Method, device and equipment for determining ground surface settlement contribution of double-line shield construction

Abstract

The application relates to a method, device and equipment for determining ground surface settlement contribution of double-line shield construction, comprising the following steps: determining the geometric influence coefficient, response sensitivity vector and equivalent stratum vector of a preceding line and a following line; splicing the response sensitivity vector and the equivalent stratum vector of the preceding line and the following line to obtain a first vector and a second vector; obtaining a first adaptive query vector based on the first vector; obtaining a second adaptive query vector based on the second vector; determining a first mask matrix based on the geometric influence coefficient of the preceding line; determining a second mask matrix based on the geometric influence coefficient of the following line; obtaining a first attention matrix based on the first mask matrix and the first adaptive query vector; obtaining a second attention matrix based on the second mask matrix and the second adaptive query vector; and determining the ground surface settlement contribution weight of the preceding line and the following line according to the first attention matrix and the second attention matrix. The method can reduce construction risks.

Description

Technical Field

[0001] This application relates to the field of intelligent construction and safety risk management technology for underground engineering, and in particular to a method, apparatus and equipment for determining the contribution of surface settlement during double-line shield tunneling. Background Technology

[0002] With the acceleration of urbanization, large-diameter shield tunneling technology has become a core technology for constructing urban underground transportation networks. In practical engineering, the construction of parallel shield tunnels is a very common practice. However, when facing complex geological formations such as "soft on top and hard underneath" or alternating soft and hard strata, the control of surface settlement caused by dual-track construction faces unprecedented challenges.

[0003] Currently, although deep learning models such as LSTM (Long Short Term Memory) and RNN (Recurrent Neural Network) perform well in time series prediction, they are usually "black box" systems. That is, they take shield parameters as input and output settlement values, but cannot determine whether the current settlement anomaly is caused by the lagging settlement of the leading line or by the immediate disturbance of the trailing line. This makes it impossible to make targeted parameter adjustments on site, increasing construction risks. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, and equipment for determining the surface settlement contribution in dual-track shield tunneling that can reduce construction risks, in order to address the above-mentioned technical problems.

[0005] A method for determining the surface settlement contribution during dual-track shield tunneling, the method comprising:

[0006] S1. Determine the geometric influence coefficient, response sensitivity vector, and equivalent stratum vector of the leading line and the trailing line at time t during the construction of a double-line shield tunnel.

[0007] S2. Concatenate the response sensitivity vector and the equivalent formation vector of the leading line to obtain a first vector, and concatenate the response sensitivity vector and the equivalent formation vector of the trailing line to obtain a second vector;

[0008] S3. Based on the first vector, obtain the first adaptive query vector of the leading line; based on the second vector, obtain the second adaptive query vector of the trailing line; based on the geometric influence coefficient of the leading line, determine the first mask matrix; based on the geometric influence coefficient of the trailing line, determine the second mask matrix.

[0009] S4. Based on the first mask matrix and the first adaptive query vector, obtain the first attention matrix; based on the second mask matrix and the second adaptive query vector, obtain the second attention matrix;

[0010] S5. Based on the first attention matrix and the second attention matrix, determine the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line, so as to obtain the predicted settlement value and adjust the construction strategy through the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line.

[0011] In this application, by determining the geometric influence coefficients, response sensitivity vectors, and equivalent stratum vectors of the leading and trailing lines at time t during dual-line shield tunneling construction, a first vector is obtained by concatenating the response sensitivity vector and equivalent stratum vector of the leading line; a second vector is obtained by concatenating the response sensitivity vector and equivalent stratum vector of the trailing line; based on the first vector, a first adaptive query vector for the leading line is obtained; based on the second vector, a second adaptive query vector for the trailing line is obtained; a first mask matrix is ​​determined based on the geometric influence coefficient of the leading line; a second mask matrix is ​​determined based on the geometric influence coefficient of the trailing line; a first attention matrix is ​​obtained based on the first mask matrix and the first adaptive query vector; a second attention matrix is ​​obtained based on the second mask matrix and the second adaptive query vector; and based on the first attention matrix and the second attention matrix, the surface settlement contribution weights of the leading and trailing lines are determined. This allows the main factors causing settlement to be identified by utilizing the output surface settlement contribution weights of the leading and trailing lines, enabling construction personnel to adjust construction strategies accordingly and reduce construction risks.

[0012] In one embodiment, step S5 includes:

[0013] According to the first attention matrix Score L (t) and the second attention matrix Score T (t), through Calculate the first surface subsidence contribution weight matrix of the leading line. ,pass Calculate the second surface subsidence contribution weight matrix of the rear track;

[0014] The average value of the elements of the first surface subsidence contribution weight matrix is ​​taken as the surface subsidence contribution weight of the leading line, and the average value of the elements of the second surface subsidence contribution weight matrix is ​​taken as the surface subsidence contribution weight of the following line.

[0015] In this application, the formula is used. and The surface settlement contribution weight matrices of the leading line and the trailing line are calculated, and the surface settlement contribution weights of the leading line and the trailing line are obtained. This achieves quantitative decoupling of the dual-line coupling effect, so that the construction strategies of the leading line and the trailing line can be determined separately by monitoring the surface settlement contribution weights of the leading line and the trailing line.

[0016] In one embodiment, the process of determining the geometric influence coefficient in step S1 includes:

[0017] Based on the straight-line distance between each preset first surface monitoring point and the cutterhead of the tunnel boring machine on the leading line, and the vertical distance between the cutterhead of the tunnel boring machine on the following line and the leading line, through... Determine the geometrical influence coefficient of the excavation work of the pilot line relative to each of the first surface monitoring points;

[0018] Based on the straight-line distance between each preset second surface monitoring point and the cutterhead of the tunnel boring machine on the following line, and the vertical distance between the cutterhead of the tunnel boring machine on the leading line and the following line, through... Determine the geometrical influence coefficient of the excavation work of the subsequent line relative to each of the second surface monitoring points;

[0019] in, X is the geometrical influence coefficient of the excavation work of the pilot line at time t relative to the first surface monitoring point i. i,L (t) represents the straight-line distance between the first surface monitoring point i and the tunnel boring machine cutterhead of the pilot line at time t, H represents the vertical distance from the centerline of the pilot line to the ground surface, K represents the stratum width coefficient, z0 represents the burial depth of the pilot line, and Y represents the depth of the pilot line. T (t) represents the vertical distance between the tunnel boring machine cutterhead and the leading line at time t; X is the geometrical influence coefficient of the excavation work of the rear track at time t relative to the second surface monitoring point j. j,T (t) represents the straight-line distance between the second surface monitoring point j and the cutterhead of the tunnel boring machine on the backward line at time t. L (t) represents the vertical distance between the tunnel boring machine cutterhead on the leading line and the trailing line at time t.

[0020] In this application, based on the straight-line distance between each preset first surface monitoring point and the cutterhead of the tunnel boring machine on the leading line, and the vertical distance between the cutterhead of the tunnel boring machine on the following line and the leading line, through... Determine the geometric influence coefficient of the excavation work of the pilot line relative to each first surface monitoring point; based on the straight-line distance between each preset second surface monitoring point and the cutterhead of the tunnel boring machine (TBM) of the subsequent line, and the vertical distance between the cutterhead of the TBM of the pilot line and the subsequent line, through... By determining the geometric influence coefficient of the excavation work of the subsequent line relative to each second surface monitoring point, the specific impact of the preceding and subsequent lines on the first and second surface monitoring points during construction can be accurately quantified.

[0021] In one embodiment, obtaining the predicted subsidence value using the surface subsidence contribution weight of the leading line and the surface subsidence contribution weight of the trailing line includes:

[0022] Determine the construction disturbance energy vectors of the leading line and the trailing line at time t;

[0023] The construction disturbance energy vector of the leading line is linearly transformed by a first linear transformation matrix to obtain the value vector of the leading line; the construction disturbance energy vector of the following line is linearly transformed by a second linear transformation matrix to obtain the value vector of the following line.

[0024] Based on the value vector V of the leading line L (t) and surface subsidence contribution weight The value vector V of the following line T (t) and surface subsidence contribution weight ,pass Obtain the comprehensive eigenvector V at time t all (t);

[0025] The comprehensive feature vector V based on time t all (t), which predicts the surface subsidence value from time t+1 to time t+N, where N is a non-negative integer.

[0026] In this application, the construction disturbance energy vectors of the leading line and the trailing line at time t are determined; the construction disturbance energy vector of the leading line is linearly transformed using a first linear transformation matrix to obtain the value vector of the leading line; the construction disturbance energy vector of the trailing line is linearly transformed using a second linear transformation matrix to obtain the value vector of the trailing line; based on the value vector V of the leading line... L (t) and the weight of surface subsidence contribution, and the value vector V of the backward line. T (t) and the weight of surface subsidence contribution, through Obtain the comprehensive eigenvector V at time t all (t); this allows us to base our comprehensive feature vector V on time t. all (t) predicts the predicted settlement value of the ground surface from time t+1 to time t+N, so as to adjust the construction strategy in time when the predicted settlement value is greater than the settlement threshold.

[0027] In one embodiment, the predicted settlement value is obtained through a dual-stream spatiotemporal attention network, and the method further includes:

[0028] Obtain the actual settlement values ​​from time t+1 to time t+N;

[0029] The actual settlement value S based on time t+1 to time t+N real and the predicted settlement value ,pass Calculate the model loss Loss of the dual-stream spatiotemporal attention network. total Loss Weight To punish the loss, Loss Shape For shape loss, and For hyperparameters;

[0030] The model parameters of the dual-stream spatiotemporal attention network are optimized based on the model loss to predict the settlement value using the optimized dual-stream spatiotemporal attention network.

[0031] In this application, a total loss function is constructed, comprising a data fidelity term (predicted settlement loss), a weight sparsity constraint term (penalty loss), and a morphological consistency constraint term (shape loss). This function guides the dual-stream spatiotemporal attention network to learn the laws of geotechnical mechanics, enabling settlement prediction while adhering to these laws. Furthermore, it ensures that the dual-stream spatiotemporal attention network maintains its predictive ability consistent with geotechnical mechanics even in areas with scarce data or abrupt geological changes, avoiding predictions that violate mechanical principles, such as reverse uplift.

[0032] In one embodiment, step S4 includes:

[0033] Determine the construction disturbance energy vectors of the leading line and the trailing line at time t;

[0034] The construction disturbance energy vector of the leading line is linearly transformed by the third linear transformation matrix to obtain the key vector of the leading line; the construction disturbance energy vector of the following line is linearly transformed by the fourth linear transformation matrix to obtain the key vector of the following line.

[0035] Based on the key vector K of the preceding line L (t), the first mask matrix M L (t) and the first adaptive query vector Q L (t), through Obtain the first attention matrix Score at time t L (t); based on the key vector K of the following line T (t), the second mask matrix M T (t) and the second adaptive query vector Q T (t), through Obtain the second attention matrix Score at time t T (t), where d is the dimension of the key vector.

[0036] In this application, the construction disturbance energy vectors of the leading line and the trailing line at time t are determined; the construction disturbance energy vector of the leading line is linearly transformed using a third linear transformation matrix to obtain the key vector of the leading line; the construction disturbance energy vector of the trailing line is linearly transformed using a fourth linear transformation matrix to obtain the key vector of the trailing line; based on the key vector K of the leading line... L (t), First mask matrix M L (t) and the first adaptive query vector Q L (t), through Obtain the first attention matrix Score at time t L (t); based on the key vector K of the backward line T (t), the second mask matrix M T (t) and the second adaptive query vector Q T (t), through Obtain the second attention matrix Score at time t T (t), so that the settlement contribution weights of the leading line and the trailing line can be determined based on the first attention matrix and the second attention matrix, and thus the excavation line with a greater impact on settlement can be determined based on the settlement contribution weights of the leading line and the trailing line.

[0037] In one embodiment, determining the construction disturbance energy vectors of the leading line and the trailing line at time t includes:

[0038] Obtain the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the tunnel boring machine for the leading and trailing lines at time t. and standard deviation of earth pressure and weighting coefficients and ;

[0039] Based on the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the aforementioned pilot line. and standard deviation of earth pressure and weighting coefficients and ,pass Calculate the specific energy excavation index (SEI) of the leading line at time t. L (t), through Calculate the soil pressure volatility index of the leading line. ;

[0040] Based on the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the aforementioned backward travel path. and standard deviation of earth pressure and weighting coefficients and ,pass Calculate the specific energy excavation index (SEI) of the backward travel line at time t. T (t), through Calculate the volatility index of soil pressure on the trailing line. ;

[0041] The construction disturbance energy vector of the leading line is determined based on the specific energy tunneling index and the soil pressure fluctuation rate index of the leading line; the construction disturbance energy vector of the following line is determined based on the specific energy tunneling index and the soil pressure fluctuation rate index of the following line.

[0042] In this application, the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil chamber pressure are used as the basis. and standard deviation of earth pressure and weighting coefficients and Calculate the specific energy tunneling index (SEI) L (t) and the soil pressure fluctuation index can quantify the degree of damage to the strata caused by total thrust, cutterhead torque, tunneling speed, cutterhead rotation speed, excavation cross-sectional area, mean soil pressure and standard deviation of soil pressure. Thus, the settlement contribution weights of the leading line and the trailing line can be accurately identified based on the quantified construction disturbance energy vector.

[0043] In one embodiment, the process of determining the first mask matrix and the second mask matrix in step S3 includes:

[0044] Obtain the preset attenuation enhancement coefficient and target constant;

[0045] Based on the attenuation enhancement coefficient The target constant and the set of geometric influence coefficients of the leading line. ,pass Determine the first mask matrix M L (t);

[0046] Based on the attenuation enhancement coefficient The target constant and the set of geometric influence coefficients of the following line. ,pass Determine the second mask matrix M T (t).

[0047] In this application, by obtaining a preset attenuation enhancement coefficient and a target constant, it is possible to base the attenuation enhancement coefficient on... Target constant and the geometric influence coefficient of the leading line ,pass Determine the first mask matrix M L (t), based on the attenuation enhancement coefficient Target constant and the geometric influence coefficient of the following line ,pass Determine the second mask matrix M T (t).

[0048] A surface settlement contribution determination device for dual-track shield tunneling, used to perform the above-described method, the device comprising:

[0049] The data determination module is used to determine the geometric influence coefficient, response sensitivity vector, and equivalent stratum vector of the leading line and the trailing line at time t during the construction of a double-line shield tunnel.

[0050] The splicing module is used to splice the response sensitivity vector and the equivalent formation vector of the leading line to obtain a first vector, and to splice the response sensitivity vector and the equivalent formation vector of the trailing line to obtain a second vector;

[0051] A vector and matrix determination module is used to obtain a first adaptive query vector for the preceding line based on the first vector; obtain a second adaptive query vector for the following line based on the second vector; determine a first mask matrix based on the geometric influence coefficient of the preceding line; and determine a second mask matrix based on the geometric influence coefficient of the following line.

[0052] The score calculation module is used to obtain a first attention matrix based on the first mask matrix and the first adaptive query vector; and to obtain a second attention matrix based on the second mask matrix and the second adaptive query vector.

[0053] The weight determination module is used to determine the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line based on the first attention matrix and the second attention matrix, so as to obtain the predicted settlement value and adjust the construction strategy through the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line.

[0054] A computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the method described above.

[0055] The aforementioned surface settlement contribution determination device for dual-track shield tunneling determines the geometric influence coefficients, response sensitivity vectors, and equivalent stratum vectors of the leading and trailing tracks at time t during the dual-track shield tunneling process. It then concatenates the response sensitivity vector and equivalent stratum vector of the leading track to obtain a first vector, and concatenates the response sensitivity vector and equivalent stratum vector of the trailing track to obtain a second vector. Based on the first vector, it obtains a first adaptive query vector for the leading track; based on the second vector, it obtains a second adaptive query vector for the trailing track; based on the geometric influence coefficients of the leading track, it determines a first mask matrix; and based on the geometric influence coefficients of the trailing track, it determines a second adaptive query vector. The influence coefficient is used to determine the second mask matrix. Based on the first mask matrix and the first adaptive query vector, the first attention matrix is ​​obtained. Based on the second mask matrix and the second adaptive query vector, the second attention matrix is ​​obtained. According to the first attention matrix and the second attention matrix, the surface settlement contribution weights of the leading line and the following line are determined. In this way, the main factors causing settlement can be identified by using the output surface settlement contribution weights of the leading line and the following line, so that construction personnel can adjust the construction strategy accordingly and reduce construction risks. Attached Figure Description

[0056] Figure 1 This is an application environment diagram of a method for determining the surface settlement contribution during dual-track shield tunneling in one embodiment.

[0057] Figure 2 This is a flowchart illustrating a method for determining the contribution of surface settlement during dual-track shield tunneling in one embodiment.

[0058] Figure 3 This is a schematic diagram showing the spatial location relationship of surface monitoring points in one embodiment;

[0059] Figure 4 This is a schematic diagram illustrating the definition of an active response window in one embodiment;

[0060] Figure 5 This is a statistical chart showing the settlement contribution weights in ring units in one embodiment;

[0061] Figure 6 This is a schematic diagram illustrating the physical meaning of the Gaussian-Mindlin mixture kernel function in one embodiment;

[0062] Figure 7 This is a comparison example of measured data and predicted data (predicted settlement values) for each ring in one embodiment;

[0063] Figure 8This is a diagram of a physically guided dual-stream spatiotemporal attention network architecture in one embodiment.

[0064] Figure 9 A schematic diagram of the overall process for determining the surface settlement contribution during dual-track shield tunneling in another embodiment;

[0065] Figure 10 A structural block diagram of a device for determining the contribution of surface settlement during dual-track shield tunneling in one embodiment;

[0066] Figure 11 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0068] The method for determining the surface settlement contribution during dual-track shield tunneling provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 interacts with server 104 via a wired / wireless channel. A data storage system can store the data that server 104 needs to process. The process involves determining the geometric influence coefficients, response sensitivity vectors, and equivalent stratum vectors of the leading and trailing lines at time t during dual-line shield tunneling; concatenating the response sensitivity vector and equivalent stratum vector of the leading line to obtain a first vector; concatenating the response sensitivity vector and tunneling state vector of the trailing line to obtain a second vector; obtaining a first adaptive query vector for the leading line based on the first vector; obtaining a second adaptive query vector for the trailing line based on the second vector; determining a first mask matrix based on the geometric influence coefficients of the leading line; determining a second mask matrix based on the geometric influence coefficients of the trailing line; obtaining a first attention matrix based on the first mask matrix and the first adaptive query vector; obtaining a second attention matrix based on the second mask matrix and the second adaptive query vector; and determining the surface settlement contribution weights of the leading and trailing lines based on the first and second attention matrices. The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, etc. The server 104 can be a single server, a server cluster consisting of multiple servers, or a cloud computing center consisting of multiple servers.

[0069] In one embodiment, such as Figure 2 As shown, a method for determining the surface settlement contribution during dual-track shield tunneling is provided, and this method is applied to... Figure 1 Taking server 104 as an example, the following steps are included:

[0070] S1. Determine the geometric influence coefficient, response sensitivity vector, and equivalent stratum vector of the leading line and the trailing line at time t during the construction of a double-line shield tunnel.

[0071] In this context, dual-track shield tunneling refers to constructing two parallel underground tunnels. The first tunnel boring machine (TBM) excavates one of the tunnels first, known as the lead tunnel. After the first tunnel has been excavated to a certain distance, the second TBM begins excavating the second tunnel, known as the follow tunnel.

[0072] The geometric influence coefficient characterizes the spatial attenuation of the mechanical disturbance energy generated by the tunnel boring machine cutting through the strata at time t, which is transmitted along the strata medium to the surface monitoring point.

[0073] The response sensitivity vector is a vector that quantifies the degree to which formation parameters affect settlement. The response sensitivity vector includes the relative stiffness ratio, strength stability index, formation heterogeneity index, and formation transition gradient. The relative stiffness ratio characterizes the formation's ability to resist deformation, the strength stability index characterizes whether the soil is close to a plastic yield state, the formation heterogeneity index is used to quantify the inhomogeneity of the formation, and the formation transition gradient is used to identify subsurface "abrupt changes."

[0074] The response sensitivity vector of the leading line Represented as The response sensitivity vector of the trailing line Represented as , The relative stiffness ratio of the leading line. This is the strength and stability index of the leading line. This is the stratigraphic heterogeneity index of the leading line. This represents the stratigraphic transition gradient of the leading line. The relative stiffness ratio of the trailing line. This is the strength and stability index of the trailing line. This is the stratigraphic heterogeneity index for the subsequent path. This represents the stratigraphic transition gradient of the subsequent line.

[0075] Relative stiffness ratio R stiff The calculation formula is E s,ref For the compression modulus reference value, in E seq When the equivalent compressive modulus of the leading line is , the relative stiffness ratio R stiff The relative stiffness ratio of the leading line, in E seq When the relative stiffness ratio R is the equivalent compressive modulus of the backward line, stiff The relative stiffness ratio of the trailing line.

[0076] Strength stability index I strength The calculation formula is , in c eq For the equivalent cohesion of the leading line, The equivalent gravity density of the leading line, When the equivalent internal friction angle of the leading line is , the strength stability index I strength The strength and stability index of the leading line; in c eq For the equivalent cohesion of the trailing line, The equivalent gravity density of the rearward line, When the effective internal friction angle of the trailing line is , the strength stability index I strength This is the strength and stability index of the trailing line.

[0077] Stratigraphic Heterogeneity Index U hetero The calculation formula is ,exist The standard deviation of the equivalent compressive modulus of each ring in the leading line, The standard deviation of the equivalent cohesion of each ring in the pilot line, The standard deviation of the equivalent internal friction angle of each ring in the leading line, The standard deviation of the equivalent gravity density of each ring in the pilot line, The mean of the equivalent compressive modulus of each ring in the leading line, The mean of the equivalent cohesion of each ring in the leading line, The mean of the equivalent internal friction angles of each ring in the leading line. When the effective gravity density of each ring in the preliminary line is the average, the formation heterogeneity index U hetero The stratigraphic heterogeneity index of the leading line; in The standard deviation of the equivalent compressive modulus of each ring in the subsequent line. The standard deviation of the equivalent cohesion of each ring in the subsequent line, The standard deviation of the equivalent internal friction angle of each ring in the trailing line, The standard deviation of the equivalent gravity density of each ring in the subsequent line. The mean of the equivalent compressive modulus of each ring in the subsequent line. The mean of the equivalent cohesion of each ring in the subsequent line. The mean of the equivalent internal friction angles of each ring in the trailing line. When the effective gravity density of each ring in the subsequent line is the mean, the formation heterogeneity index U hetero This is the stratigraphic heterogeneity index of the backward-moving line;

[0078] stratigraphic transition gradient G gradient The calculation formula is In E seq Let n be the equivalent compression modulus of the leading line, and n be the ring number excavated in the leading line at time t. For the ring width of the leading line, The mean of the equivalent compressive modulus of each ring in the leading line, The mean of the equivalent cohesion of each ring in the leading line, c eq When the equivalent cohesion of the leading line is , the formation transition gradient G gradient The stratigraphic transition gradient of the leading line; in E seq Let n be the equivalent compression modulus of the backward line, and n be the ring number excavated in the backward line at time t. For the ring width of the rear lane, The mean of the equivalent compressive modulus of each ring in the subsequent line. The mean of the equivalent cohesion of each ring in the subsequent line, c eq When the equivalent cohesion of the backward line is , the formation transition gradient G gradient This represents the stratigraphic transition gradient of the subsequent excavation line. A ring in the leading line refers to a series of physical discrete units, starting from the excavation start point of the leading line and extending horizontally forward along the leading line according to a predetermined ring width. Each physical discrete unit is numbered sequentially from the starting point. Similarly, a ring in the subsequent excavation line refers to a series of physical discrete units, starting from the excavation start point of the subsequent excavation line and extending horizontally forward along the subsequent line according to a predetermined ring width. Each physical discrete unit is numbered sequentially from the starting point. The length (ring width) of each ring is fixed. The formulas for calculating the equivalent parameters of each ring (equivalent compression modulus, equivalent cohesion, equivalent internal friction angle, equivalent gravity density) are as follows: Param eq Here are the equivalent parameters for the currently calculated ring, K1 is the number of soil layer types at the excavation section of the currently calculated ring, and h is the equivalent parameter for the currently calculated ring. k Let Param be the vertical thickness of the k-th soil layer. k The formation parameters of the ring being calculated are: compression modulus, cohesion, internal friction angle, and gravity density.

[0079] The equivalent stratum vector includes the equivalent lateral earth pressure coefficient K. ep Equivalent lag coefficient And the burial depth z0. The equivalent lateral earth pressure coefficient represents the compressive force of the soil on the tunnel sidewall, and the equivalent hysteresis coefficient represents how "sluggish" the impact of subsequent line construction on the preceding line is. Burial depth refers to the vertical distance of the preceding or subsequent line from the ground surface. The expression for the equivalent stratum vector D is: .

[0080] The equivalent stratigraphic vector is the equivalent stratigraphic vector of the ring excavated by the leading / following line at time t. Specifically, the equivalent stratigraphic vector of the ring excavated by the leading line at time t is represented as D. L (t), the equivalent formation vector of the ring excavated by the backward line at time t is represented as D. T (t).

[0081] Furthermore, longitudinal monitoring points are deployed at predetermined intervals along the surface projection direction of the tunnel design centerlines of the leading and trailing lines. Lateral monitoring points perpendicular to the centerlines of the dual tunnels are deployed at key construction nodes or environmentally sensitive areas (such as adjacent elevated ramp pile foundations, intersections of important underground pipelines, etc.). The active response window W along the tunnel axis for both longitudinal and lateral monitoring points is determined. spatial =[L start ,L end Based on the active response window W of the tunnel boring machine's cutterhead coordinates at time t, and the lateral and longitudinal monitoring points. spatial and the tunneling state vector V dig (t), determining the physical response pattern at time t; longitudinal monitoring points are used to capture the longitudinal surface subsidence evolution along the tunnel boring machine cutterhead propulsion line, and transverse monitoring points are used to capture the transverse Gaussian settlement trough morphology induced by dual-line construction. Both longitudinal and transverse monitoring points are surface monitoring points, and the spatial relationship diagram of the surface monitoring points is shown in the figure. Figure 3 As shown in the diagram, the active response window definition is illustrated below. Figure 4 As shown. The specific physical response modes are as follows:

[0082] (1) Mode A (Independent Dominant Mode for Pilot Line): Only the cutterhead coordinates of the tunnel boring machine in the pilot line are located at a certain monitoring point W. spatial Inside, and S L (t)=1;

[0083] (2) Mode B (Independent Dominant Mode for Backward Lines): Only the cutterhead coordinates of the tunnel boring machine in the backward line are located at a certain monitoring point W. spatial Inside, and S T (t)=1;

[0084] (3) Mode C (Dual-line coupling superposition mode): The tool head coordinates of both the leading line and the trailing line are located at a certain monitoring point W. spatial The inner or rearward traveler's tool head coordinates are located at a certain monitoring point W. spatial The tunnel boring machine (TBM) of the inner and rear lines traverses the active zone of delayed settlement of the leading line. In this mode, the mechanical disturbance of the two lines and the dissipation of pore water pressure in the soil generate a strong nonlinear coupling. The mode label will send a trigger command to the system, fully activating the physically guided dual-stream spatiotemporal attention network and simultaneously activating the temporal convolutional network to perform lag effect feature compensation. Among them, the active zone of delayed settlement is [L]. end , ], V is the hysteresis coefficient that is positively correlated with the clay content of the formation. avg Z is the average tunneling speed of the tunnel boring machine, and z0 is the burial depth.

[0085] Furthermore, the proactive response window W spatial=[-30m,+50m], the active response window W at a certain monitoring point where the cutterhead coordinates of the tunnel boring machine are located. spatial At that time, it was determined that the tunnel boring machine caused a direct "active construction disturbance" to the monitoring point.

[0086] S2. Combine the response sensitivity vector and equivalent formation vector of the leading line to obtain the first vector, and combine the response sensitivity vector and equivalent formation vector of the trailing line to obtain the second vector.

[0087] Among them, the response sensitivity vector of the splicing leading line and equivalent stratigraphic vector D L (t), to obtain the first vector The expression is The response sensitivity vector of the concatenated line. and equivalent stratigraphic vector D T (t), to obtain the second vector The expression is . This indicates a concatenation operation. Concatenation operation This is a concatenation operation performed on the two sets of heterogeneous vectors (response sensitivity vector and equivalent stratigraphic vector) along the feature channel dimension. The first and second vectors are combined feature matrices containing joint information of "geological background - dynamic state".

[0088] After determining the geometric influence coefficients, response sensitivity vectors, and equivalent formation vectors for the leading and trailing lines at time t, these parameters are input into a two-stream spatiotemporal attention network (SPCN). The SPCN then performs a concatenation operation between the response sensitivity vectors and the equivalent formation vectors. The SPCN consists of parallel leading and trailing line feature branches. The leading line feature branch performs vector concatenation for the leading line, and the trailing line feature branch performs vector concatenation for the trailing line. SPCNs include, but are not limited to, LSTM and TCN (Temporal Convolutional Network).

[0089] S3. Based on the first vector, obtain the first adaptive query vector of the leading line; based on the second vector, obtain the second adaptive query vector of the trailing line; based on the geometric influence coefficient of the leading line, determine the first mask matrix; based on the geometric influence coefficient of the trailing line, determine the second mask matrix.

[0090] The first adaptive query vector represents the sensitivity of the formation to disturbances caused by the tunnel boring machine (TBM) excavating the lead tunnel. The second adaptive query vector represents the sensitivity of the formation to disturbances caused by disturbances caused by the TBM excavating the follow tunnel.

[0091] The determination of the first adaptive query vector is completed in the feature branch of the preceding line, and the determination of the second adaptive query vector is completed in the feature branch of the following line. The first and second adaptive query vectors can be determined simultaneously. The first mask matrix and the second mask matrix can also be determined simultaneously. The first mask matrix is ​​used to correct the first adaptive query vector of the preceding line, obtaining the first attention matrix representing the attention energy distribution of the preceding line. The second mask matrix is ​​used to correct the second adaptive query vector of the following line, obtaining the second attention matrix representing the attention energy distribution of the following line.

[0092] Specifically, the first vector C of the leading line L (t) Input is fed into the nonlinear mapping layer of the dual-stream spatiotemporal attention network (containing a neural network layer of a multilayer perceptron), and passes through the feature branch of the leading line. Obtain the first adaptive query vector Q L (t); the second vector C of the subsequent line. T (t) is input to the nonlinear mapping layer of the dual-stream spatiotemporal attention network, and passes through the backward feature branch. Obtain the second adaptive query vector Q T (t). Among them, W q b is the learnable weight matrix of the nonlinear mapping layer. q For bias terms, It is a non-linear activation function. The learnable weight matrix W q During the model training phase, the system can implicitly learn and memorize geotechnical coupling principles such as "the extremely high vulnerability of weak strata during dual-line excavation" or "the stability of hard rock strata during shutdown." This enables the generated first / second adaptive query vectors to possess geological adaptive perception capabilities, providing accurate addressing signals for subsequent physical-guided attention layers. The learnable weight matrix and bias terms are model parameters of the dual-stream spatiotemporal attention network.

[0093] S4. Based on the first mask matrix and the first adaptive query vector, obtain the first attention matrix; based on the second mask matrix and the second adaptive query vector, obtain the second attention matrix.

[0094] The determination of the first attention matrix is ​​completed in the leading feature branch of the two-stream spatiotemporal attention network, and the determination of the second attention matrix is ​​completed in the trailing feature branch of the two-stream spatiotemporal attention network. The first attention matrix and the second attention matrix can be determined simultaneously.

[0095] S5. Based on the first attention matrix and the second attention matrix, determine the surface settlement contribution weights of the leading line and the following line, so as to obtain the predicted settlement value and adjust the construction strategy through the surface settlement contribution weights of the leading line and the following line.

[0096] The surface settlement contribution weight of the leading line is used to characterize the impact of the construction behavior of the leading line on surface settlement. The surface settlement contribution weight of the following line is used to characterize the impact of the construction behavior of the following line on surface settlement.

[0097] Furthermore, if the surface settlement contribution weight of the leading line is less than that of the following line, then the main controlling factor for settlement is determined to be the construction disturbance of the following line, rather than the delayed settlement of the leading line, and the excavation command for the following line at the next moment is adjusted. For example, the excavation command for the following line at the next moment is adjusted to "reduce the thrust of the following line to 12000kN and increase the synchronous grouting pressure to 0.3MPa". If the surface settlement contribution weight of the leading line is greater than that of the following line, then the main controlling factor for settlement is determined to be the construction disturbance of the leading line, and the excavation command for the leading line at the next moment is adjusted. For example, the excavation command for the leading line at the next moment is adjusted to "perform secondary grouting reinforcement in the area of ​​the leading line".

[0098] Furthermore, the tunnel boring machine excavates one ring for every unit distance (ring width) it travels. The settlement contribution weight of each ring is the average of the settlement contribution weights at all times during the excavation of that ring. The statistical graph of the settlement contribution weight per ring is shown below. Figure 5 As shown, the horizontal axis represents the tunneling ring number, and the contribution weight is the settlement contribution weight.

[0099] In the aforementioned method for determining the surface settlement contribution during dual-track shield tunneling, the geometric influence coefficients, response sensitivity vectors, and equivalent stratum vectors of the leading and trailing tracks at time t are determined. The response sensitivity vector and equivalent stratum vector of the leading track are then concatenated to obtain a first vector. The response sensitivity vector and equivalent stratum vector of the trailing track are then concatenated to obtain a second vector. Based on the first vector, a first adaptive query vector for the leading track is obtained; based on the second vector, a second adaptive query vector for the trailing track is obtained; a first mask matrix is ​​determined based on the geometric influence coefficients of the leading track; and a second mask matrix is ​​determined based on the geometric influence coefficients of the trailing track. The influence coefficient is used to determine the second mask matrix. Based on the first mask matrix and the first adaptive query vector, the first attention matrix is ​​obtained. Based on the second mask matrix and the second adaptive query vector, the second attention matrix is ​​obtained. According to the first attention matrix and the second attention matrix, the surface settlement contribution weights of the leading line and the following line are determined. In this way, the main factors causing settlement can be identified by using the output surface settlement contribution weights of the leading line and the following line, so that construction personnel can adjust the construction strategy accordingly and reduce construction risks.

[0100] In one embodiment, step S5 includes:

[0101] Based on the first attention matrix Score L (t) and the second attention matrix Score T (t), through Calculate the first surface subsidence contribution weight matrix of the pilot line. ,pass Calculate the second surface subsidence contribution weight matrix of the rear track;

[0102] The average value of the elements of the first surface subsidence contribution weight matrix is ​​used as the surface subsidence contribution weight of the leading line, and the average value of the elements of the second surface subsidence contribution weight matrix is ​​used as the surface subsidence contribution weight of the trailing line.

[0103] The surface settlement contribution weight of the leading line represents the impact of its construction on settlement, while the surface settlement contribution weight of the following line represents the impact of its construction on settlement. Both the surface settlement contribution weights of the leading and following lines are obtained using the normalized exponential function Softmax.

[0104] Furthermore, the maximum value of the first surface subsidence contribution weight matrix is ​​used as the surface subsidence contribution weight of the leading line, and the maximum value of the second surface subsidence contribution weight matrix is ​​used as the surface subsidence contribution weight of the trailing line.

[0105] Furthermore, the median of the first surface subsidence contribution weight matrix is ​​taken as the surface subsidence contribution weight of the leading line, and the median of the second surface subsidence contribution weight matrix is ​​taken as the surface subsidence contribution weight of the trailing line.

[0106] In this embodiment, the formula is used. and The surface settlement contribution weight matrices of the leading line and the trailing line are calculated, and the surface settlement contribution weights of the leading line and the trailing line are obtained. This achieves quantitative decoupling of the dual-line coupling effect, so that the construction strategies of the leading line and the trailing line can be determined separately by monitoring the surface settlement contribution weights of the leading line and the trailing line.

[0107] In one embodiment, the process of determining the geometric influence coefficient in step S1 includes:

[0108] Based on the straight-line distance between each preset first surface monitoring point and the cutterhead of the tunnel boring machine on the leading line, and the vertical distance between the cutterhead of the tunnel boring machine on the following line and the leading line, through... Determine the geometrical influence coefficient of the excavation work of the pilot line relative to each first surface monitoring point;

[0109] Based on the straight-line distance between each preset second surface monitoring point and the cutterhead of the tunnel boring machine on the following line, and the vertical distance between the cutterhead of the tunnel boring machine on the leading line and the following line, through Determine the geometric influence coefficient of the excavation work of the subsequent route relative to each second surface monitoring point;

[0110] in, X represents the geometrical influence coefficient of the excavation work of the pilot line at time t relative to the first surface monitoring point i. i,L (t) represents the straight-line distance between the first surface monitoring point i and the tunnel boring machine cutterhead of the pilot line at time t, H is the vertical distance from the centerline of the pilot line to the ground surface, K is the stratum width coefficient, z0 is the burial depth of the pilot line, and Y... T (t) represents the vertical distance between the tunnel boring machine cutterhead and the leading line on the subsequent line at time t; X is the geometric influence coefficient of the excavation work of the subsequent line at time t relative to the second surface monitoring point j. j,T (t) represents the straight-line distance between the second surface monitoring point j and the cutterhead of the tunnel boring machine on the following line at time t. L (t) represents the vertical distance between the tunnel boring machine cutterhead on the leading line and the trailing line at time t.

[0111] The first surface monitoring point can be any of the preset surface monitoring points, or it can be a monitoring point located directly above the central axis of the leading line. The second surface monitoring point can be any of the preset surface monitoring points, or it can be a monitoring point located directly above the central axis of the trailing line.

[0112] First item Simulation of longitudinal displacement attenuation of point source force in a semi-infinite space based on Mindlin solution, first term The physical meaning of the Gaussian-Mindlin hybrid kernel function is illustrated in the diagram below, which simulates the Gaussian decay characteristics of a transverse settling channel. Figure 6 As shown.

[0113] In this embodiment, based on the straight-line distance between each preset first surface monitoring point and the cutterhead of the tunnel boring machine on the leading line, and the vertical distance between the cutterhead of the tunnel boring machine on the following line and the leading line, through... Determine the geometric influence coefficient of the excavation work of the pilot line relative to each first surface monitoring point; based on the straight-line distance between each preset second surface monitoring point and the cutterhead of the tunnel boring machine (TBM) of the subsequent line, and the vertical distance between the cutterhead of the TBM of the pilot line and the subsequent line, through... By determining the geometric influence coefficient of the excavation work of the subsequent line relative to each second surface monitoring point, the specific impact of the preceding and subsequent lines on the first and second surface monitoring points during construction can be accurately quantified.

[0114] In one embodiment, obtaining the predicted subsidence value using the surface subsidence contribution weights of the leading line and the following line includes:

[0115] Determine the construction disturbance energy vectors of the leading line and the trailing line at time t;

[0116] The construction disturbance energy vector of the leading line is linearly transformed by the first linear transformation matrix to obtain the value vector of the leading line; the construction disturbance energy vector of the trailing line is linearly transformed by the second linear transformation matrix to obtain the value vector of the trailing line.

[0117] Based on the value vector V of the leading line L (t) and surface subsidence contribution weight The value vector V of the following line T (t) and surface subsidence contribution weight ,pass Obtain the comprehensive eigenvector V at time t all (t);

[0118] The comprehensive feature vector V based on time t all (t), which predicts the surface subsidence value from time t+1 to time t+N, where N is a non-negative integer.

[0119] The construction disturbance energy vector includes the specific energy tunneling index and the soil pressure fluctuation rate index. The specific energy tunneling index refers to the energy consumed in breaking a unit volume of rock (or soil), and is a key indicator used to describe the energy consumption level of the tunnel boring machine (TBM) or during tunneling. The soil pressure fluctuation rate index characterizes the stability of the support pressure at the tunnel face. A large soil pressure fluctuation rate index indicates that repeated loading or unloading of the soil by the TBM can easily cause ground fatigue and loosening.

[0120] The first and second linear transformation matrices are both model parameters in a two-stream spatiotemporal attention network.

[0121] The value vector V of the leading line at time t L The definite expression for (t) is: The value vector V of the backward line at time t T The definite expression for (t) is: , Let be the first linear transformation matrix. This is the second linear transformation matrix. This represents the energy vector of the construction disturbance on the pilot line. This represents the energy vector of construction disturbance on the subsequent line.

[0122] After determining the construction disturbance energy vectors of the leading and trailing lines at time t, these vectors are input into a dual-stream spatiotemporal attention network (SPCN), where value vector calculations are performed. The SPCN comprises parallel feature branches for the leading and trailing lines. The value vectors of the leading and trailing lines are determined in their respective feature branches, and both can be determined simultaneously.

[0123] Comprehensive feature vector V all (t) is an eigenvector containing the double-line coupling effect. The surface area during double-line shield tunneling.

[0124] Furthermore, it is determined whether each predicted settlement value is greater than the settlement threshold, and the construction strategy is adjusted when at least one predicted settlement value is greater than the settlement threshold. Further, the construction strategy for the leading line is adjusted when at least one predicted settlement value is greater than the settlement threshold and the surface settlement contribution weight of the leading line is greater than that of the following line; the construction strategy for the following line is adjusted when the predicted settlement value is greater than the settlement threshold and the surface settlement contribution weight of the leading line is less than that of the following line.

[0125] Furthermore, the comprehensive feature vector V based on time t all (t), the predicted surface subsidence values ​​from time t+1 to time t+N include: if the pattern at time t is a dual-line coupled superposition mode, then a hysteresis compensation term is introduced into the comprehensive feature vector, that is, the temporal convolutional network (TCN) is used to capture the rheological hysteresis characteristics of the leading line disturbance in the strata. The predicted surface subsidence values ​​from time t+1 to time t+N are predicted using rheological hysteresis characteristics and integrated eigenvectors. For lagging windows, , V is the hysteresis coefficient that is positively correlated with the clay content of the formation. avg Let z be the average tunneling speed of the tunnel boring machine, and z0 be the burial depth. When predicting ground settlement, since "the current settlement may be caused by excavation before time t", a rheological hysteresis feature is introduced. This feature can extract the disturbance data before time t, calculate a "hysteresis compensation value", and add it to the dual-stream spatiotemporal attention network, thereby improving the accuracy of the prediction.

[0126] In a specific scenario, a comparison example of measured data and predicted data (predicted settlement values) for each ring is shown in the figure below. Figure 7 As shown.

[0127] In this embodiment, the construction disturbance energy vectors of the leading line and the trailing line at time t are determined; the construction disturbance energy vector of the leading line is linearly transformed using a first linear transformation matrix to obtain the value vector of the leading line; the construction disturbance energy vector of the trailing line is linearly transformed using a second linear transformation matrix to obtain the value vector of the trailing line; based on the value vector V of the leading line... L (t) and the weight of surface subsidence contribution, and the value vector V of the backward line. T (t) and the weight of surface subsidence contribution, through Obtain the comprehensive eigenvector V at time t all (t); this allows us to base our comprehensive feature vector V on time t. all (t) predicts the predicted settlement value of the ground surface from time t+1 to time t+N, so as to adjust the construction strategy in time when the predicted settlement value is greater than the settlement threshold.

[0128] In one embodiment, the predicted settlement value is obtained through a two-stream spatiotemporal attention network, and the method further includes:

[0129] Obtain the actual settlement values ​​from time t+1 to time t+N;

[0130] Based on the actual settlement value S from time t+1 to time t+N real and predicted settlement value ,pass Calculate the model loss of a two-stream spatiotemporal attention network. total Loss Weight To punish the loss, Loss Shape For shape loss, and For hyperparameters;

[0131] The model parameters of the dual-stream spatiotemporal attention network are optimized based on model loss, so as to predict the settlement value through the optimized dual-stream spatiotemporal attention network.

[0132] The penalty loss is the sum of the leading line penalty loss and the trailing line penalty loss. The leading line penalty loss is... The penalty loss for the subsequent lane is , Let be the geometric influence coefficient of the leading line at time t. Let be the geometric influence coefficient of the backward line at time t. For any one of the geometric influence coefficients of the excavation work of the pilot line relative to each of the first surface monitoring points. Let be any one of the geometric influence coefficients of the excavation work of the preceding and following lines relative to each of the second surface monitoring points. The penalty loss aims to use the geometric influence coefficients of the preceding and following lines as masks to force the model weights of the non-influenced areas to zero. By introducing the geometric influence coefficients of the preceding and following lines, the dual-stream spatiotemporal attention network can be forced to follow the "spatiotemporal proximity" principle during training, that is, only construction parameters that are spatially close and temporally synchronized have high weights, and the predicted curves corresponding to the predicted settlement values ​​are constrained to conform to the evolution law of Peck's formula in the time domain.

[0133] M represents the total number of surface monitoring points deployed on the ground during the double-shield tunneling construction, and y represents... i Let be the lateral coordinate of surface monitoring point i perpendicular to the axis of the leading / following line (the horizontal distance of surface monitoring point i from the axis of the leading / following line), and Penalty be the penalty function. Shape loss can severely penalize the predicted curve formed by the predicted settlement values ​​from time t+1 to time t+N when it shows an abnormal "upward convex" shape, forcing the predicted curve to always maintain the standard "downward concave" settlement trough shape.

[0134] The optimized dual-stream spatiotemporal attention network can be used to predict land subsidence values ​​at time t+N+1 and thereafter.

[0135] In this embodiment, a total loss function is constructed that includes a data fidelity term (predicted settlement loss), a weight sparsity constraint term (penalty loss), and a morphological consistency constraint term (shape loss). This function guides the dual-stream spatiotemporal attention network to learn the laws of geotechnical mechanics, enabling settlement prediction while adhering to these laws. Furthermore, it ensures that the dual-stream spatiotemporal attention network maintains its predictive capabilities consistent with geotechnical mechanics even in areas with scarce data or abrupt geological changes, avoiding predictions that violate mechanical principles, such as reverse uplift.

[0136] In one embodiment, step S4 includes:

[0137] Determine the construction disturbance energy vectors of the leading line and the trailing line at time t;

[0138] The construction disturbance energy vector of the leading line is linearly transformed by the third linear transformation matrix to obtain the key vector of the leading line; the construction disturbance energy vector of the trailing line is linearly transformed by the fourth linear transformation matrix to obtain the key vector of the trailing line.

[0139] Based on the key vector K of the leading line L (t), First mask matrix M L (t) and the first adaptive query vector Q L (t), through Obtain the first attention matrix Score at time t L (t); based on the key vector K of the backward line T (t), the second mask matrix M T (t) and the second adaptive query vector Q T (t), through Obtain the second attention matrix Score at time t T (t), where d is the dimension of the key vector.

[0140] The construction disturbance energy vector includes the specific energy tunneling index and the soil pressure fluctuation rate index.

[0141] The third and fourth linear transformation matrices are both model parameters in a two-stream spatiotemporal attention network.

[0142] The key vector K of the leading line at time t L The definite expression for (t) is - The key vector K of the backward line at time t T The definite expression for (t) is: , The third linear transformation matrix, This is the fourth linear transformation matrix. This represents the energy vector of the construction disturbance on the pilot line. This represents the energy vector of construction disturbance on the subsequent line.

[0143] After determining the construction disturbance energy vectors of the leading and trailing lines at time t, these vectors are input into a dual-stream spatiotemporal attention network (SPCN), where key vectors are calculated. The SPCN comprises parallel feature branches for the leading and trailing lines. The key vectors for the leading and trailing lines are determined in their respective feature branches, and both can be determined simultaneously.

[0144] The first attention matrix is ​​obtained by performing a dot product operation between the first adaptive query vector and the key vector of the preceding line to obtain data relevance features, and then explicitly superimposing this onto the first mask matrix. The second attention matrix is ​​obtained by performing a dot product operation between the second adaptive query vector and the key vector of the following line to obtain data relevance features, and then explicitly superimposing this onto the second mask matrix. The first attention matrix is ​​determined in the preceding line feature branch, and the second attention matrix is ​​determined in the following line feature branch. The first and second attention matrices can be determined simultaneously.

[0145] In this embodiment, the construction disturbance energy vectors of the leading line and the trailing line at time t are determined; the construction disturbance energy vector of the leading line is linearly transformed using a third linear transformation matrix to obtain the key vector of the leading line; the construction disturbance energy vector of the trailing line is linearly transformed using a fourth linear transformation matrix to obtain the key vector of the trailing line; based on the key vector K of the leading line... L (t), First mask matrix M L (t) and the first adaptive query vector Q L (t), through Obtain the first attention matrix Score at time t L (t); based on the key vector K of the backward line T (t), the second mask matrix M T (t) and the second adaptive query vector Q T (t), through Obtain the second attention matrix Score at time t T (t), so that the settlement contribution weights of the leading line and the trailing line can be determined based on the first attention matrix and the second attention matrix, and thus the excavation line with a greater impact on settlement can be determined based on the settlement contribution weights of the leading line and the trailing line.

[0146] In one embodiment, determining the construction disturbance energy vectors of the leading line and the trailing line at time t includes:

[0147] Obtain the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the tunnel boring machine for both the leading and trailing lines at time t. and standard deviation of earth pressure and weighting coefficients and ;

[0148] Based on the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the pilot line. and standard deviation of earth pressure and weighting coefficients and ,pass Calculate the specific energy excavation index (SEI) of the leading line at time t. L (t), through Calculate the soil pressure volatility index of the leading line. ;

[0149] Based on the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the back track. and standard deviation of earth pressure and weighting coefficients and ,pass Calculate the specific energy excavation index (SEI) of the backward line at time t. L (t), through Calculate the volatility index of soil pressure on the trailing line. ;

[0150] The construction disturbance energy vector of the leading line is determined based on the specific energy tunneling index and the soil pressure volatility index of the leading line; the construction disturbance energy vector of the following line is determined based on the specific energy tunneling index and the soil pressure volatility index of the following line.

[0151] The excavation cross-sectional area is a fixed value. Total thrust, cutterhead torque, tunneling speed, cutterhead rotation speed, and soil chamber pressure can be collected in real time by sensors. Soil chamber pressure is the pressure formed by the excavated soil at various points in the soil chamber in front of the tunnel boring machine (TBM). The mean soil chamber pressure is the average pressure at various points in the soil chamber in front of the TBM, and the standard deviation of soil chamber pressure is the standard deviation of the pressure at various points in the soil chamber in front of the TBM.

[0152] In soft soil, small thrust can cause significant damage; in hard rock, large thrust is common. Therefore, simple thrust or torque values ​​cannot reflect the extent of damage caused by the tunnel boring machine (TBM) to the strata. Thus, total thrust, cutterhead torque, tunneling speed, cutterhead rotation speed, excavation cross-sectional area, mean soil pressure, and standard deviation of soil pressure need to be quantified into a specific energy tunneling index and a soil pressure fluctuation rate index. Only in this way can the extent of damage caused by the TBM to the strata be reflected through these indices.

[0153] The specific energy tunneling index (SETI) represents the total energy consumed by a tunnel boring machine (TBM) in cutting and expelling a unit volume of soil. In complex strata, abrupt changes in the SETI usually indicate changes in the stratum interface or the occurrence of over-excavation / under-excavation. The soil pressure fluctuation rate index can characterize the stability of the face support pressure. Therefore, using the SETI and soil pressure fluctuation rate index as energy vectors of construction disturbance can accurately describe the degree of damage caused by the TBM to the strata.

[0154] Construction disturbance energy vector of the pilot line Represented as .

[0155] Construction disturbance energy vector of the rear line Represented as .

[0156] Furthermore, based on the specific energy excavation index SEI of the leading line at time t... L (t), Earth Warehouse Pressure Volatility Index Cutter head rotation speed n L(t), cutter head torque T L (t), Total thrust F L (t), propulsion speed AR L (t), Penetration Pr L (t), Grouting volume Gv L (t), Earth pressure P L (t), Grouting pressure Gp L (t), cumulative days of downtime St L (t), determine the construction disturbance energy vector of the pilot line. Based on the specific energy excavation index SEI of the backward line at time t T (t), Earth Warehouse Pressure Volatility Index Cutter head rotation speed n T (t), cutter head torque T T (t), Total thrust F T (t), propulsion speed AR T (t), Penetration Pr T (t), Grouting volume Gv T (t), Earth pressure P T (t), Grouting pressure Gp T (t), cumulative days of downtime St T (t), determine the construction disturbance energy vector of the subsequent line. By concatenating the geometric influence coefficients, response sensitivity vectors, and construction disturbance energy vectors of the leading and trailing lines on the time axis, the spatiotemporal coupling characteristic matrices corresponding to the leading and trailing lines are obtained. The spatiotemporal coupling characteristic matrix X of the leading line is... L (t) is The spatiotemporal coupling feature matrix X of the subsequent line T (t) is , For any one of the geometric influence coefficients of the leading line, For any one of the geometric influence coefficients of the trailing line, Let be the response sensitivity vector of the leading line. This is the response sensitivity vector for the trailing line. The cumulative number of downtime days is determined based on the tunneling state vector, and the tunneling state vector for the leading line is... =S L (t), the tunneling state vector of the subsequent line is =S T (t). S L (t)=1 indicates that the tunnel boring machine on the leading line is in a cutting and tunneling operation state at time t (with mechanical energy input), S L (t)=0 indicates that the tunnel boring machine on the pilot line is in a state of shutdown maintenance or segment assembly at time t (mechanical energy input is suspended); S T(t)=1 indicates that the tunnel boring machine in the following line is in a cutting and tunneling operation state (with mechanical energy input) at time t; S T (t)=0 indicates that the tunnel boring machine (TBM) on the subsequent line is in a state of shutdown maintenance or segment assembly at time t (mechanical energy input is suspended). The construction disturbance energy vector of the preceding line is determined based on the specific energy tunneling index, soil pressure fluctuation rate index, cutterhead speed, cutterhead torque, total thrust, propulsion speed, penetration depth, grouting volume, soil pressure, grouting pressure, and cumulative shutdown days at time t. The subsequent line's specific energy tunneling index, soil pressure fluctuation rate index, cutterhead speed, cutterhead torque, total thrust, propulsion speed, and penetration depth Pr are then determined based on these parameters. T (t), grouting volume, soil pressure, grouting pressure, and cumulative number of downtime days are used to determine the construction disturbance energy vector of the subsequent line, which can make the construction disturbance energy vector more in line with the actual situation.

[0157] Furthermore, the spatiotemporal coupling feature matrix of the leading line and the trailing line are encapsulated into a binary dual-stream spatiotemporal feature set X(t). This binary dual-stream spatiotemporal feature set is then input into a dual-stream spatiotemporal attention network, enabling the network to predict settlement values ​​and identify settlement contributions based on this feature set. The expression for the binary dual-stream spatiotemporal feature set is as follows: , This represents the binary encapsulation operation. By encapsulating the spatiotemporal coupling feature matrix of the preceding and following lines, the input requirements of the two-stream spatiotemporal attention network can be met, ensuring synchronous input and physical isolation of the features from both lines at the same time step t. Simultaneously, the binary two-stream spatiotemporal feature set also achieves strict alignment of multi-source data in the time dimension.

[0158] In this embodiment, the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil chamber pressure are used as the basis. and standard deviation of earth pressure and weighting coefficients and Calculate the specific energy tunneling index (SEI) L (t) and the soil pressure fluctuation index can quantify the degree of damage to the strata caused by total thrust, cutterhead torque, tunneling speed, cutterhead rotation speed, excavation cross-sectional area, mean soil pressure and standard deviation of soil pressure. Thus, the settlement contribution weights of the leading line and the trailing line can be accurately identified based on the quantified construction disturbance energy vector.

[0159] In one embodiment, the process of determining the first mask matrix and the second mask matrix in step S3 includes:

[0160] Obtain the preset attenuation enhancement coefficient and target constant;

[0161] Based on attenuation enhancement coefficient Target constant and the set of geometric influence coefficients of the leading line ,pass Determine the first mask matrix M L (t);

[0162] Based on attenuation enhancement coefficient Target constant and the set of geometric influence coefficients of the following lines ,pass Determine the second mask matrix M T (t).

[0163] When the tunnel boring machine is far from the surface monitoring point, the geometric influence coefficient approaches zero, and the elements in the first mask matrix and the second mask matrix approach infinity.

[0164] The set of geometric influence coefficients for the leading line includes the geometric influence coefficients of the excavation work of the leading line at time t relative to each of the first surface monitoring points; the set of geometric influence coefficients for the following line includes the geometric influence coefficients of the excavation work of the following leading line at time t relative to each of the second surface monitoring points.

[0165] One geometric influence coefficient from the set of geometric influence coefficients of the leading line can be used to obtain an element value in the first mask matrix, and so on. Based on each geometric influence coefficient in the set of geometric influence coefficients of the leading line, the value of each element in the first mask matrix can be obtained. Similarly, one geometric influence coefficient from the set of geometric influence coefficients of the trailing line can be used to obtain an element value in the second mask matrix, and so on. Based on each geometric influence coefficient in the set of geometric influence coefficients of the trailing line, the value of each element in the second mask matrix can be obtained.

[0166] In this embodiment, by obtaining a preset attenuation enhancement coefficient and a target constant, the attenuation enhancement coefficient can be used as a basis for... Target constant and the geometric influence coefficient of the leading line ,pass Determine the first mask matrix M L (t), based on the attenuation enhancement coefficient Target constant and the geometric influence coefficient of the following line ,pass Determine the second mask matrix M T (t).

[0167] In one embodiment, the physically guided dual-stream spatiotemporal attention network architecture is shown in the diagram below. Figure 8As shown. Specifically, the geometric influence coefficients, response sensitivity vectors, specific energy tunneling index, and soil pressure fluctuation rate index of the leading and trailing lines are concatenated on the time axis to obtain the spatiotemporal coupling feature matrices corresponding to the leading and trailing lines. These matrices are then input into a dual-stream spatiotemporal attention network. The temporal feature extraction layer in the dual-stream spatiotemporal attention network determines the value vector, key vector, and first adaptive query vector of the leading line, as well as the value vector, key vector, and second adaptive query vector of the trailing line, based on the spatiotemporal coupling feature matrices of the leading and trailing lines. In the physical guidance attention layer, the key vector K of the leading line is used as the basis for further analysis. L (t), First mask matrix M L (t) and the first adaptive query vector Q L (t), through Obtain the first attention matrix Score at time t L (t); based on the key vector K of the backward line T (t), the second mask matrix M T (t) and the second adaptive query vector Q T (t), through Obtain the second attention matrix Score at time t T (t), the formulas for the first attention matrix and the second attention matrix include the dot product operation ( ), scaling operations ( The system uses the first attention matrix and the second attention matrix to obtain and output the settlement contribution weights (dual-line contribution weights) of the leading and trailing lines. Based on the value vectors of the leading and trailing lines, the system then... Weighted summation is performed to obtain a comprehensive feature vector. The comprehensive feature vector is then passed through a feature fusion layer and a fully connected layer to output the predicted settlement value (surface settlement prediction).

[0168] This application also provides an application scenario in which the above-described method for determining the surface settlement contribution during dual-track shield tunneling is applied. Specifically, the application of this method for determining the surface settlement contribution during dual-track shield tunneling in this scenario is as follows:

[0169] like Figure 9As shown, the server obtains geological survey data (strata parameters) from the geological survey report, acquires real-time tunneling parameters of the dual-line shield tunneling machine (total thrust, cutterhead torque, tunneling speed, cutterhead rotation speed, grouting pressure, etc.) and surface monitoring time-series data (historical settlement values, coordinates of surface monitoring points) through sensors, and performs data preprocessing and spatiotemporal alignment on the geological survey data, real-time tunneling parameters of the dual-line shield tunneling machine, and surface monitoring time-series data (step 1). Then, based on the Gaussian-Mindlin hybrid kernel function, the geometric influence coefficients of the leading line and the trailing line are calculated. The geometric influence coefficients of the leading line and the trailing line are called geometric space decay vectors (step 2: physical bias term construction). The physical-guided dual-stream spatiotemporal attention network is used to predict short-term surface settlement and identify the dynamic contribution weights of the dual-line shield tunneling machine (step 3: deep learning network prediction).

[0170] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0171] Based on the same inventive concept, this application also provides a surface settlement contribution determination device for double-line shield tunneling, used to implement the surface settlement contribution determination method for double-line shield tunneling as described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the surface settlement contribution determination device for double-line shield tunneling provided below can be found in the limitations of the surface settlement contribution determination method for double-line shield tunneling described above, and will not be repeated here.

[0172] In one embodiment, such as Figure 10 As shown, a device for determining the surface settlement contribution during dual-track shield tunneling is provided, comprising:

[0173] The data determination module is used to determine the geometric influence coefficient, response sensitivity vector, and equivalent stratum vector of the leading line and the trailing line at time t during the construction of a double-line shield tunnel.

[0174] The splicing module is used to splice the response sensitivity vector and equivalent formation vector of the leading line to obtain the first vector, and to splice the response sensitivity vector and equivalent formation vector of the trailing line to obtain the second vector;

[0175] The vector and matrix determination module is used to obtain the first adaptive query vector of the preceding line based on the first vector; to obtain the second adaptive query vector of the following line based on the second vector; to determine the first mask matrix based on the geometric influence coefficient of the preceding line; and to determine the second mask matrix based on the geometric influence coefficient of the following line.

[0176] The score calculation module is used to obtain a first attention matrix based on a first mask matrix and a first adaptive query vector; and to obtain a second attention matrix based on a second mask matrix and a second adaptive query vector.

[0177] The weight determination module is used to determine the surface settlement contribution weight of the leading line and the subsequent line based on the first attention matrix and the second attention matrix, so as to obtain the predicted settlement value and adjust the construction strategy through the surface settlement contribution weight of the leading line and the subsequent line.

[0178] The various modules in the aforementioned device for determining the surface settlement contribution during dual-track shield tunneling can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0179] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 11 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores various types of data. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer program implements a method for determining surface settlement contribution during dual-shield tunneling.

[0180] Those skilled in the art will understand that Figure 11The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0181] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0182] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0183] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0184] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0185] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0186] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0187] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining the surface settlement contribution during dual-track shield tunneling, characterized in that, The method includes: S1. Determine the geometric influence coefficient, response sensitivity vector, and equivalent stratum vector of the leading line and the trailing line at time t during the construction of a double-line shield tunnel. S2. Concatenate the response sensitivity vector and the equivalent formation vector of the leading line to obtain a first vector, and concatenate the response sensitivity vector and the equivalent formation vector of the trailing line to obtain a second vector; S3. Based on the first vector, obtain the first adaptive query vector of the leading line; based on the second vector, obtain the second adaptive query vector of the trailing line; based on the geometric influence coefficient of the leading line, determine the first mask matrix; based on the geometric influence coefficient of the trailing line, determine the second mask matrix. S4. Based on the first mask matrix and the first adaptive query vector, obtain the first attention matrix; based on the second mask matrix and the second adaptive query vector, obtain the second attention matrix; S5. Based on the first attention matrix and the second attention matrix, determine the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line, so as to obtain the predicted settlement value and adjust the construction strategy through the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line.

2. The method according to claim 1, characterized in that, Step S5 includes: According to the first attention matrix Score L (t) and the second attention matrix Score T (t), through Calculate the first surface subsidence contribution weight matrix of the leading line. ,pass Calculate the second surface subsidence contribution weight matrix of the rear track; The average value of the elements of the first surface subsidence contribution weight matrix is ​​taken as the surface subsidence contribution weight of the leading line, and the average value of the elements of the second surface subsidence contribution weight matrix is ​​taken as the surface subsidence contribution weight of the following line.

3. The method according to claim 1, characterized in that, The process of determining the geometric influence coefficient in step S1 includes: Based on the straight-line distance between each preset first surface monitoring point and the cutterhead of the tunnel boring machine on the leading line, and the vertical distance between the cutterhead of the tunnel boring machine on the following line and the leading line, through... Determine the geometrical influence coefficient of the excavation work of the pilot line relative to each of the first surface monitoring points; Based on the straight-line distance between each preset second surface monitoring point and the cutterhead of the tunnel boring machine on the following line, and the vertical distance between the cutterhead of the tunnel boring machine on the leading line and the following line, through... Determine the geometrical influence coefficient of the excavation work of the subsequent line relative to each of the second surface monitoring points; in, X is the geometrical influence coefficient of the excavation work of the pilot line at time t relative to the first surface monitoring point i. i,L (t) represents the straight-line distance between the first surface monitoring point i and the tunnel boring machine cutterhead of the pilot line at time t, H represents the vertical distance from the centerline of the pilot line to the ground surface, K represents the stratum width coefficient, z0 represents the burial depth of the pilot line, and Y represents the depth of the pilot line. T (t) represents the vertical distance between the tunnel boring machine cutterhead and the leading line at time t; X is the geometrical influence coefficient of the excavation work of the rear track at time t relative to the second surface monitoring point j. j,T (t) represents the straight-line distance between the second surface monitoring point j and the cutterhead of the tunnel boring machine on the backward line at time t. L (t) represents the vertical distance between the tunnel boring machine cutterhead on the leading line and the trailing line at time t.

4. The method according to claim 1, characterized in that, The method of obtaining the predicted subsidence value through the surface subsidence contribution weight of the leading line and the surface subsidence contribution weight of the following line includes: Determine the construction disturbance energy vectors of the leading line and the trailing line at time t; The construction disturbance energy vector of the leading line is linearly transformed by a first linear transformation matrix to obtain the value vector of the leading line; the construction disturbance energy vector of the following line is linearly transformed by a second linear transformation matrix to obtain the value vector of the following line. Based on the value vector V of the leading line L (t) and weight of surface subsidence contribution The value vector V of the following line T (t) and weight of surface subsidence contribution ,pass Obtain the comprehensive eigenvector V at time t all (t); The comprehensive feature vector V based on time t all (t), which predicts the surface subsidence value from time t+1 to time t+N, where N is a non-negative integer.

5. The method according to claim 4, characterized in that, The predicted settlement value is obtained through a dual-stream spatiotemporal attention network, and the method further includes: Obtain the actual settlement values ​​from time t+1 to time t+N; The actual settlement value S based on time t+1 to time t+N real and the predicted settlement value ,pass Calculate the model loss Loss of the dual-stream spatiotemporal attention network. total Loss Weight To punish the loss, Loss Shape For shape loss, and For hyperparameters; The model parameters of the dual-stream spatiotemporal attention network are optimized based on the model loss to predict the settlement value using the optimized dual-stream spatiotemporal attention network.

6. The method according to claim 1, characterized in that, Step S4 includes: Determine the construction disturbance energy vectors of the leading line and the trailing line at time t; The construction disturbance energy vector of the leading line is linearly transformed by the third linear transformation matrix to obtain the key vector of the leading line; the construction disturbance energy vector of the following line is linearly transformed by the fourth linear transformation matrix to obtain the key vector of the following line. Based on the key vector K of the preceding line L (t), the first mask matrix M L (t) and the first adaptive query vector Q L (t), through Obtain the first attention matrix Score at time t L (t); based on the key vector K of the following line T (t), the second mask matrix M T (t) and the second adaptive query vector Q T (t), through Obtain the second attention matrix Score at time t T (t), where d is the dimension of the key vector.

7. The method according to claim 6, characterized in that, Determining the construction disturbance energy vectors of the leading line and the trailing line at time t includes: Obtain the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the tunnel boring machine for the leading and trailing lines at time t. and standard deviation of earth pressure and weighting coefficients and ; Based on the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the aforementioned pilot line. and standard deviation of earth pressure and weighting coefficients and ,pass Calculate the specific energy excavation index (SEI) of the leading line at time t. L (t), through Calculate the soil pressure volatility index of the leading line. ; Based on the total thrust F(t), cutterhead torque T(t), tunneling speed v(t), cutterhead rotation speed n(t), excavation cross-sectional area A, and average soil pressure of the aforementioned backward travel path. and standard deviation of earth pressure and weighting coefficients and ,pass Calculate the specific energy excavation index (SEI) of the backward travel line at time t. T (t), through Calculate the volatility index of soil pressure on the trailing line. ; The construction disturbance energy vector of the leading line is determined based on the specific energy tunneling index and the soil pressure fluctuation rate index of the leading line; the construction disturbance energy vector of the following line is determined based on the specific energy tunneling index and the soil pressure fluctuation rate index of the following line.

8. The method according to claim 1, characterized in that, The process of determining the first and second mask matrices in step S3 includes: Obtain the preset attenuation enhancement coefficient and target constant; Based on the attenuation enhancement coefficient The target constant and the set of geometric influence coefficients of the leading line. ,pass Determine the first mask matrix M L (t); Based on the attenuation enhancement coefficient The target constant and the set of geometric influence coefficients of the following line. ,pass Determine the second mask matrix M T (t).

9. A device for determining the surface settlement contribution during dual-track shield tunneling, used to execute the method according to any one of claims 1-8, characterized in that, The device includes: The data determination module is used to determine the geometric influence coefficient, response sensitivity vector, and equivalent stratum vector of the leading line and the trailing line at time t during the construction of a double-line shield tunnel. The splicing module is used to splice the response sensitivity vector and the equivalent formation vector of the leading line to obtain a first vector, and to splice the response sensitivity vector and the equivalent formation vector of the trailing line to obtain a second vector; A vector and matrix determination module is used to obtain a first adaptive query vector for the preceding line based on the first vector; obtain a second adaptive query vector for the following line based on the second vector; determine a first mask matrix based on the geometric influence coefficient of the preceding line; and determine a second mask matrix based on the geometric influence coefficient of the following line. The score calculation module is used to obtain a first attention matrix based on the first mask matrix and the first adaptive query vector; and to obtain a second attention matrix based on the second mask matrix and the second adaptive query vector. The weight determination module is used to determine the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line based on the first attention matrix and the second attention matrix, so as to obtain the predicted settlement value and adjust the construction strategy through the surface settlement contribution weight of the leading line and the surface settlement contribution weight of the following line.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images