A method, system, device and storage medium for deformation early warning of shield crossing embankment

Abstract

The application provides a shield crossing embankment deformation early warning method, system, device and storage medium, and relates to the technical field of tunnel and underground engineering monitoring and prediction. The method clearly takes the incision pressure, tunneling speed, synchronous grouting pressure and synchronous grouting volume of shield tunneling as the BP neural network input layer parameters, takes the embankment settlement value as the output layer parameter, trains, verifies and optimizes the BP neural network through engineering measured data, and constructs a high-precision settlement prediction model. The model is used to predict the embankment settlement value in real time, and the early warning threshold is dynamically set in combination with the shield tunneling stage division, so that the advanced prediction and graded early warning of the embankment settlement are realized. The application realizes the deep coupling of construction parameters and settlement prediction, has high prediction accuracy and fast early warning response, can accurately guide the adjustment of shield construction parameters, effectively controls the embankment settlement deformation, and guarantees the safety of shield crossing embankment construction.

Description

Technical Field

[0001] This invention relates to the field of tunnel and underground engineering monitoring and prediction technology, and in particular to a method, system, equipment and storage medium for early warning of deformation when a shield tunnel crosses a dam. Background Technology

[0002] With the rapid development of transportation infrastructure, large-diameter slurry balance shield tunneling machines (SMTs) are widely used in tunnel construction due to their high construction efficiency and relatively controllable ground disturbance. In some tunnel construction projects, the tunnels need to cross rivers and lakes. As core water conservancy facilities for flood control and drainage, the structural stability of the embankment directly affects the safety of people's lives and property within the basin. During the tunneling process, the shield tunneling inevitably disturbs the surrounding ground, causing deformation problems such as settlement and horizontal displacement of the embankment. If the deformation exceeds the safety threshold, it can easily lead to cracks, water leakage, or even structural instability, causing serious water conservancy safety accidents. Therefore, deformation monitoring and accurate early warning during the shield tunneling process across embankments have become core technical points in engineering construction.

[0003] In existing technologies, surface deformation monitoring and early warning for shield tunneling mainly involves setting up settlement monitoring points in the monitoring area to obtain the settlement of the embankment, setting a fixed early warning threshold based on engineering experience, and issuing an early warning signal when the monitoring data exceeds the threshold. At the same time, measures such as adjusting shield tunneling parameters and strengthening grouting reinforcement are used to control the development of deformation.

[0004] Existing methods for predicting embankment settlement are limited and have certain limitations. For example, while a settlement prediction model for shield tunneling through an embankment can be established considering shield disturbance and seepage, it requires experimental determination of key parameters and exhibits a certain time lag. Some technologies attempt to use backpropagation neural networks for settlement prediction, but they do not specifically select core parameters as inputs and do not set warning thresholds, resulting in problems such as warning lag, fixed thresholds, and a disconnect between prediction and construction parameters. Summary of the Invention

[0005] Purpose of the invention: To propose a method, system, equipment and storage medium for early warning of deformation when shield tunneling crosses a levee, to solve the technical problems of existing levee settlement monitoring mainly relying on manual on-site monitoring data with lag and the prediction method not being deeply integrated with construction parameters, to achieve deep coupling between construction parameters and settlement prediction, and to improve the accuracy and timeliness of levee settlement prediction and early warning.

[0006] To achieve the above objectives, this invention proposes a method for early warning of deformation when a shield tunnel crosses a levee, comprising the following steps:

[0007] S1. Collect settlement data from monitoring points on the embankment and shield tunneling parameter data during the shield tunneling construction process. Preprocess the settlement data and the shield tunneling parameter data to obtain a preprocessed dataset. Divide the preprocessed dataset into a training set, a validation set, and a test set.

[0008] S2. Construct a BP neural network settlement prediction model based on the training set. The BP neural network settlement prediction model includes an input layer, a hidden layer, and an output layer. The input layer is used to input shield tunneling parameters, and the output layer is used to output the predicted settlement value of the embankment.

[0009] S3. Input the training set into the BP neural network settlement prediction model for model training, and use the validation set to validate the trained BP neural network settlement prediction model to obtain the optimal settlement prediction model.

[0010] S4. Input the test set into the optimal settlement prediction model for model testing and accuracy evaluation. When the optimal settlement prediction model meets the preset accuracy requirements, determine the optimal settlement prediction model as the target prediction model.

[0011] S5. During the shield tunneling construction, real-time data of shield tunneling parameters are collected, and the real-time data of shield tunneling parameters are preprocessed and then input into the target prediction model to obtain the real-time predicted value of embankment settlement.

[0012] S6. Set corresponding settlement warning thresholds based on different tunneling stages, and compare the real-time predicted settlement value of the embankment with the settlement warning threshold of the corresponding tunneling stage to trigger the corresponding warning level.

[0013] S7. Feed back the construction adjustment parameters after the early warning is triggered and the actual embankment settlement monitoring values ​​to the dynamic update database of engineering data, and iteratively train the target prediction model based on the newly added actual embankment settlement monitoring values ​​to update the model parameters.

[0014] As a preferred embodiment, the BP neural network settlement prediction model has 4 nodes in the input layer, which correspond to the cutting pressure, tunneling speed, synchronous grouting pressure, and synchronous grouting volume, respectively.

[0015] The output layer has 1 node, corresponding to the levee settlement value;

[0016] The number of nodes in the hidden layer was determined using a trial-and-error method.

[0017] As a preferred embodiment, the hidden layer uses the Sigmoid function as the activation function, and the output layer uses the linear Purelin function as the activation function.

[0018] The BP neural network settlement prediction model is trained using the LM optimization algorithm.

[0019] As a preferred embodiment, in step S3, the training set is input into the BP neural network settlement prediction model for model training, specifically including:

[0020] The data input from the input layer is forward-propagated to obtain the predicted value of the levee settlement;

[0021] Calculate the mean square error between the predicted and measured values ​​of the embankment settlement, and then backpropagate the error.

[0022] Adjust network weights and thresholds based on backpropagation results;

[0023] Repeat the forward and backward propagation processes until the training error reaches the preset accuracy requirement or the maximum number of training iterations is reached.

[0024] As a preferred embodiment, in step S3, the prediction error of the BP neural network settlement prediction model is calculated using the validation set;

[0025] When the prediction accuracy does not meet the preset standard, the number of hidden layer nodes and the learning rate are adjusted, and the BP neural network settlement prediction model is retrained.

[0026] As a preferred option, in step S4, the mean squared error (MSE) and the coefficient of determination (R²) are used. 2 As an accuracy evaluation index for the target prediction model;

[0027] When the determination coefficient R 2 If the mean square error (MSE) is greater than or equal to 0.9 and less than or equal to 0.5, the target prediction model is determined to meet the preset accuracy requirements.

[0028] As a preferred embodiment, in step S6, the settlement early warning threshold includes a first-level early warning threshold, a second-level early warning threshold, and a third-level early warning threshold;

[0029] Among them, the first-level warning threshold, the second-level warning threshold, and the third-level warning threshold for the tail settlement stage are 60%, 80%, and 100% of the allowable settlement value of the embankment, respectively.

[0030] As a preferred embodiment, in step S6, the real-time predicted value of the embankment settlement is compared with the settlement early warning threshold of the corresponding tunneling stage to trigger the corresponding early warning level, specifically including:

[0031] When the real-time predicted value of the embankment settlement is less than or equal to the first-level early warning threshold, the current shield tunneling parameters are maintained.

[0032] When the real-time predicted value of the embankment settlement is greater than the first-level early warning threshold and less than or equal to the second-level early warning threshold, an early warning is issued and the shield tunneling parameters are adjusted.

[0033] When the real-time predicted value of the embankment settlement is greater than the secondary warning threshold and less than or equal to the tertiary warning threshold, an audible and visual alarm is issued and the shield tunneling parameters are adjusted.

[0034] When the real-time predicted value of the embankment settlement exceeds the level 3 early warning threshold, an emergency alarm is issued, and the tunnel boring machine is stopped and grouting reinforcement measures are taken.

[0035] As a preferred option, in step S7, according to the progress of the tunnel boring machine, every 5 to 10 rings, the newly added actual embankment settlement monitoring values ​​are added to the training set, and the target prediction model is iteratively trained to readjust the network weights and thresholds.

[0036] Furthermore, this invention also proposes a shield tunneling deformation early warning system for crossing a levee, the system comprising:

[0037] The data acquisition module is used to collect settlement data from monitoring points on the embankment and shield tunneling parameter data during the shield tunneling construction process. The settlement data and the shield tunneling parameter data are preprocessed to obtain a preprocessed dataset, and the preprocessed dataset is divided into a training set, a validation set, and a test set.

[0038] The BP neural network settlement prediction model is constructed based on the training set. The BP neural network settlement prediction model includes an input layer, a hidden layer, and an output layer. The input layer is used to input shield tunneling parameters, and the output layer is used to output the predicted settlement value of the embankment.

[0039] The training module is used to input the training set into the BP neural network settlement prediction model for model training, and to use the validation set to validate the trained BP neural network settlement prediction model to obtain the optimal settlement prediction model.

[0040] The model selection module is used to input the test set into the optimal settlement prediction model for model testing and accuracy evaluation. When the optimal settlement prediction model meets the preset accuracy requirements, the optimal settlement prediction model is determined as the target prediction model.

[0041] The prediction output module is used to collect real-time data of shield tunneling parameters during the shield tunneling construction process, and input the preprocessed real-time data of shield tunneling parameters into the target prediction model to obtain the real-time prediction value of embankment settlement.

[0042] The early warning triggering module sets corresponding settlement early warning thresholds based on different tunneling stages, compares the real-time predicted settlement value of the embankment with the settlement early warning threshold of the corresponding tunneling stage, and triggers the corresponding early warning level.

[0043] The module iteration module is used to feed back the construction adjustment parameters after the early warning is triggered and the actual embankment settlement monitoring values ​​to the dynamic update database of engineering data, and to iteratively train the target prediction model based on the newly added actual embankment settlement monitoring values ​​in order to update the model parameters.

[0044] Furthermore, the present invention also proposes an electronic device, which includes a processor, a memory, a communication interface, and a communication bus. The processor, the memory, and the communication interface communicate with each other through the communication bus. The memory is used to store a number of executable instructions, which cause the processor to execute the aforementioned method for predicting the deformation of foundation pit maintenance structures based on fractal theory.

[0045] Furthermore, the present invention also proposes a computer-readable storage medium storing a plurality of executable instructions, which, when executed on an electronic device, cause the electronic device to execute the aforementioned method for predicting the deformation of foundation pit maintenance structures based on fractal theory.

[0046] Compared with the prior art, the present invention has at least the following beneficial effects:

[0047] (1) This invention uses the cutting pressure, tunneling speed, synchronous grouting pressure, and synchronous grouting volume of the shield tunneling as the core parameters of the input layer of the BP neural network, accurately capturing the key construction factors affecting the settlement of the embankment, fitting the complex nonlinear relationship between tunneling parameters and settlement values, and significantly improving the prediction accuracy. The coefficient of determination R 2 With a value ≥0.9, accurate and advanced prediction of levee settlement can be achieved.

[0048] (2) This invention constructs a closed-loop settlement prediction and early warning system that integrates data acquisition, model prediction, dynamic early warning, construction adjustment, and feedback iteration. It realizes the integrated operation of real-time acquisition of construction parameters, advance prediction of settlement, graded early warning triggering, and dynamic adjustment of construction parameters. The early warning response is fast and can issue an early warning in time before the settlement of the dike exceeds the safe range, so as to make the best time for construction adjustment and effectively control the settlement and deformation of the dike.

[0049] (3) The present invention establishes a model iterative update mechanism. By continuously supplementing the engineering measured data, the BP neural network model is iteratively trained, so that the model can adapt to the dynamic changes of construction conditions and geological conditions, maintain high prediction accuracy, and improve the engineering adaptability and practicality of the method.

[0050] (4) The prediction and early warning method of the present invention is simple to operate and easy to implement. It can be seamlessly connected with the existing shield construction monitoring system, collect construction parameters in real time and output prediction and early warning results, and at the same time provide targeted construction adjustment suggestions, providing intuitive and accurate construction guidance for engineering technicians, effectively ensuring the safety of large-diameter slurry shield tunneling through the dike and reducing the risk of dike instability. Attached Figure Description

[0051] Figure 1 This is a flowchart of shield tunneling construction data processing and data group division according to the present invention.

[0052] Figure 2 This is a schematic diagram illustrating the construction process of the BP neural network embankment settlement / uplift deformation prediction model of the present invention.

[0053] Figure 3 This is a schematic diagram of the dam deformation prediction process of the present invention.

[0054] Figure 4 This is a flowchart of the three-level early warning process for dike deformation in this invention. Detailed Implementation

[0055] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0056] This embodiment proposes a method for early warning of deformation when a shield tunnel crosses a levee. The specific scheme is as follows:

[0057] S1. Data Acquisition and Processing. Using settlement data collected from existing embankment monitoring points, the average settlement value y (unit: mm) for each tunneling cycle is obtained based on the collection time period; measured data of core tunneling parameters X=[x] are collected during the construction of the large-diameter slurry shield tunnel across the embankment. 1, x 2, x 3, x4] T This includes cutting pressure x1 (unit: bar), tunneling speed x2 (unit: mm / min), synchronous grouting pressure x3 (unit: bar), and synchronous grouting volume x4 (unit: m³ / ring). The collected raw data underwent data cleaning and normalization: outliers were identified and removed using the 3σ criterion; missing values ​​were filled using nearest neighbor interpolation to ensure data integrity; and all data were normalized to the interval [-1, 1] using the maximum-minimum standardization method, with the formula: , The data are normalized, where x represents the original data for cutting pressure, tunneling speed, synchronous grouting pressure, and synchronous grouting volume, respectively. The maximum value of the original data, This is the minimum value of the original data.

[0058] The preprocessed dataset was randomly divided into training, validation, and test sets in a ratio of 7:2:1, which were used for training, parameter optimization, and accuracy evaluation of the BP neural network, respectively.

[0059] S2. Construct a BP neural network settlement prediction model. The BP neural network is configured as a three-layer feedforward neural network with an input layer, hidden layers, and an output layer. The core parameters are designed as follows:

[0060] Input layer: There are 4 nodes, which correspond to the four core shield tunneling parameters: cutting pressure, tunneling speed, synchronous grouting pressure, and synchronous grouting volume, providing basic data on construction conditions for the model;

[0061] Output layer: 1 node, corresponding to the levee settlement value, which is the model's prediction target;

[0062] Hidden layer: The number of nodes is optimized based on trial and error, with the number of nodes ranging from 6 to 12. The number of nodes with the highest prediction accuracy on the validation set is selected as the optimal number of nodes in the hidden layer, thereby achieving the fitting of the complex nonlinear relationship between the input parameters and the settlement value.

[0063] Activation function: The hidden layer uses the Sigmoid function to achieve a non-linear mapping of parameters, as shown in the formula. The output layer uses a linear Purelin function to ensure the continuity and linear output of the settlement prediction results.

[0064] Training algorithm and basic parameters: The LM optimization algorithm is used for model training to improve training efficiency and convergence speed; the learning rate is set to 0.001~0.01, the number of training iterations is set to 1000~5000, and the initial values ​​of weights and thresholds are assigned random values ​​in the range of [-1,1].

[0065] Training and Validation of S3 and BP Neural Network Models

[0066] Input the training set from step S1 into the constructed BP neural network model to train the model:

[0067] Forward propagation: Input layer data is passed to the hidden layer via a weight matrix, processed by the activation function of the hidden layer, and then passed to the output layer to obtain the predicted value of levee settlement;

[0068] Backpropagation: Calculate the mean square error (MSE) between the predicted and measured values, propagate the error back from the output layer to the hidden and input layers, and continuously adjust the weights and thresholds of the network using gradient descent to reduce the prediction error;

[0069] Model convergence: Repeat the forward and backward propagation process until the training error of the model reaches the preset accuracy requirement or the maximum number of training iterations is completed, and a preliminary trained BP neural network model is obtained.

[0070] The model trained in the initial stage is validated using the validation set obtained in step S1. The prediction error of the validation set is calculated. If the prediction accuracy of the validation set does not meet the preset standard (coefficient of determination R), the model is considered to be at fault. 2 If the value is ≥0.9, return to step S2 to adjust the number of hidden layer nodes, learning rate, and other network parameters, and retrain the model until the optimal settlement prediction model that meets the actual accuracy requirements of the engineering is obtained.

[0071] S4. Model Testing and Accuracy Evaluation. Input the test set from step S1 into the optimal settlement prediction model to obtain the predicted settlement values ​​of the levee for the test set. Calculate the mean square error (MSE) and coefficient of determination (R²) of the model. 2 As an evaluation metric for model prediction accuracy:

[0072] Mean Square Error (MSE): Reflects the average squared error between predicted and measured values. A smaller value indicates higher model accuracy. The formula is: Where n is the number of test set samples, These are the measured values ​​of levee settlement / uplift. This is the predicted deformation value;

[0073] Coefficient of determination R 2 This reflects how well the model fits the data. The closer the value is to 1, the better the model fit. The formula is: ,in, This represents the average value of the measured settlement / uplift of the embankment.

[0074] If the model's R 2 If the MSE is ≥0.9 and MSE≤0.5, the model is deemed to meet the actual prediction requirements of the project and can be used for subsequent real-time settlement prediction; if the requirements are not met, return to step S2 to re-optimize the model parameters.

[0075] S5. Real-time Settlement Prediction. During the actual construction of a large-diameter slurry shield tunneling through a dam, the shield tunneling monitoring system and data acquisition terminal collect real-time data on the shield tunneling cut pressure, tunneling speed, synchronous grouting pressure, and synchronous grouting volume. After the real-time data is cleaned and normalized in step S1, it is input into the optimal settlement / uplift deformation prediction model obtained in step S4. Through forward propagation of the model, the real-time predicted value of dam settlement is quickly output, realizing the advanced prediction of dam deformation.

[0076] S6. Dynamic Graded Early Warning. Based on the "Design Code for Embankment Engineering" and the structural characteristics of the embankment, corresponding settlement early warning thresholds are dynamically set for different tunneling stages. The thresholds are divided into three levels: Level 1 (early warning notification), Level 2 (audible and visual alarm), and Level 3 (emergency alarm). The tail-end settlement stage has the highest settlement rate and the greatest risk, so the thresholds are set more strictly: 60% (Level 1), 80% (Level 2), and 100% (Level 3) of the allowable settlement value of the embankment. For other stages, the thresholds are adjusted appropriately according to the settlement risk. Furthermore, in practical applications, deformation acceleration values ​​can be considered as part of the early warning criteria, with the following preferred options:

[0077] No warning: |Predicted settlement value| ≤ Level 1 warning threshold or the absolute value of settlement deformation acceleration is less than 2 mm / d 2 ;

[0078] Level 1 warning: Level 1 warning threshold <|predicted settlement value| ≤ Level 2 warning threshold or the absolute value of settlement deformation acceleration is greater than or equal to 2 mm / d 2 Less than 4mm / d 2 ;

[0079] Level II warning: Level II warning threshold <|predicted settlement value| ≤ Level III warning threshold or absolute value of settlement deformation acceleration greater than or equal to 4 mm / d 2 Less than 6mmd 2 ;

[0080] Level 3 warning: The predicted settlement value is greater than the Level 3 warning threshold or the absolute value of the settlement deformation acceleration is greater than or equal to 6 mm / d. 2 The real-time settlement prediction value obtained in step S5 is compared with the early warning threshold of the corresponding tunneling stage in real time. The corresponding early warning level is triggered, and targeted construction adjustment suggestions are output. The specific triggering rules are as follows:

[0081] No warning: |Predicted settlement value| ≤ Level 1 warning threshold, the settlement / uplift of the embankment is within a safe range, maintain the current shield tunneling parameters, and conduct settlement monitoring at the regular frequency;

[0082] Level 1 warning: If the Level 1 warning threshold is less than |predicted settlement value| and less than or equal to the Level 2 warning threshold, there is a slight risk of abnormal settlement / heaving deformation of the embankment. A text / voice warning will be issued, tunneling parameters will be adjusted, and monitoring will be intensified.

[0083] Level II warning: If the Level II warning threshold is less than |predicted settlement value| and less than or equal to the Level III warning threshold, there is a moderate risk of settlement / uplift deformation of the embankment. An audible and visual alarm will be issued, tunneling parameters will be adjusted, and monitoring will be intensified.

[0084] Level 3 warning: If the predicted settlement value exceeds the Level 3 warning threshold, the embankment faces the risk of instability. An emergency alarm is issued, tunnel boring is immediately stopped, and emergency measures such as grouting reinforcement are taken until the settlement value stabilizes.

[0085] Specifically, Level 1 warning: Prioritize fine-tuning of tunneling speed and grouting parameters: reduce tunneling speed by 10%~20%; during settlement, increase synchronous grouting pressure x3 by 0.1~0.3 bar, and slightly increase grouting volume x4 by 5%~10%; during uplift, adjust accordingly.

[0086] Level 2 warning: reduce tunneling speed by 20% to 35% to reduce the rate of ground disturbance; fine-tune the cutting pressure to the upper limit of the range according to the burial depth and soil and water pressure; increase the synchronous grouting pressure by 0.2 to 0.4 bar and increase the grouting volume by 10% to 20% during settlement to strengthen ground filling and support; reduce the pressure accordingly during uplift.

[0087] Level 3 warning: Stop tunneling. During settlement, the cutting pressure should be adjusted to the upper limit of the range according to the burial depth and soil and water pressure, with grouting pressure +0.3~0.6 bar and grouting volume +20%~35%. During uplift, the pressure should be adjusted downward accordingly.

[0088] S7. Early Warning Feedback and Model Iteration. Establish a dynamic update database for engineering data, and feed back the construction adjustment parameters and actual embankment settlement monitoring values ​​to the database in real time after the early warning is triggered during shield tunneling. According to the shield tunneling progress, every 5 to 10 rings, the newly added measured data in the database is added to the training set of step S1, and the optimal settlement prediction model is iteratively trained, the network weights and thresholds are readjusted, and the model parameters are continuously optimized.

[0089] Through model iteration, the prediction model can adapt to the dynamic changes in engineering geology and construction conditions during shield tunneling, continuously improving the prediction accuracy and engineering adaptability of the model, and realizing the dynamic optimization of the prediction and early warning system.

[0090] The underlying logic of the shield tunneling deformation early warning method for crossing a levee disclosed in the above embodiments can be implemented entirely or partially through software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented entirely or partially in the form of a computer program product. A computer program product includes one or more computer instructions or computer programs.

[0091] When implemented in hardware, the methods disclosed in the above embodiments can be used to build a shield tunneling deformation early warning system for crossing a levee. The system consists of an acquisition module, a BP neural network settlement prediction model, a training module, a model optimization module, a prediction value output module, an early warning triggering module, and a module iteration module.

[0092] The data acquisition module is used to collect settlement data from monitoring points on the embankment and shield tunneling parameter data during the shield tunneling construction process. The settlement data and the shield tunneling parameter data are preprocessed to obtain a preprocessed dataset, and the preprocessed dataset is divided into a training set, a validation set, and a test set.

[0093] The BP neural network settlement prediction model is constructed based on the training set; the BP neural network settlement prediction model includes an input layer, a hidden layer and an output layer, the input layer is used to input shield tunneling parameters, and the output layer is used to output the predicted settlement value of the embankment;

[0094] The training module is used to input the training set into the BP neural network settlement prediction model for model training, and to use the validation set to validate the trained BP neural network settlement prediction model to obtain the optimal settlement prediction model.

[0095] The model selection module is used to input the test set into the optimal settlement prediction model for model testing and accuracy evaluation. When the optimal settlement prediction model meets the preset accuracy requirements, the optimal settlement prediction model is determined as the target prediction model.

[0096] The prediction output module is used to collect real-time data of shield tunneling parameters during the shield tunneling construction process, and input the preprocessed real-time data of shield tunneling parameters into the target prediction model to obtain the real-time prediction value of embankment settlement.

[0097] The early warning triggering module sets corresponding settlement early warning thresholds based on different tunneling stages, and compares the real-time predicted settlement value of the embankment with the settlement early warning threshold of the corresponding tunneling stage to trigger the corresponding early warning level.

[0098] The module iteration module is used to feed back the construction adjustment parameters after the early warning is triggered and the actual embankment settlement monitoring values ​​to the dynamic update database of engineering data, and to iteratively train the target prediction model based on the newly added actual embankment settlement monitoring values ​​in order to update the model parameters.

[0099] When computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives (SSDs).

[0100] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0102] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] In summary, this method uses the shield tunneling cutting pressure, tunneling speed, synchronous grouting pressure, and synchronous grouting volume as input layer parameters of a BP neural network, and the embankment settlement value as the output layer parameter. The BP neural network is trained, validated, and optimized using engineering measurement data to construct a high-precision settlement prediction model. This model is used to predict embankment settlement in real time, and early warning thresholds are dynamically set based on the shield tunneling stages, achieving advanced prediction and tiered early warning of embankment settlement. This invention achieves deep coupling between construction parameters and settlement prediction, resulting in high prediction accuracy and rapid early warning response. It can precisely guide the adjustment of shield tunneling construction parameters, effectively control embankment settlement and deformation, and ensure the safety of shield tunneling through embankments.

[0104] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for early warning of deformation when a shield tunnel crosses a levee, characterized in that, Includes the following steps: S1. Collect settlement data from monitoring points on the embankment and shield tunneling parameter data during the shield tunneling construction process. Preprocess the settlement data and the shield tunneling parameter data to obtain a preprocessed dataset. Divide the preprocessed dataset into a training set, a validation set, and a test set. S2. Construct a BP neural network settlement prediction model based on the training set. The BP neural network settlement prediction model includes an input layer, a hidden layer, and an output layer. The input layer is used to input shield tunneling parameters, and the output layer is used to output the predicted settlement value of the embankment. S3. Input the training set into the BP neural network settlement prediction model for model training, and use the validation set to validate the trained BP neural network settlement prediction model to obtain the optimal settlement prediction model. S4. Input the test set into the optimal settlement prediction model for model testing and accuracy evaluation. When the optimal settlement prediction model meets the preset accuracy requirements, determine the optimal settlement prediction model as the target prediction model. S5. During the shield tunneling construction, real-time data of shield tunneling parameters are collected, and the real-time data of shield tunneling parameters are preprocessed and then input into the target prediction model to obtain the real-time predicted value of embankment settlement. S6. Set corresponding settlement warning thresholds based on different tunneling stages, and compare the real-time predicted settlement value of the embankment with the settlement warning threshold of the corresponding tunneling stage to trigger the corresponding warning level. S7. Feed back the construction adjustment parameters after the early warning is triggered and the actual embankment settlement monitoring values ​​to the dynamic update database of engineering data, and iteratively train the target prediction model based on the newly added actual embankment settlement monitoring values ​​to update the model parameters.

2. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 1, characterized in that, In the BP neural network settlement prediction model, the number of nodes in the input layer is 4, which correspond to the cutting pressure, tunneling speed, synchronous grouting pressure and synchronous grouting volume, respectively. The output layer has 1 node, corresponding to the levee settlement value; The number of nodes in the hidden layer was determined using a trial-and-error method.

3. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 2, characterized in that, The hidden layer uses the Sigmoid function as the activation function, and the output layer uses the linear Purelin function as the activation function. The BP neural network settlement prediction model is trained using the LM optimization algorithm.

4. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 1, characterized in that, In step S3, the training set is input into the BP neural network settlement prediction model for model training, specifically including: The data input from the input layer is forward-propagated to obtain the predicted value of the levee settlement; Calculate the mean square error between the predicted and measured values ​​of the embankment settlement, and then backpropagate the error. Adjust network weights and thresholds based on backpropagation results; Repeat the forward and backward propagation processes until the training error reaches the preset accuracy requirement or the maximum number of training iterations is reached.

5. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 1, characterized in that, In step S3, the prediction error of the BP neural network settlement prediction model is calculated using the validation set; When the prediction accuracy does not meet the preset standard, the number of hidden layer nodes and the learning rate are adjusted, and the BP neural network settlement prediction model is retrained.

6. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 1, characterized in that, In step S4, the mean square error (MSE) and the coefficient of determination (R²) are used. 2 As an accuracy evaluation index for the target prediction model; When the determination coefficient R 2 If the mean square error (MSE) is greater than or equal to 0.9 and less than or equal to 0.5, the target prediction model is determined to meet the preset accuracy requirements.

7. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 1, characterized in that, In step S6, the settlement early warning threshold includes a first-level early warning threshold, a second-level early warning threshold, and a third-level early warning threshold; Among them, the first-level warning threshold, the second-level warning threshold, and the third-level warning threshold for the tail settlement stage are 60%, 80%, and 100% of the allowable settlement value of the embankment, respectively.

8. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 7, characterized in that, In step S6, the real-time predicted value of embankment settlement is compared with the settlement early warning threshold of the corresponding tunneling stage to trigger the corresponding early warning level, specifically including: When the real-time predicted value of the embankment settlement is less than or equal to the first-level early warning threshold, the current shield tunneling parameters are maintained. When the real-time predicted value of the embankment settlement is greater than the first-level early warning threshold and less than or equal to the second-level early warning threshold, an early warning is issued and the shield tunneling parameters are adjusted. When the real-time predicted value of the embankment settlement is greater than the secondary warning threshold and less than or equal to the tertiary warning threshold, an audible and visual alarm is issued and the shield tunneling parameters are adjusted. When the real-time predicted value of the embankment settlement exceeds the level 3 early warning threshold, an emergency alarm is issued, and the tunnel boring machine is stopped and grouting reinforcement measures are taken.

9. The method for early warning of deformation when a shield tunnel crosses a levee according to claim 1, characterized in that, In step S7, according to the progress of the tunnel boring machine, every 5 to 10 rings, the newly added actual embankment settlement monitoring values ​​are added to the training set, and the target prediction model is iteratively trained to readjust the network weights and thresholds.

10. A shield tunneling deformation early warning system for crossing a levee, characterized in that, include: The data acquisition module is used to collect settlement data from monitoring points on the embankment and shield tunneling parameter data during the shield tunneling construction process. The settlement data and the shield tunneling parameter data are preprocessed to obtain a preprocessed dataset, and the preprocessed dataset is divided into a training set, a validation set, and a test set. The BP neural network settlement prediction model is constructed based on the training set. The BP neural network settlement prediction model includes an input layer, a hidden layer, and an output layer. The input layer is used to input shield tunneling parameters, and the output layer is used to output the predicted settlement value of the embankment. The training module is used to input the training set into the BP neural network settlement prediction model for model training, and to use the validation set to validate the trained BP neural network settlement prediction model to obtain the optimal settlement prediction model. The model selection module is used to input the test set into the optimal settlement prediction model for model testing and accuracy evaluation. When the optimal settlement prediction model meets the preset accuracy requirements, the optimal settlement prediction model is determined as the target prediction model. The prediction output module is used to collect real-time data of shield tunneling parameters during the shield tunneling construction process, and input the preprocessed real-time data of shield tunneling parameters into the target prediction model to obtain the real-time prediction value of embankment settlement. The early warning triggering module sets corresponding settlement early warning thresholds based on different tunneling stages, compares the real-time predicted settlement value of the embankment with the settlement early warning threshold of the corresponding tunneling stage, and triggers the corresponding early warning level. The module iteration module is used to feed back the construction adjustment parameters after the early warning is triggered and the actual embankment settlement monitoring values ​​to the dynamic update database of engineering data, and to iteratively train the target prediction model based on the newly added actual embankment settlement monitoring values ​​in order to update the model parameters.

11. An electronic device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store a plurality of executable instructions, which cause the processor to execute the method for predicting the deformation of foundation pit maintenance structures based on fractal theory as described in any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The storage medium stores a plurality of executable instructions, which, when executed on an electronic device, cause the electronic device to perform the deformation prediction method for foundation pit maintenance structure based on fractal theory as described in any one of claims 1 to 9.

Images