Bridge-tunnel combined construction method based on construction parameter self-adaptation

Abstract

This invention discloses a bridge-tunnel co-construction method based on adaptive construction parameters, specifically relating to the field of bridge-tunnel collaborative construction technology. It collects disturbance response parameters such as foundation settlement rate, pile foundation lateral displacement, and tunnel arch convergence rate to form a set of construction disturbance parameters. This set is input into a structural coupling identification model to extract the pile-arch relative stiffness ratio, pile end support reaction difference, and tunnel deformation anomaly values, constructing a structural coupling state feature vector. Based on this, an optimal construction sequence combination strategy is matched to implement bridge-tunnel coupled construction. During construction, the disturbance response parameters are updated in real time, compared with historical curves, and the disturbance offset is calculated. When the offset exceeds a preset tolerance threshold, iterative updates of features and strategies are automatically triggered until the disturbance stabilizes. This method enables dynamic adjustment of construction sequences and structural collaborative control of bridge-tunnel connection sections, significantly improving the adaptability, safety, and construction efficiency of bridge-tunnel co-construction projects under complex geological conditions.

Description

Technical Field

[0001] This invention relates to the field of bridge and tunnel collaborative construction technology, and specifically to a bridge and tunnel co-construction method based on adaptive construction parameters. Background Technology

[0002] In transportation infrastructure construction, the combined construction of bridges and tunnels (referred to as "bridge-tunnel co-construction") is becoming increasingly common, especially in mountainous and coastal areas with complex terrain and variable hydrological conditions. Bridge-tunnel co-construction projects often face the problem of mutual interference between tunnel excavation and bridge foundation construction procedures. During construction, it is necessary to frequently switch processes, allocate equipment, adjust construction intensity and pace, which places extremely high demands on construction management.

[0003] Especially in geological transition zones (such as transitions from soft soil to isolated rock layers, fault zones, or expansive rock layers), tunnel excavation may cause changes in foundation settlement or soil disturbance, which in turn affects the bearing capacity of bridge foundations and the verticality of pile foundations, resulting in structural stress concentration, increased settlement differences, and even serious problems such as misalignment at bridge ends and deformation of bridge-tunnel transition sections.

[0004] Currently, most construction schemes are based on pre-set geological models and design parameters, lacking the ability to identify and adaptively adjust the response characteristics of construction disturbances in real time. This leads to problems such as unplanned interruptions, post-construction structural instability, and frequent maintenance during the operation period during the bridge-tunnel collaborative construction process, which seriously restricts the efficiency and reliability of bridge-tunnel co-construction projects.

[0005] Therefore, there is an urgent need for a bridge-tunnel co-construction method that can automatically identify the coupling state of bridge and tunnel structures based on disturbance response parameters during construction, and dynamically adjust the construction sequence, construction rhythm and structural connection form, so as to fundamentally improve the adaptability and stability of bridge-tunnel collaborative construction and ensure the long-term service performance of the structure. Summary of the Invention

[0006] The purpose of this invention is to provide a bridge-tunnel co-construction method based on adaptive construction parameters to address the shortcomings of the prior art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a bridge-tunnel co-construction method based on adaptive construction parameters, comprising:

[0008] S100. Set up a sensor array in the bridge-tunnel transition section to obtain a set of construction disturbance response parameters P0, including the foundation settlement rate, pile foundation lateral displacement and tunnel arch convergence rate.

[0009] S200. Input the set of construction disturbance response parameters P0 into the structural coupling identification model to obtain the structural coupling state feature vector F1 of the bridge-tunnel connection section, wherein the structural coupling state includes the pile-arch relative stiffness ratio, the pile end support reaction force difference and the tunnel deformation cooperative anomaly value.

[0010] S300. Based on the structural coupling state feature vector F1, call the construction procedure combination strategy that best matches it, including construction sequence, procedure interval time and temporary support type.

[0011] S400. Implement the coupled construction of bridge and tunnel according to the construction procedure combination strategy, and update the set of construction disturbance response parameters P0 in real time, compare it with the historical disturbance curve, and calculate the disturbance offset ΔP.

[0012] S500. When the disturbance offset ΔP is greater than or equal to the preset tolerance threshold T, the structural coupling state feature vector F1 is readjusted and the construction procedure combination strategy is iteratively updated until the disturbance offset is less than the preset tolerance threshold T, and the adaptive co-construction of the bridge-tunnel transition section is completed.

[0013] Preferably, the calculation of the pile-arch relative stiffness ratio includes the following steps:

[0014] Based on the load-displacement data of the pile foundation and arch collected in the construction disturbance response parameter set P0, a structural response time series matrix D is constructed;

[0015] The structural response time series matrix D is preprocessed to form a set of state variable sequences X and its derivative sequences for modeling.

[0016] The sparse mapping relationship between state derivatives and variables is constructed by a sparse identification nonlinear dynamics algorithm to identify the principal stiffness control coefficients of pile foundation and arch.

[0017] Based on the coefficient ratio of the identified control items, the pile-arch relative stiffness ratio is calculated.

[0018] Preferably, the calculation of the pile end support reaction difference includes the following steps:

[0019] The data related to pile displacement and foundation deformation in the set of construction disturbance response parameters P0 are used to construct the input tensor, and the coordinate axis in the pile length direction is used as the input variable to construct the input space of the pile foundation response prediction model.

[0020] A physical information neural network model is constructed by embedding the control differential equations of the pile foundation. The differential equations include the equilibrium equations and boundary conditions of the pile-soil system, which are embedded in the network structure as loss function terms.

[0021] The physical information neural network model is trained to output the predicted displacement distribution and internal force response of the pile foundation along the depth direction, thereby obtaining the theoretical support reaction force value at the pile end;

[0022] The difference between the theoretical support reaction force value at the pile end and the measured support reaction force is calculated to obtain the pile end support reaction force difference value.

[0023] Preferably, the calculation of tunnel deformation co-anomalies includes the following steps:

[0024] A sensor graph structure is constructed based on multi-node displacement sensors deployed at the tunnel arch and shoulder. Each sensor is a graph node, and the nodes are connected by weighted edges based on structural distance and force path to form a tunnel structure sensing graph.

[0025] The temporal deformation variables collected by each node in the sensor graph are used as input features to construct a graph attention network model;

[0026] By training the graph attention network model, deformation coordination features between nodes are extracted, local abnormal deformation areas of the arch are identified, and the degree of coordination deviation between the local abnormal deformation areas of the arch and the whole structure is identified.

[0027] The tunnel deformation co-anomaly value is calculated based on the deviation between the co-displacement degree and the global average response.

[0028] Preferably, the step of invoking the most suitable construction procedure combination strategy includes:

[0029] The structural coupling state feature vector F1 is input into a pre-constructed multi-dimensional process strategy matching database, which is constructed based on historical bridge and tunnel collaborative construction data through structural feature clustering and process mapping.

[0030] Using a feature matching algorithm based on cosine similarity, the matching degree between F1 and each strategy feature template in the database is calculated, and the strategy template with the highest similarity is selected as the current recommended process combination strategy.

[0031] The construction process combination strategy includes the construction sequence of bridge pile foundations and tunnel arches, the interval between processes, and the transition structure of connecting sections, which are automatically configured according to the structural coupling level.

[0032] Preferably, the step of updating the construction disturbance response parameter set P0 in real time includes:

[0033] According to the matching construction sequence combination strategy, the construction tasks of bridge pile foundation and tunnel arch are carried out in sequence, and the ground settlement rate, pile foundation lateral displacement and tunnel arch convergence rate are continuously collected at the key construction nodes to form an updated set of disturbance response parameters P1.

[0034] The updated disturbance response parameter set P1 is standardized and the current disturbance response curve is constructed in a time series manner. It is then compared point-to-point with the disturbance reference curve under the corresponding coupling state in the historical construction data.

[0035] By comparing the deformation, response speed, and fluctuation range of the curve over multiple time periods, the disturbance offset ΔP is calculated based on the Euclidean distance.

[0036] Preferably, the step of calculating the disturbance offset ΔP includes the following:

[0037] The disturbance response parameter set P1 collected during the current construction phase is divided into multiple equal-length analysis periods according to the time series, and the deformation, response speed and fluctuation range in each period are extracted as three-dimensional feature vectors.

[0038] The three-dimensional feature vectors for each analysis period are normalized and compared with the feature vectors of the same dimension under the corresponding coupling state in the historical disturbance reference curves on a time-by-time basis.

[0039] The Euclidean distance algorithm is used to calculate the multidimensional difference between the current feature vector and the historical feature vectors, and the sum is used to obtain the perturbation offset ΔP.

[0040] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0041] 1. This invention integrates multiple intelligent analysis methods, including sparse nonlinear modeling, physically constrained neural networks, and graph neural networks, to accurately identify the structural coupling state between bridge pile foundations and tunnel arches. It dynamically extracts key indicators such as the pile-arch relative stiffness ratio, pile end support reaction difference, and tunnel deformation anomalies. By combining a process strategy matching database constructed from historical engineering big data, this invention achieves adaptive matching of optimal process combination strategies driven by construction disturbances, significantly improving the intelligent decision-making level and structural response coordination of bridge-tunnel collaborative construction.

[0042] 2. This invention constructs a closed-loop control system for the construction process by real-time acquisition of disturbance response parameters and dynamic iterative updates of structural characteristics. It possesses the capabilities for full-process disturbance perception, structural risk prediction, and rapid strategy adjustment. This solution effectively reduces the risks of misalignment, settlement differences, and structural instability caused by geological abrupt changes or structural mismatches during the construction of bridge-tunnel connection sections, significantly improving the safety, continuity, and long-term service performance of bridge-tunnel combined construction projects in complex geological environments. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0044] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. 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.

[0046] For examples, please refer to Figure 1 As shown in this embodiment, a bridge-tunnel co-construction method based on adaptive construction parameters includes:

[0047] S100. Set up a sensor array in the bridge-tunnel transition section to obtain a set of construction disturbance response parameters P0, including the foundation settlement rate, pile foundation lateral displacement and tunnel arch convergence rate.

[0048] In this invention, a multi-type, high-density sensor array is installed along the path from the bridge pile foundation to the tunnel arch in the bridge-tunnel transition section to acquire the set of construction disturbance response parameters P0 in real time. The set of parameters includes, but is not limited to, the following three key physical quantities:

[0049] Foundation settlement rate (unit: mm / d) is used to reflect the vertical deformation trend of bridge foundations under different geological disturbances;

[0050] Lateral displacement of pile foundation (unit: mm) is used to reflect the stability risk of bridge caused by weak layers or lateral disturbances;

[0051] The tunnel arch convergence rate (unit: mm / d) is used to identify the deformation evolution state of the surrounding rock before and after support.

[0052] To ensure the real-time performance and accuracy of the data, the sensor array employs a multi-source information acquisition system combining a high-sensitivity distributed fiber optic sensor, a GNSS settlement monitoring module, and a MEMS miniature displacement sensor. Specifically:

[0053] The settlement monitoring module is deployed at equal intervals of 5 meters, centered on the bridge-tunnel junction area, covering at least twice the influence range of the bridge foundation.

[0054] Lateral displacement sensors are mainly deployed in the lower and middle sections of the main pile foundation and in adjacent strata to capture horizontal shear trends.

[0055] The tunnel arch convergence sensors are arranged in an arc shape, covering the top and both sides of the arch shoulders to form a continuous convergence monitoring zone.

[0056] In the initial data acquisition phase, the system calibrates all sensors, eliminates initial offsets, and generates standardized disturbance response reference curves. During construction, the sensor array samples periodically at a frequency of 5–10 minutes, and various disturbance parameters are preliminarily processed by edge computing nodes before being transmitted to the main construction information control system for aggregation, forming a construction disturbance response parameter set P0. This set P0 serves as the fundamental input for determining the coupling relationship between bridge and tunnel structures and is a crucial prerequisite for achieving adaptive co-construction. The real-time nature of its acquisition and processing directly determines the accuracy and stability of subsequent coupling identification and strategy adjustment.

[0057] In addition, to prevent data loss caused by extreme geological disturbances or equipment failures, this system is equipped with a multi-channel redundant communication mechanism and a local caching strategy to ensure that the evolution trend of construction disturbances can still be traced in the event of communication abnormalities, thus ensuring the full-cycle responsiveness of the bridge-tunnel co-construction process.

[0058] S200. Input the set of construction disturbance response parameters P0 into the structural coupling identification model to obtain the structural coupling state feature vector F1 of the bridge-tunnel connection section, wherein the structural coupling state includes the pile-arch relative stiffness ratio, the pile end support reaction difference and the tunnel deformation cooperative anomaly value.

[0059] In this invention, in order to realize dynamic coupling identification during the bridge-tunnel collaborative construction process, a structural coupling identification model is constructed based on the set of construction disturbance response parameters P0 obtained in step S100, and the structural coupling state feature vector F1 of the bridge-tunnel connection section is output through the model to reflect the dynamic matching state of the bridge and tunnel structures under disturbance conditions.

[0060] Specifically, the construction disturbance response parameter set P0 consists of multi-dimensional structural response data collected by a high-precision sensor array deployed at the construction site, including at least: foundation settlement rate, pile lateral displacement, pile vertical load, tunnel arch convergence, and arch shoulder deformation. This dataset not only reflects the direct impact of construction disturbance on structural stability but also constitutes the basic data source for bridge-tunnel coupling behavior modeling.

[0061] To extract key state features of bridge-tunnel coupling behavior, P0 is input into a pre-constructed structural coupling identification model. This model integrates sparse dynamics identification, physical constraint neural networks, and graph neural network methods to extract the following three core structural feature parameters from the perturbation parameters, forming a feature vector F1.

[0062] The calculation steps for the pile-arch relative stiffness ratio include:

[0063] The vertical loads and corresponding vertical displacements of the bridge pile foundations and tunnel arches are extracted from the set of construction disturbance response parameters P0 collected by the sensors, and are denoted as follows:

[0064] Pile foundation: Load sequence Lp(t), displacement sequence Up(t);

[0065] Arch section: Load sequence La(t), displacement sequence Ua(t);

[0066] Align Lp(t), Up(t), La(t), and Ua(t) within a unified sampling time t∈[t0,tn] to form a 4×n-dimensional time series matrix D, which is used to describe the structural response evolution process.

[0067] The formation of the state variable sequence set X and the derivative sequence dX / dt: Apply the first-order central difference formula to each column of matrix D for differentiation, calculate the state derivative dU / dt, and form the state derivative sequence dX / dt;

[0068] The original displacement variables such as Up(t) and Ua(t) are retained as a set of state variable sequences X;

[0069] The obtained data is standardized (mean 0, variance 1) to form the input dataset for modeling.

[0070] Construction of a nonlinear dynamic model for sparse identification:

[0071] Construct a candidate function library Θ(X), which includes polynomial terms (such as first-order and second-order terms), cross terms (such as Up×Ua), and trigonometric function terms;

[0072] Using linear regression method , where Ξ is the sparse coefficient matrix to be identified; the stepwise minimum angular regression (LARS) algorithm is used to solve the sparse problem, eliminating low contribution coefficients and retaining the dominant stiffness term.

[0073] Calculate the pile-arch relative stiffness ratio:

[0074] Extract the principal control coefficients Kp of pile foundation stiffness and Ka of arch stiffness from the sparse matrix;

[0075] Calculate the ratio This ratio is used as the pile-arch relative stiffness ratio, and after normalization, it is input into the structural coupling state feature vector F1.

[0076] Constructing the input tensor and depth space mapping:

[0077] Extract data from P0 related to pile displacement U(z,t), foundation settlement S(z,t), and horizontal displacement V(z,t). Construct an input tensor T1 with the pile length direction coordinate z (0 represents the pile top, H represents the pile end) as a continuous spatial variable, in the format: T1(z,t)=[U(z,t),S(z,t),V(z,t)]. For each time point, construct the input space with z as the input and T1(z,t) as the feature vector.

[0078] Constructing a physical information neural network model:

[0079] Construct a feedforward neural network model N, with z and perturbation parameters as inputs and displacement prediction value U(z) as output;

[0080] Simultaneously, the pile-soil mechanics governing differential equations are embedded as physical constraints into the network loss function, including:

[0081] Vertical equilibrium equations: In the formula, The second derivative of the bending moment with respect to the pile depth z represents the distribution of internal forces in the vertical direction; q(z) is the distributed reaction force of the soil per unit length on the pile (unit: kN / m), i.e., the lateral earth pressure function.

[0082] Support reaction boundary condition: The support reaction at the pile tip is equal to the support reaction at the pile bottom. Rb is the pile tip support reaction force (unit: kN), which is the magnitude of the ground reaction force at the pile bottom; ks is the pile tip ground elastic reaction coefficient (unit: kN / m), which reflects the stiffness of the soil layer at the pile tip, and the value can be set according to the geological survey data. The displacement value at the pile bottom (z=H) predicted by the model is used to calculate the reaction force.

[0083] The network loss function is: L is the total loss function value, used to measure the overall error of the model output; λ1 is the weight coefficient of the data fitting term, used to control the proportion of fitting error in the total loss. λ1 is the mean squared error term, representing the average squared difference between the displacement value predicted by the model and the measured displacement, used to evaluate the prediction accuracy of the neural network; λ2 is the weight coefficient of the physical constraint term, used to control the importance of physical consistency in model optimization; PDE_loss is the partial differential equation residual loss term, i.e. whether the model output satisfies the pile foundation control differential equation (such as the equilibrium equation or deformation equation).

[0084] Training the model and deriving the theoretical support reaction force:

[0085] The Adam optimization algorithm is used to iteratively train the model, adjusting the weights to minimize the loss function;

[0086] The predicted displacement curve U(z) along the continuous depth direction is obtained, and the theoretical support reaction value Rt is derived from the mechanical relationship, where: Where E represents the elastic modulus of the pile concrete, i.e., the linear deformation stiffness of the material. It reflects the stress intensity under unit strain, and its value can be determined according to the concrete grade, such as approximately 30 GPa for C30. I represents the moment of inertia of the pile section, which depends on the geometry of the pile section.

[0087] Calculation of support reaction difference:

[0088] Obtain the measured support reaction value Rm;

[0089] Calculate the difference , as the difference in pile end support reaction force;

[0090] Input this value into the structural coupling state feature vector F1.

[0091] The calculation steps for tunnel deformation co-anomalies include:

[0092] Constructing the sensor graph structure: Number the displacement sensors deployed on the tunnel arch and shoulders as nodes V={v1,v2,...,vn};

[0093] A weighted edge E1 is established based on the structural distance dij between sensors and the theoretical force flow path, with the edge weight being... ;in, This represents the distance decay adjustment factor, used to control the sensitivity of distance to weight decay;

[0094] Construct a graph structure G=(V,E1) to represent the tunnel structure deformation network.

[0095] Graph attention network model construction: The input node features are the time series xi(t) of the shape variables of each measurement point;

[0096] The network structure consists of two graph attention computation layers: each layer calculates the attention coefficient between nodes based on the features of adjacent nodes and edge weights; the embedding feature hi of the output node represents its local deformation pattern; the model is trained using historical normal construction phase data, and the optimization objective is to reconstruct the average deformation response of the entire structure.

[0097] Local anomaly region identification and collaborative deviation calculation: By comparing the embedded features hi of the node in the current stage with the feature distribution center μ in the training stage, Mahalanobis distance is used to determine whether it is an anomaly point;

[0098] The anomalies are grouped into a local anomaly region A. V;

[0099] The Euclidean distance between the average embedded features of this region and the average features of the whole image is calculated as the co-coord.

[0100] Tunnel deformation coordination anomaly calculation: The coordination deviation Dcoord is compared with a set threshold θ; if it exceeds the threshold, it is determined to be a coordination anomaly. The outlier intensity is represented as a term in the structural coupling state feature vector F1 and is input into the subsequent construction control model.

[0101] S300. Based on the structural coupling state feature vector F1, call the construction procedure combination strategy that best matches it, including construction sequence, procedure interval time and temporary support type.

[0102] In this invention, to achieve adaptive control during the construction of the bridge-tunnel connection section, an optimal construction sequence combination strategy is automatically matched based on the structural coupling state feature vector F1. This strategy guides the sequential and coordinated construction of the bridge pile foundation and the tunnel arch, ensuring the coordinated stability of the structural response. Specific technical steps include the following:

[0103] First, a multi-dimensional process strategy matching database for bridge and tunnel collaborative construction is established to store the optimal configuration relationship between different structural coupling states and corresponding construction processes.

[0104] The database construction process is as follows:

[0105] Collect historical monitoring data and construction logs of bridge-tunnel connection projects over the past 10 years, covering different geological conditions, bridge-tunnel combination forms and construction disturbances;

[0106] Based on the bridge and tunnel response data in each project, the pile-arch relative stiffness ratio, pile end support reaction difference, and tunnel deformation anomaly values ​​are extracted to form a set of historical feature vectors.

[0107] The density peak clustering algorithm is used to cluster the above feature vector set and divide it into different structural coupling types, with each class representing a typical structural response mode.

[0108] For the structural characteristics of each cluster center, the construction sequence, process interval time, temporary support type, and connection section structure form used in the corresponding project are statistically analyzed to form a strategy template corresponding to the structural characteristics, which is then used as one of the entries in the database.

[0109] The structural coupling state feature vector F1 obtained at the current construction stage is input into the multi-dimensional process strategy matching database constructed above. The optimal strategy search is performed using a feature matching algorithm based on cosine similarity. The specific steps are as follows:

[0110] Let F1 be the feature vector of the current structural state. Let F1 be the feature vector corresponding to the i-th policy template in the database; calculate the cosine similarity between F1 and Si, using the formula: similarity = inner product of vectors F1 and Si divided by the product of their magnitudes; traverse all policy templates in the database and select the policy template with the highest similarity value. This means finding the best combination of construction procedures for the current structural state. This matching method can identify the nearest neighboring case of the current structural state in historical data, thereby quickly selecting the validated optimal construction configuration and ensuring a high degree of consistency between the construction plan and the structural response.

[0111] Selected construction process combination strategy Including the following key construction parameters:

[0112] Construction sequence: Define the order of construction of bridge pile foundation and tunnel arch, with priority determined by a combination of pile-arch relative stiffness ratio and tunnel anomalies;

[0113] Process interval time: refers to the minimum interval between key construction stages of bridges and tunnels, in hours or days, to avoid structural interference;

[0114] Connection section transition structure forms: including the setting of flexible transition sections, settlement reduction sections or additional pile caps, etc., which are called according to the structural coupling state level;

[0115] Temporary support type: refers to the material and structural type used for phased support in bridge-tunnel connection sections, such as steel supports, advanced small pipes or long pipe roof support, and the configuration is selected based on the tunnel deformation synergy index.

[0116] This construction process combination strategy is not a static rule preset, but a dynamic generation, and is classified according to the structural coupling state level: when the pile-arch stiffness ratio is too high, a flexible connection structure is given priority and the tunnel excavation time is delayed; when the pile end support reaction force difference exceeds the warning threshold, the tunnel excavation is adjusted to proceed first and pile end reaction force leveling support is added; when the tunnel coordination abnormal value increases, a temporary support scheme is automatically inserted and the process interval is shortened.

[0117] Through the above-mentioned structured matching and dynamic configuration methods, this invention realizes intelligent adaptation and risk feedforward control in the construction process, significantly improving the coordination, continuity and structural stability of bridge-tunnel connection section construction.

[0118] S400. Implement the coupled construction of bridge and tunnel according to the construction procedure combination strategy, and update the set of construction disturbance response parameters P0 in real time, compare it with the historical disturbance curve, and calculate the disturbance offset ΔP.

[0119] In this invention, to achieve continuous disturbance monitoring and response feedback adjustment during the construction of bridge-tunnel connection sections, the coupled construction of the bridge and tunnel is dynamically executed based on a matched construction procedure combination strategy. Disturbance response parameters are continuously collected and analyzed, and the disturbance offset ΔP is calculated to quantify the degree of difference between the actual construction disturbance and the historical reference disturbance. Specifically, the following steps are included:

[0120] During the coupled construction of bridge pile foundations and tunnel arches according to a pre-matched construction sequence, sensors continuously deployed at key structural nodes collect construction disturbance response parameters, including but not limited to:

[0121] Foundation settlement rate: the amount of settlement of the ground surface or the area around the pile foundation per unit time, measured in millimeters per hour;

[0122] Lateral displacement of pile foundation: The structural displacement of the pile in the horizontal direction, reflecting lateral stability;

[0123] Tunnel arch convergence rate: The rate of change of the distance between the arch crown and the arch shoulder, reflecting the degree of deformation of the surrounding rock.

[0124] The continuously collected data are integrated according to the time series to form an updated set of disturbance response parameters P1, which serves as a digital representation of the construction disturbance characteristics at the current stage.

[0125] To eliminate the impact of dimensional differences and fluctuations between different sensors on subsequent analysis, all parameter data in set P1 were normalized. A linear proportional normalization formula was used to compress all values ​​into the [0,1] interval to ensure the comparability of features in different dimensions.

[0126] Subsequently, the normalized parameters were arranged dimensionally according to the sampling time points to construct a multidimensional disturbance response curve, where the horizontal axis represents the sampling time and the vertical axis represents the normalized parameter values. The curve is used to represent the changing trend of the structural disturbance response.

[0127] To establish a benchmark, reference perturbation response curves corresponding to the current structural coupling state are extracted from historical databases to ensure that subsequent comparisons are consistent with the structural scenario.

[0128] The disturbance response curve of the current construction phase is divided into several equal-length analysis periods along the time axis. The length of each period is determined based on the sampling frequency and monitoring response cycle, preferably ranging from 15 minutes to 60 minutes.

[0129] Within each analysis period, the following three key indicators are extracted from the disturbance response curve to form a three-dimensional feature vector:

[0130] Deformation: refers to the difference between the maximum and minimum values ​​of a parameter within a certain period, reflecting the intensity of the response;

[0131] Response speed: the average rate of change of deformation per unit time, indicating the trend of disturbance change;

[0132] Fluctuation range: This is the standard deviation of the data within this period, reflecting the stability and fluctuation amplitude of the disturbance.

[0133] Each three-dimensional feature vector formed is used to characterize the comprehensive features of the structural perturbation state during that time period.

[0134] Subsequently, the three-dimensional feature vectors of all analysis periods of the current curve are normalized, and then paired and compared with the feature vectors at the same position in the historical disturbance reference curve.

[0135] To quantitatively represent the degree of difference between the current perturbation state and the historical reference state, the Euclidean distance algorithm is used to calculate the degree of difference for each pair of corresponding feature vectors.

[0136] Let the feature vector of the current time period be A=[a1,a2,a3], and the corresponding historical vector be B=[b1,b2,b3], and their Euclidean distance D be defined as: ;

[0137] The cumulative difference value obtained by summing the Euclidean distances calculated for all time periods is the disturbance offset ΔP, which is used to quantify the overall deviation between the current disturbance and the expected disturbance.

[0138] The larger the value of ΔP, the more significant the difference between the current structural response state and the historical structural coupling scenario, providing a key criterion for subsequent dynamic adjustment of the strategy.

[0139] S500. When the disturbance offset ΔP is greater than or equal to the preset tolerance threshold T, the structural coupling state feature vector F1 is readjusted and the construction procedure combination strategy is iteratively updated until the disturbance offset is less than the preset tolerance threshold T, and the adaptive co-construction of the bridge-tunnel transition section is completed.

[0140] In this invention, to achieve adaptive combined construction of bridge-tunnel connection sections under complex construction disturbance environments, the disturbance offset ΔP is used as the basis for judging structural response anomalies. Based on this, the structural coupling state feature vector F1 is dynamically reconstructed, and the construction procedure combination strategy is iteratively updated until the structural response meets the preset stability control standard. The specific technical steps are as follows:

[0141] Once the disturbance offset ΔP calculated according to the aforementioned steps is generated, it is compared with a preset tolerance threshold T. The tolerance threshold T defines the maximum acceptable disturbance deviation range for the structural response and is an empirical parameter determined based on statistical analysis of structural stability safety boundaries in historical bridge-tunnel combined projects. The specific definition method is as follows:

[0142] The average value of the maximum disturbance offset under known safe operating conditions in historical engineering projects is taken and multiplied by the safety reduction factor;

[0143] Preferably, the tolerance threshold T is set between 0.75 and 1.25, in units of normalized Euclidean distance;

[0144] When ΔP is greater than or equal to the tolerance threshold T, it indicates that the current structural response has deviated from the expected response behavior, and there is a risk of structural incoordination, which requires triggering the feature re-identification and strategy adjustment process.

[0145] After the disturbance offset ΔP exceeds the tolerance threshold T, the structural coupling state needs to be re-identified based on the latest construction disturbance response parameter set P1, generating an updated structural coupling state feature vector F1'. The specific steps are as follows:

[0146] Key physical quantities were re-extracted from parameter set P1, including foundation settlement rate, pile foundation lateral displacement, and tunnel convergence rate;

[0147] The sparse nonlinear dynamics identification algorithm, the physical information neural network model, and the graph attention network model are respectively invoked, and the following steps (S200) are recalculated: pile-arch relative stiffness ratio; pile end support reaction difference; tunnel deformation co-anomaly value; and the above three structural response indices are recombined to form a new structural coupling state feature vector F1'. This step ensures that the extracted structural state information reflects the true situation of the current construction disturbance, and has timeliness and accuracy guarantees.

[0148] After reconstructing the structural coupling state feature vector F1', it is input into the multi-dimensional process strategy matching database, and the strategy matching step is re-executed. The specific method is as follows:

[0149] Using F1' as the input feature, the cosine similarity algorithm is used to recalculate the similarity with the feature templates in the policy database;

[0150] Get the strategy template with the highest similarity And extract its corresponding construction sequence, process interval time, connection section structure form and temporary support configuration;

[0151] New process combination strategy This serves as the basis for construction instructions at the current stage, and is used to replace the original strategy. This enables iterative updates to the strategy.

[0152] The goal of the strategy update is to mitigate the uncoordinated trend of structural disturbances through process adjustments, thereby enabling the construction process to return to a stable state.

[0153] New strategy Under guidance, bridge-tunnel coupled construction continued, a new set of disturbance response parameters P2 was collected, and the aforementioned ΔP calculation process was repeated. This process, consisting of "updating parameters—identifying features—matching strategies—executing construction" as a complete closed loop, constitutes an iterative control link.

[0154] Each iteration calculates a new perturbation offset ΔP2, ΔP3, ... and compares it with the tolerance threshold T. When ΔP is less than T, the following relationship is satisfied: perturbation offset ΔP < tolerance threshold T;

[0155] If the current structural response is consistent with the expected coupling behavior, the construction process is within a controllable and stable range. At this point, the adaptive co-construction process of the bridge-tunnel connection section is complete.

[0156] In this embodiment, the dynamic feedback control method realizes multi-stage responsive scheduling of bridge and tunnel collaborative construction through disturbance-driven structural identification and strategy self-adjustment, effectively copes with structural risks caused by complex geological conditions and uncertain disturbances, and significantly improves structural adaptability and construction safety.

[0157] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A bridge-tunnel combined construction method based on construction parameter self-adaptation, characterized in that: include: S100. Set up a sensor array in the bridge-tunnel transition section to obtain a set of construction disturbance response parameters P0, including the foundation settlement rate, pile foundation lateral displacement and tunnel arch convergence rate. S200. Input the set of construction disturbance response parameters P0 into the structural coupling identification model to obtain the structural coupling state feature vector F1 of the bridge-tunnel connection section, wherein the structural coupling state includes the pile-arch relative stiffness ratio, the pile end support reaction force difference and the tunnel deformation cooperative anomaly value. S300. Based on the structural coupling state feature vector F1, call the construction procedure combination strategy that best matches it, including construction sequence, procedure interval time and temporary support type. S400. Implement the coupled construction of bridge and tunnel according to the construction procedure combination strategy, and update the set of construction disturbance response parameters P0 in real time, compare it with the historical disturbance curve, and calculate the disturbance offset ΔP. S500. When the disturbance offset ΔP is greater than or equal to the preset tolerance threshold T, the structural coupling state feature vector F1 is readjusted and the construction procedure combination strategy is iteratively updated until the disturbance offset is less than the preset tolerance threshold T, and the adaptive co-construction of the bridge-tunnel transition section is completed.

2. The bridge-tunnel combined construction method based on construction parameters self-adaptation according to claim 1, characterized in that: The calculation of the pile-arch relative stiffness ratio includes the following steps: Based on the load-displacement data of the pile foundation and arch collected in the construction disturbance response parameter set P0, a structural response time series matrix D is constructed; The structural response time series matrix D is preprocessed to form a set of state variable sequences X and its derivative sequences for modeling. The sparse mapping relationship between state derivatives and variables is constructed by a sparse identification nonlinear dynamics algorithm to identify the principal stiffness control coefficients of pile foundation and arch. Based on the coefficient ratio of the identified control items, the pile-arch relative stiffness ratio is calculated.

3. The method of claim 1, wherein the method is characterized by: The calculation of the pile end support reaction difference includes the following steps: The data related to pile displacement and foundation deformation in the set of construction disturbance response parameters P0 are used to construct the input tensor, and the coordinate axis in the pile length direction is used as the input variable to construct the input space of the pile foundation response prediction model. A physical information neural network model is constructed by embedding the control differential equations of the pile foundation. The differential equations include the equilibrium equations and boundary conditions of the pile-soil system, which are embedded in the network structure as loss function terms. The physical information neural network model is trained to output the predicted displacement distribution and internal force response of the pile foundation along the depth direction, thereby obtaining the theoretical support reaction force value at the pile end; The difference between the theoretical support reaction force value at the pile end and the measured support reaction force is calculated to obtain the pile end support reaction force difference value.

4. The method for bridge-tunnel co-construction based on adaptive construction parameters according to claim 1, characterized in that: The calculation of tunnel deformation anomalies includes the following steps: A sensor graph structure is constructed based on multi-node displacement sensors deployed at the tunnel arch and shoulder. Each sensor is a graph node, and the nodes are connected by weighted edges based on structural distance and force path to form a tunnel structure sensing graph. The temporal deformation variables collected by each node in the sensor graph are used as input features to construct a graph attention network model; By training the graph attention network model, deformation coordination features between nodes are extracted, local abnormal deformation areas of the arch are identified, and the degree of coordination deviation between the local abnormal deformation areas of the arch and the whole structure is identified. The tunnel deformation co-anomaly value is calculated based on the deviation between the co-displacement degree and the global average response.

5. The method for bridge-tunnel co-construction based on adaptive construction parameters according to claim 1, characterized in that: The steps for invoking the best-matching construction procedure combination strategy include: The structural coupling state feature vector F1 is input into a pre-constructed multi-dimensional process strategy matching database, which is constructed based on historical bridge and tunnel collaborative construction data through structural feature clustering and process mapping. Using a feature matching algorithm based on cosine similarity, the matching degree between F1 and each strategy feature template in the database is calculated, and the strategy template with the highest similarity is selected as the current recommended process combination strategy. The construction process combination strategy includes the construction sequence of bridge pile foundations and tunnel arches, the interval between processes, and the transition structure of connecting sections, which are automatically configured according to the structural coupling level.

6. The method for bridge-tunnel co-construction based on adaptive construction parameters according to claim 1, characterized in that: The steps for real-time updating of the construction disturbance response parameter set P0 include: According to the matching construction sequence combination strategy, the construction tasks of bridge pile foundation and tunnel arch are carried out in sequence, and the ground settlement rate, pile foundation lateral displacement and tunnel arch convergence rate are continuously collected at the key construction nodes to form an updated set of disturbance response parameters P1. The updated disturbance response parameter set P1 is standardized and the current disturbance response curve is constructed in a time series manner. It is then compared point-to-point with the disturbance reference curve under the corresponding coupling state in the historical construction data. By comparing the deformation, response speed, and fluctuation range of the curve over multiple time periods, the disturbance offset ΔP is calculated based on the Euclidean distance.

7. A bridge-tunnel co-construction method based on adaptive construction parameters according to claim 6, characterized in that: The steps for calculating the disturbance offset ΔP include the following: The disturbance response parameter set P1 collected during the current construction phase is divided into multiple equal-length analysis periods according to the time series, and the deformation, response speed and fluctuation range in each period are extracted as three-dimensional feature vectors. The three-dimensional feature vectors for each analysis period are normalized and compared with the feature vectors of the same dimension under the corresponding coupling state in the historical disturbance reference curves on a time-by-time basis. The Euclidean distance algorithm is used to calculate the multidimensional difference between the current feature vector and the historical feature vectors, and the sum is used to obtain the perturbation offset ΔP.

Images