation object and quantitative index based on the dynamic response characteristics of the tunnel end; the quantitative index is used to characterize the dynamic response level of the tunnel end region. The process of obtaining the input parameters required for calculating tunnel end effects and determining the evaluation object and quantitative indicators based on the dynamic response characteristics of the tunnel end includes: obtaining the engineering parameters and dynamic action parameters of the tunnel to be analyzed; the engineering parameters include at least the tunnel geometry, portal or end location, surrounding rock and support structure parameters, portal topographic parameters, damping parameters, and end boundary conditions; the end boundary condition parameters include portal free surface conditions, portal constraint form, and end transition section structural abrupt change parameters; the dynamic action parameters include at least one of seismic motion parameters, blasting vibration parameters, or train dynamic load parameters; determining the evaluation range and key locations of the tunnel end area as the evaluation object, and selecting dynamic response quantitative indicators to characterize the end effects; the indicators include the dynamic response quantity at the key end locations, and indicators reflecting the amplification degree or influence range of the end effects. The model building module is used to calculate the dynamic response of the tunnel end under different working conditions or obtain historical data based on the evaluation object and quantitative indicators, construct a sample dataset containing input parameters and corresponding quantitative indicator results, and use machine learning methods to train an intelligent calculation model of tunnel end effect. The model calculation module is used to input the input parameters of the tunnel to be evaluated into the trained intelligent calculation model, obtain the dynamic response quantification results of the evaluation object under the quantification index, and use the results for engineering analysis and optimization.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent calculation method for tunnel end effect as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the intelligent calculation method for tunnel end effects as described in any one of claims 1 to 5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent calculation method for tunnel end effect as described in any one of claims 1 to 5.

Images