Large-diameter shield non-uniform grouting amount setting method and system considering settlement development process

Abstract

The invention relates to the technical field of shield engineering, in particular to a large-diameter shield non-uniform grouting amount setting method and system considering the settlement development process, and the method comprises the following steps: collecting construction data of historical engineering; obtaining a micro working condition database matched with the current working condition from the construction data of the historical engineering by utilizing a clustering algorithm; using variational mode decomposition to extract ground surface settlement development information; non-uniform grouting parameters are extracted from the micro working condition database; constructing a ground surface settlement prediction model by using ground surface settlement development information, non-uniform grouting parameters and the micro-working condition database; in the actual shield tunneling process, the ground surface settlement prediction model is used for predicting ground surface settlement of the current shield tail ring to obtain a prediction result, the particle swarm optimization is used for optimizing the non-uniform grouting amount, and then the non-uniform grouting amount of the next ring is generated. According to the method, the ground surface settlement development process is extracted by using the deep learning model, the consideration range in practical application is wider, and the universality is higher.

Description

Technical Field

[0001] The invention relates to the technical field of shield engineering, and in particular to a method and system for setting uneven grouting amount of a large-diameter shield taking into account a settlement development process. Background Art

[0002] Shield tunnel construction is an important means to alleviate urban road congestion and infrastructure shortage. It has high requirements for the control of surface settlement, among which the synchronous grouting volume is a key parameter. Compared with small-diameter shield tunnels (diameter <6m), large-diameter shield tunnels have large excavation surfaces, more exposed strata, poor excavation control, and the gap between the pipe segments and the surrounding soil is often greater than the design value. Therefore, the synchronous grouting control of large-diameter tunnels is more complicated. The grouting holes of large-diameter shield tunnels can be divided into three parts: upper, middle, and lower according to their locations. In actual construction, the grouting volumes of these three parts are often different and are often set based on manual experience. This method relies too much on the construction experience of technicians, and it is difficult to adjust parameters in time when the geological conditions are complex and changeable; unreasonable uneven grouting volume will also affect surface settlement and reduce construction quality. Therefore, it is necessary to predict the uneven grouting volume through reasonable methods to guide construction and avoid accidents. Summary of the invention

[0003] The purpose of the present invention is to overcome the defects of the prior art and provide a method and system for setting the uneven grouting amount of a large-diameter shield taking into account the settlement development process, so as to solve the problem that the existing uneven grouting parameters rely on the construction experience of technicians and unreasonable parameters will also affect the surface settlement and reduce the construction quality.

[0004] The technical solution to achieve the above purpose is: The present invention provides a method for setting uneven grouting amount of a large-diameter shield taking into account the settlement development process, comprising the following steps: Collect construction data from historical projects; According to the current construction conditions, a clustering algorithm is used to obtain a matching micro-condition database from the collected construction data of historical projects; Based on the surface settlement data in the micro-condition database, variational mode decomposition is used to extract the surface settlement development information. Extracting uneven grouting parameters from the micro-condition database; Constructing a surface settlement prediction model using the extracted surface settlement development information, uneven grouting parameters and the micro-condition database; During the actual tunneling process of the shield machine, the constructed surface settlement prediction model is used to predict the surface settlement of the current shield tail ring to obtain the prediction result. Based on the prediction result, the particle swarm algorithm is used to optimize the uneven grouting amount, and then the uneven grouting amount of the next ring is generated.

[0005] The present invention further improves the method for setting uneven grouting amount of large-diameter shield considering the settlement development process, and further includes clustering algorithm analysis of the collected construction data of historical projects: S11, define the domain radius parameter and the minimum number of sample points parameter, and initialize all points to an unvisited state; S12, randomly select an unvisited point P from the collected historical engineering construction data. If the number of points in the domain of P is greater than or equal to the minimum sample point number parameter, P is marked as a core point, a new cluster is created, all points in the domain of P are expanded, and points that meet the density requirements are added to the new cluster; if the number of points in the domain of P is less than the minimum sample point number parameter, P is marked as a noise point; S13, check the domains of all points in the current cluster. If a point is a core point and has not been visited, add the points in its domain to the current cluster; if a point is a boundary point, assign it to the current cluster without expanding it. Repeat this step until the current cluster cannot be expanded any further. S14, select the next point from the remaining unvisited points, and repeat steps S12 and S13 until all points are processed; S15, outputting sample points of the same cluster as the same micro-operating condition database, wherein points not assigned to any cluster are marked as noise points.

[0006] The further improvement of the method for setting uneven grouting amount of a large-diameter shield considering the settlement development process of the present invention is that the step of extracting surface settlement development information using variational mode decomposition includes: Initialize the candidate range of modal number K [1,2,3,…,N]; Perform variational mode decomposition to decompose the original surface settlement data into K modes IMF1, IMF2, …, IMF k ; The superposition of modal components gives ; Calculate the decomposed signal With the original signal The error between ,in is the Euclidean norm of the signal; The optimal mode number K is determined using the elbow method; The original surface settlement data are extracted by variational mode decomposition to obtain 1 trend component and K-1 fluctuation components.

[0007] The present invention further improves the method for setting uneven grouting amount of a large-diameter shield considering the settlement development process in that the steps of constructing a surface settlement prediction model using the extracted surface settlement development information, uneven grouting parameters and the micro-condition database include: The surface settlement development information extracted by using the modified mode decomposition is input into the GRU model to extract the time series data of surface settlement development; Extracting uneven grouting parameters, tunnel geometry parameters and geological parameters from the micro-condition database using a BP neural network model; The extracted surface settlement development time series data, uneven grouting parameters, tunnel geometric parameters and geological parameters are used as the input of the deep neural network model, and the final settlement of the shield tail ring is used as the output of the deep neural network model. The deep neural network model is trained and then a surface settlement prediction model is constructed.

[0008] The further improvement of the method for setting uneven grouting amount of a large-diameter shield considering the settlement development process of the present invention is that the step of optimizing the uneven grouting amount by using a particle swarm algorithm based on the obtained prediction results includes: Initialize the candidate set of tunneling parameters and set the uneven grouting volume x=[x 上 ,x 中 ,x 下 ], adjust the range , randomly initialize m particles within the adjustment range, and the position of each particle corresponds to an uneven grouting value , the adjustment speed of each particle is ; Calculate the individual and global optimal solutions. For each particle, calculate its fitness according to the objective function, and set the initial position to its optimal position. At the same time, record the minimum value of the optimal positions of all individuals in the group as the global optimal solution, where the objective function is ;

[0009] In the above formula, The final settlement of the surface of the shield tail ring, are other grouting parameters except uneven grouting amount, is the geological parameter corresponding to the shield tail ring, are the tunnel geometry parameters;

[0010] Update the speed and position of each particle according to the following formula: ,

[0011] In the above formula, is the velocity of the particle in the next time step, is the inertia weight, and is the learning factor, and is a random number between [0,1], is the position of particle i at time t, is the optimal position found between particle i and time t; is the global optimal position;

[0012] Update the candidate set of uneven grouting volume, and update the candidate set of uneven grouting volume at time t according to the updated particle speed and position Updated to the candidate set of uneven grouting volume at time t+1 ;

[0013] Update individual and global optimal solutions, calculate the objective function value of each element in the excavation parameter candidate set according to the updated excavation parameter candidate set, and update the individual optimal solution of each element and the global optimal solution of the element set;

[0014] The steps of updating the candidate set of uneven grouting amount and updating the individual and global optimal solutions are repeated until the iteration termination condition is reached, and the optimized uneven grouting amount is output.

[0015] The present invention also provides a large-diameter shield uneven grouting amount setting system considering the settlement development process, comprising:

[0016] A general database for storing the collected construction data of historical projects;

[0017] A cluster analysis unit connected to the master database is used to obtain a matching micro-condition database from the collected construction data of historical projects using a clustering algorithm according to the current construction conditions;

[0018] A settlement extraction unit, connected to the cluster analysis unit, is used to extract surface settlement development information using variational mode decomposition for the surface settlement data in the obtained micro-condition database;

[0019] A parameter extraction unit, connected to the cluster analysis unit, for extracting uneven grouting parameters from the micro-condition database;

[0020] A model building unit, connected to the cluster analysis unit, the settlement extraction unit and the parameter extraction unit, for building a surface settlement prediction model using the extracted surface settlement development information, the non-uniform grouting parameters and the micro-condition database;

[0021] The processing unit is connected to the model building unit and is used to predict the surface settlement of the current shield tail ring using the constructed surface settlement prediction model during the actual shield tunneling process to obtain a prediction result, and optimize the uneven grouting amount using a particle swarm algorithm based on the obtained prediction result, thereby generating an uneven grouting amount for the next ring.

[0022] A further improvement of the large-diameter shield uneven grouting amount setting system considering the settlement development process of the present invention is that the cluster analysis unit is also used to perform clustering algorithm analysis on the collected construction data of historical projects to obtain multiple micro-condition databases.

[0023] The further improvement of the large-diameter shield uneven grouting amount setting system considering the settlement development process of the present invention is that the settlement extraction unit uses variational mode decomposition to extract surface settlement development information including:

[0024] Initialize the candidate range of modal number K [1,2,3,…,N];

[0025] Perform variational mode decomposition to decompose the original surface settlement data into K modes IMF1, IMF2, …, IMF k ;

[0026] The superposition of modal components gives ;

[0027] Calculate the decomposed signal With the original signal The error between ,in is the Euclidean norm of the signal;

[0028] The optimal mode number K is determined using the elbow method;

[0029] The original surface settlement data are extracted by variational mode decomposition to obtain 1 trend component and K-1 fluctuation components.

[0030] The further improvement of the large-diameter shield uneven grouting amount setting system considering the settlement development process of the present invention is that the model building unit includes a GRU model, a BP neural network model, a deep neural network model and a training model module;

[0031] The GRU model is connected to the settlement extraction unit, and is used to receive the surface settlement development information extracted by the settlement extraction unit, and extract the surface settlement development time series data;

[0032] The BP neural network model is used to extract uneven grouting parameters, tunnel geometry parameters and geological parameters from the micro-condition database;

[0033] The training model module is connected to the GRU model, the BP neural network model and the deep neural network model, and is used to use the extracted surface settlement development time series data, uneven grouting parameters, tunnel geometric parameters and geological parameters as the input of the deep neural network model, and the final settlement of the shield tail ring as the output of the deep neural network model to train the deep neural network model, and then construct a surface settlement prediction model.

[0034] A further improvement of the system for setting uneven grouting amount of a large-diameter shield considering the settlement development process of the present invention is that the processing unit is also used for:

[0035] Initialize the candidate set of tunneling parameters and set the uneven grouting volume x=[x 上 ,x 中 ,x 下 ], adjust the range , randomly initialize m particles within the adjustment range, and the position of each particle corresponds to an uneven grouting value , the adjustment speed of each particle is ;

[0036] Calculate the individual and global optimal solutions. For each particle, calculate its fitness according to the objective function, and set the initial position to its optimal position. At the same time, record the minimum value of the optimal positions of all individuals in the group as the global optimal solution, where the objective function is ;

[0037] In the above formula, The final settlement of the surface of the shield tail ring, are other grouting parameters except uneven grouting amount, is the geological parameter corresponding to the shield tail ring, are the tunnel geometry parameters;

[0038] Update the speed and position of each particle according to the following formula: ,

[0039] In the above formula, is the velocity of the particle in the next time step, is the inertia weight, and is the learning factor, and is a random number between [0,1], is the position of particle i at time t, is the optimal position found between particle i and time t; is the global optimal position;

[0040] Update the candidate set of uneven grouting volume, and update the candidate set of uneven grouting volume at time t according to the updated particle speed and position Updated to the candidate set of uneven grouting volume at time t+1 ;

[0041] Update individual and global optimal solutions, calculate the objective function value of each element in the excavation parameter candidate set according to the updated excavation parameter candidate set, and update the individual optimal solution of each element and the global optimal solution of the element set;

[0042] The steps of updating the candidate set of uneven grouting amount and updating the individual and global optimal solutions are repeated until the iteration termination condition is reached, and the optimized uneven grouting amount is output.

[0043] The beneficial effects of the method and system for setting uneven grouting amount of a large-diameter shield considering the settlement development process of the present invention are as follows:

[0044] The present invention collects the construction data of large-diameter shield tunnels under different working conditions, and divides the micro-condition database according to different working conditions; establishes the GRU-BP model to extract the development process of surface settlement, constructs the relationship between the surface settlement of the shield tail, geological parameters, tunnel geometric parameters and uneven grouting amount, and realizes the accurate prediction of the final surface settlement; establishes a multi-objective optimization strategy for uneven grouting amount based on the PSO algorithm (particle swarm algorithm), realizes the real-time optimization of uneven grouting amount in various parts during shield excavation, and is more suitable for actual construction operations.

[0045] The present invention uses a deep learning model to extract the development process of surface subsidence, which takes into account a wider range of practical applications and is more versatile.

[0046] The micro-operating condition database established by the present invention is trained for a specific operating condition of a large-diameter shield tunnel, and is more targeted in practical applications.

[0047] The multi-objective optimization strategy for uneven grouting amount based on the PSO algorithm of the present invention integrates the surface settlement prediction model, and can provide the optimization value of uneven grouting amount in various parts during shield tunneling, which has greater guiding significance for actual construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention is a flow chart of a method for setting uneven grouting amount of a large-diameter shield taking into account the settlement development process.

[0049] Figure 2 The present invention is a method for setting uneven grouting volume of a large-diameter shield taking into account the settlement development process and a schematic diagram of the process of constructing a micro-condition database in the system.

[0050] Figure 3The present invention discloses a method for setting uneven grouting amount of a large-diameter shield taking into account the settlement development process and a variational modal decomposition settlement development process diagram in the system.

[0051] Figure 4 This is a network structure diagram of a large-diameter shield uneven grouting amount setting method and a surface settlement prediction model in the system in consideration of the settlement development process of the present invention.

[0052] Figure 5 This is a multi-objective optimization flow chart of uneven grouting amount based on PSO in the method for setting uneven grouting amount of large-diameter shield considering the settlement development process of the present invention. DETAILED DESCRIPTION

[0053] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0054] See also Figure 1 The present invention provides a method and system for setting uneven grouting amount of a large-diameter shield taking into account the development process of settlement, which is used to automatically set the uneven grouting amount of each ring according to the change of surface settlement during the shield excavation process (the uneven grouting amount includes the grouting amount of the upper grouting hole position, the grouting amount of the middle grouting hole position, and the grouting amount of the lower grouting hole position). First, according to the historical shield geometry data and geological data, the data set under each micro-condition is divided; then, the corresponding micro-condition database is matched from the total database for the project under construction, and the surface settlement development time series data is extracted using GRU, and the uneven grouting characteristics, tunnel geometry characteristics and geological characteristics are extracted using BP neural network to construct a surface settlement prediction model; finally, the uneven grouting amount optimization value of each ring is obtained through the PSO-based uneven grouting amount optimization strategy. The present invention takes into account the uneven grouting amount of the upper, middle and lower three different parts, and is more suitable for the grouting amount control of large-diameter shield tunnels, which is conducive to on-site construction adjustments. The present invention comprehensively considers the development process of surface settlement and the changes in tunnel construction parameters, and can dynamically adjust the uneven grouting amount in real time during the shield construction process. The following is a description of the method and system for setting uneven grouting amount of a large-diameter shield considering the development process of settlement in conjunction with the accompanying drawings.

[0055] A large-diameter shield uneven grouting amount setting system considering the settlement development process of the present invention comprises a general database, a cluster analysis unit, a settlement extraction unit, a parameter extraction unit, a model building unit and a processing unit; wherein the general database is used to store the collected construction data of historical projects, and the collected construction data of historical projects include surface settlement data, uneven grouting data, shield geometry data and geological data in the historical projects. Specifically, the surface settlement data includes the cumulative settlement data monitored by the measuring point before the shield tail is out of the measuring point and the cumulative settlement data (i.e., the final settlement) monitored by the measuring point one week after the shield tail is out of the measuring point. The uneven grouting data includes the upper grouting amount, the middle grouting amount, the lower grouting amount and the slurry slump. The shield geometry data includes the shield diameter and the tunnel burial depth. The geological data includes the weight of the overburden layer, the cohesion of the excavated soil layer, the internal friction angle of the excavated soil layer and the overburden compression modulus. Preferably, the collected surface settlement data, uneven grouting data, shield geometry data and geological data of historical projects constitute a data set and are stored in a general database.

[0056] The cluster analysis unit is connected to the overall database. The cluster analysis unit is used to obtain a matching micro-condition database from the collected historical engineering construction data based on the current construction conditions using a clustering algorithm. The impact of shield tunneling on surface settlement under different geological conditions varies greatly. Before building a surface settlement prediction model, it is necessary to first match historical data that is relatively close to the current construction conditions from the overall database. Specifically, Figure 2 As shown, by using the DBSCAN clustering algorithm to perform cluster analysis on the data sets in the total database, the distance between samples is calculated using the Euclidean distance, the minimum number of sample points is determined by twice the data set dimension, and the field radius is determined by the elbow method, and finally m micro-condition databases are obtained. m is a positive integer greater than 1. In the division of the micro-condition database, the geological data of the current shield tail ring and the tunnel geometry data are input into the cluster analysis unit, which can automatically match the corresponding micro-condition database from the total database.

[0057] The settlement extraction unit is connected to the cluster analysis unit. The settlement extraction unit is used to extract surface settlement development information using variational mode decomposition (VMD) for the surface settlement data in the obtained micro-condition database. After matching the micro-condition database, it is necessary to use VMD to extract the settlement trend signal and fluctuation signal from the cumulative settlement data monitored before the shield tail escapes from the measuring point, so as to explore the settlement development process.

[0058] The parameter extraction unit is connected with the cluster analysis unit, and is used to extract the non-uniform grouting parameters from the micro-condition database, and the extracted non-uniform grouting parameters are the data used for model training.

[0059] The model building unit is connected with the cluster analysis unit, the settlement extraction unit and the parameter extraction unit, and the model building unit is used to build a surface settlement prediction model using the extracted surface settlement development information, uneven grouting parameters and micro-condition database;

[0060] The processing unit is connected to the model building unit, which is used to predict the surface settlement of the current shield tail ring using the constructed surface settlement prediction model during the actual shield excavation process to obtain a prediction result, and optimize the uneven grouting amount using the particle swarm algorithm based on the obtained prediction result, thereby generating the uneven grouting amount of the next ring.

[0061] In a specific implementation of the present invention, the cluster analysis unit is also used to perform clustering algorithm analysis on the collected construction data of historical projects to obtain multiple micro-condition databases.

[0062] Specifically, the specific steps of the cluster analysis unit dividing the micro-operating condition database include:

[0063] S11, define the field radius parameter (eps) and the minimum number of sample points parameter (min_samples), and initialize all points to an unvisited state;

[0064] S12, randomly select an unvisited point P from the collected historical construction data. If the number of points in the domain of P is greater than or equal to the minimum number of sample points parameter, P is marked as a core point, a new cluster is created, all points in the domain of P are expanded, and points that meet the density requirements are added to the new cluster; if the number of points in the domain of P is less than the minimum number of sample points parameter, P is marked as a noise point (temporary); if the subsequent expansion step finds that P belongs to a cluster, P is reallocated;

[0065] S13, check the domains of all points in the current cluster (starting from the core point). If a point is a core point and has not been visited, add the points in its domain to the current cluster; if a point is a boundary point, assign it to the current cluster without expanding it. Repeat this step until the current cluster cannot be expanded any further.

[0066] S14, select the next point from the remaining unvisited points, and repeat steps S12 and S13 until all points are processed;

[0067] S15, outputting sample points of the same cluster as the same micro-operating condition database, wherein points not assigned to any cluster are marked as noise points.

[0068] like Figure 1 As shown, the construction data of the historical projects in the overall database are continuously updated data. After the update, the cluster analysis unit of the present invention can re-divide the construction data in the overall database into micro-condition databases.

[0069] In a specific embodiment of the present invention, Figure 3 As shown, the settlement extraction unit uses the signal restoration error to determine the optimal mode number K. Specifically, the settlement extraction unit uses variational mode decomposition to extract the surface settlement development information including:

[0070] Initialize the candidate range of modal number K [1,2,3,…,N];

[0071] Perform variational mode decomposition to decompose the original surface settlement data into K modes IMF1, IMF2, …, IMF k ;

[0072] The superposition of modal components gives ;

[0073] Calculate the decomposed signal With the original signal The error between ,in is the Euclidean norm of the signal;

[0074] The optimal mode number K is determined using the elbow method;

[0075] The original surface settlement data are extracted by variational mode decomposition to obtain 1 trend component and K-1 fluctuation components.

[0076] In a specific embodiment of the present invention, Figure 4 As shown, the model building unit includes a GRU model, a BP neural network model, a deep neural network model and a training model module;

[0077] The GRU model is connected to the settlement extraction unit, and is used to receive the surface settlement development information extracted by the settlement extraction unit, and extract the surface settlement development time series data;

[0078] The BP neural network model is used to extract uneven grouting parameters, tunnel geometry parameters and geological parameters from the micro-condition database; the BP neural network model is the parameter extraction unit in the system of the present invention;

[0079] The training model module is connected with the GRU model, the BP neural network model and the deep neural network model. The training model module is used to use the extracted surface settlement development time series data, uneven grouting parameters, tunnel geometric parameters and geological parameters as the input of the deep neural network model, and the final settlement of the shield tail ring as the output of the deep neural network model to train the deep neural network model and then construct a surface settlement prediction model.

[0080] The present invention uses GRU to extract the time series data of surface settlement development, and uses BP neural network to extract uneven grouting characteristics, tunnel geometric characteristics and geological characteristics, and outputs the final settlement of the shield tail ring, wherein the number of network layers of GRU is 2 layers, the number of neurons is 32, and the discard rate is 0.2; the number of network layers of BP neural network is 2 layers, and the number of neurons is 16. The Adam optimizer is used and the stochastic gradient descent method is used to update the initial weight of the network, and the initial learning rate is set to 0.01.

[0081] In a specific embodiment of the present invention, Figure 5 As shown, after the surface settlement prediction model is constructed, the processing unit of the present invention adopts a multi-objective optimization strategy for uneven grouting volume based on the PSO algorithm. By integrating the PSO algorithm into the surface settlement prediction model, the upper, middle and lower uneven grouting volumes can be obtained respectively. The specific steps include:

[0082] Initialize the candidate set of tunneling parameters and set the uneven grouting volume x=[x 上 ,x 中 ,x 下 ], adjust the range , randomly initialize m particles within the adjustment range, and the position of each particle corresponds to an uneven grouting value , the adjustment speed of each particle is ;

[0083] Calculate the individual and global optimal solutions. For each particle, calculate its fitness according to the objective function, and set the initial position to its optimal position. At the same time, record the minimum value of the optimal positions of all individuals in the group as the global optimal solution, where the objective function is ;

[0084] In the above formula, The final settlement of the surface of the shield tail ring, are other grouting parameters except uneven grouting amount, is the geological parameter corresponding to the shield tail ring, are the tunnel geometry parameters;

[0085] Update the speed and position of each particle according to the following formula: ,

[0086] In the above formula, is the velocity of the particle in the next time step, is the inertia weight, and is the learning factor, and is a random number between [0,1], is the position of particle i at time t, is the optimal position found between particle i and time t; is the global optimal position;

[0087] Update the candidate set of uneven grouting volume, and update the candidate set of uneven grouting volume at time t according to the updated particle speed and position Updated to the candidate set of uneven grouting volume at time t+1 ;

[0088] Update individual and global optimal solutions, calculate the objective function value of each element in the excavation parameter candidate set according to the updated excavation parameter candidate set, and update the individual optimal solution of each element and the global optimal solution of the element set;

[0089] Repeat the steps of updating the candidate set of uneven grouting amount and updating the individual and global optimal solutions until the iteration termination condition is reached, and output the optimized uneven grouting amount. The iteration termination condition may be that the surface settlement reaches a suitable range or the number of iterations reaches an upper limit.

[0090] The large-diameter shield uneven grouting amount setting system considering the settlement development process of the present invention is an application software, which can be connected to the control system of the shield machine, and can obtain the real-time construction parameters in the shield tunneling construction process from the control system of the shield machine, and input the real-time construction parameters into the surface settlement prediction model to predict the current surface settlement, and obtain the prediction record, wherein the real-time construction parameters include grouting parameters, geometric parameters, formation parameters and the cumulative settlement parameters monitored before the shield tail escapes from the measuring point, and the settlement development information of the accumulated settlement parameters is extracted by the settlement extraction unit, and the settlement development information is used to input into the surface settlement prediction model for settlement prediction. Based on the prediction results, the particle swarm algorithm (PSO) is applied to optimize the uneven grouting amount to generate the optimal parameters for the next ring of uneven grouting.

[0091] The present invention also provides a method for setting the uneven grouting amount of a large-diameter shield taking into account the settlement development process, and the setting method is described below.

[0092] like Figure 1 As shown, the method for setting uneven grouting amount of a large-diameter shield considering the settlement development process of the present invention comprises the following steps: Collect construction data from historical projects; According to the current construction conditions, a clustering algorithm is used to obtain a matching micro-condition database from the collected construction data of historical projects; Based on the surface settlement data in the micro-condition database, variational mode decomposition is used to extract the surface settlement development information. Extracting non-uniform grouting parameters from the micro-condition database; The surface settlement prediction model is constructed using the extracted surface settlement development information, non-uniform grouting parameters and micro-condition database; During the actual tunneling process of the shield machine, the constructed surface settlement prediction model is used to predict the surface settlement of the current shield tail ring to obtain the prediction result. Based on the prediction result, the particle swarm algorithm is used to optimize the uneven grouting amount, and then the uneven grouting amount of the next ring is generated.

[0093] The setting method of the present invention mainly includes four steps: the first is historical data collection and classification, the second is database matching and settlement development process extraction, the third is surface settlement prediction model construction, and the fourth is providing uneven grouting amount optimization decision. First, the surface monitoring data, uneven grouting parameters and geological parameters in the historical construction data are extracted; then, the DB-SCAN algorithm is used to perform cluster analysis on the overall data set according to the geological parameters, and the micro-condition data sets under different geological conditions are divided, and the surface settlement development information is extracted using variational mode decomposition (VMD) for the surface settlement data under each micro-condition; the gated recurrent unit network (GRU) is used to further extract the surface settlement development characteristics, and the uneven grouting parameters and geological parameters extracted by the BP neural network are combined. The above features are spliced ​​and input into the deep learning network model to establish a prediction model that comprehensively considers the development of surface settlement. Finally, when the model is used in the actual shield excavation process, the historical settlement data of the shield tail ring is extracted and modal decomposition is performed using VMD. The decomposition results, the uneven grouting parameters of the shield tail ring and the geological parameters are input into the surface settlement prediction model to obtain the predicted value of the shield tail ring surface settlement. Based on the prediction results, the particle swarm algorithm (PSO) is used to optimize the uneven grouting amount and generate the optimal parameters for the next ring of uneven grouting.

[0094] In a specific embodiment of the present invention, the construction data of historical projects collected include surface settlement data, uneven grouting data, shield geometry data and geological data in the historical projects. Specifically, the surface settlement data include the cumulative settlement data monitored by the measuring point before the shield tail is out of the measuring point and the cumulative settlement data (i.e., the final settlement) monitored by the measuring point one week after the shield tail is out of the measuring point. The uneven grouting data include the upper grouting amount, the middle grouting amount, the lower grouting amount and the slump of the slurry. The shield geometry data include the shield diameter and the tunnel burial depth. The geological data include the weight of the overburden layer, the cohesion of the excavated soil layer, the internal friction angle of the excavated soil layer and the overburden compression modulus.

[0095] In a specific implementation of the present invention, it also includes performing clustering algorithm analysis on the collected construction data of historical projects: S11, define the domain radius parameter and the minimum number of sample points parameter, and initialize all points to an unvisited state; S12, randomly select an unvisited point P from the collected historical engineering construction data. If the number of points in the domain of P is greater than or equal to the minimum sample point number parameter, P is marked as a core point, a new cluster is created, all points in the domain of P are expanded, and points that meet the density requirements are added to the new cluster; if the number of points in the domain of P is less than the minimum sample point number parameter, P is marked as a noise point; S13, check the domains of all points in the current cluster. If a point is a core point and has not been visited, add the points in its domain to the current cluster; if a point is a boundary point, assign it to the current cluster without expanding it. Repeat this step until the current cluster cannot be expanded any further. S14, select the next point from the remaining unvisited points, and repeat steps S12 and S13 until all points are processed; S15, outputting sample points of the same cluster as the same micro-operating condition database, wherein points not assigned to any cluster are marked as noise points.

[0096] The impact of shield machine excavation on surface settlement under different geological conditions is quite different, so before building a surface settlement prediction model, it is necessary to first match historical data that is relatively close to the current construction conditions from the total database. Figure 2 As shown in the figure, by using the DBSCAN clustering algorithm to cluster the geological data and tunnel geometry data in the total database, the distance between samples is calculated using the Euclidean distance, the minimum number of sample points is determined by twice the dataset dimension, and the neighborhood radius is determined by the elbow method, and finally m micro-condition databases are obtained. After the micro-condition database is divided, the geological data and tunnel geometry data of the current shield tail ring only need to be input into the DBSCAN clustering algorithm to automatically match the corresponding micro-condition database from the total database.

[0097] In a specific embodiment of the present invention, after matching the corresponding micro-condition database, it is necessary to use VMD to extract the trend signal and fluctuation signal of the settlement from the cumulative settlement data monitored before the shield tail escapes from the measuring point, and explore the settlement development process. Figure 3 As shown, the optimal mode number K is determined using the signal restoration error. Specifically, the steps of extracting the surface settlement development information using variational mode decomposition include: Initialize the candidate range of modal number K [1,2,3,…,N]; Perform variational mode decomposition to decompose the original surface settlement data into K modes IMF1, IMF2, …, IMF k ; The superposition of modal components gives ; Calculate the decomposed signal With the original signal The error between ,in is the Euclidean norm of the signal; The optimal mode number K is determined using the elbow method; The original surface settlement data are extracted by variational mode decomposition to obtain 1 trend component and K-1 fluctuation components.

[0098] In a specific embodiment of the present invention, the steps of constructing a surface settlement prediction model using the extracted surface settlement development information, uneven grouting parameters and micro-condition database include: The surface settlement development information extracted by using the modified mode decomposition is input into the GRU model to extract the time series data of surface settlement development; The BP neural network model is used to extract the non-uniform grouting parameters, tunnel geometry parameters and geological parameters from the micro-condition database; The extracted surface settlement development time series data, uneven grouting parameters, tunnel geometric parameters and geological parameters are used as the input of the deep neural network model, and the final settlement of the shield tail ring is used as the output of the deep neural network model. The deep neural network model is trained and then a surface settlement prediction model is constructed.

[0099] The present invention uses GRU to extract the time series data of surface settlement development, and uses BP neural network to extract the uneven grouting characteristics, tunnel geometric characteristics and geological characteristics, and outputs the final settlement of the shield tail ring. The specific characteristics are consistent with the characteristics of the construction data collected from historical projects, and will not be repeated here. The model structure diagram is shown in Figure 4 As shown in the figure, the GRU has 2 layers, 32 neurons, and a dropout rate of 0.2; the BP neural network has 2 layers and 16 neurons. The Adam optimizer is used to update the initial network weights using the stochastic gradient descent method, and the initial learning rate is set to 0.01.

[0100] In a specific embodiment of the present invention, after the surface settlement prediction model is constructed, the present invention adopts a multi-objective optimization strategy of uneven grouting amount based on the PSO algorithm, such as Figure 5 As shown in FIG. 1 , by integrating the PSO algorithm into the surface settlement prediction model, the uneven grouting amounts of the upper, middle and lower parts can be obtained respectively. Specifically, the steps of optimizing the uneven grouting amount using the particle swarm algorithm based on the obtained prediction results include: Initialize the candidate set of tunneling parameters and set the uneven grouting volume x=[x 上 ,x 中 ,x 下 ], adjust the range , randomly initialize m particles within the adjustment range, and the position of each particle corresponds to an uneven grouting value , the adjustment speed of each particle is ; Calculate the individual and global optimal solutions. For each particle, calculate its fitness according to the objective function, and set the initial position to its optimal position. At the same time, record the minimum value of the optimal positions of all individuals in the group as the global optimal solution, where the objective function is ; In the above formula, The final settlement of the surface of the shield tail ring, are other grouting parameters except uneven grouting amount, is the geological parameter corresponding to the shield tail ring, are the tunnel geometry parameters; Update the speed and position of each particle according to the following formula: ,

[0101] In the above formula, is the velocity of the particle in the next time step, is the inertia weight, and is the learning factor, and is a random number between [0,1], is the position of particle i at time t, is the optimal position found between particle i and time t; Global optimal position; Update the candidate set of uneven grouting volume, and update the candidate set of uneven grouting volume at time t according to the updated particle speed and position Updated to the candidate set of uneven grouting volume at time t+1 ; Update individual and global optimal solutions, calculate the objective function value of each element in the excavation parameter candidate set according to the updated excavation parameter candidate set, and update the individual optimal solution of each element and the global optimal solution of the element set; The steps of updating the candidate set of uneven grouting amount and updating the individual and global optimal solutions are repeated until the iteration termination condition is reached, and the optimized uneven grouting amount is output.

[0102] The present invention is described in detail above in conjunction with the embodiments of the accompanying drawings. A person skilled in the art can make various variations of the present invention according to the above description. Therefore, certain details in the embodiments should not constitute a limitation of the present invention, and the scope of protection of the present invention shall be defined by the scope of the attached claims.

Claims

1. A method for setting uneven grouting volume of a large-diameter shield considering the settlement development process, characterized in that: The steps include: Collect construction data from historical projects; According to the current construction conditions, a clustering algorithm is used to obtain a matching micro-condition database from the collected construction data of historical projects; Based on the surface settlement data in the micro-condition database, variational mode decomposition is used to extract the surface settlement development information. Extracting uneven grouting parameters from the micro-condition database; Constructing a surface settlement prediction model using the extracted surface settlement development information, uneven grouting parameters and the micro-condition database; During the actual tunneling process of the shield machine, the constructed surface settlement prediction model is used to predict the surface settlement of the current shield tail ring to obtain the prediction result. Based on the prediction result, the particle swarm algorithm is used to optimize the uneven grouting amount, and then the uneven grouting amount of the next ring is generated.

2. The method for setting uneven grouting amount of a large-diameter shield considering the settlement development process as claimed in claim 1, characterized in that: It also includes clustering algorithm analysis of the collected historical engineering construction data: S11, define the domain radius parameter and the minimum number of sample points parameter, and initialize all points to an unvisited state; S12, randomly select an unvisited point P from the collected historical engineering construction data. If the number of points in the domain of P is greater than or equal to the minimum sample point number parameter, P is marked as a core point, a new cluster is created, all points in the domain of P are expanded, and points that meet the density requirements are added to the new cluster; if the number of points in the domain of P is less than the minimum sample point number parameter, P is marked as a noise point; S13, check the domains of all points in the current cluster. If a point is a core point and has not been visited, add the points in its domain to the current cluster; if a point is a boundary point, assign it to the current cluster without expanding it. Repeat this step until the current cluster cannot be expanded any further. S14, select the next point from the remaining unvisited points, and repeat steps S12 and S13 until all points are processed; S15, outputting sample points of the same cluster as the same micro-operating condition database, wherein points not assigned to any cluster are marked as noise points.

3. The method for setting uneven grouting amount of a large-diameter shield considering the settlement development process as claimed in claim 1, characterized in that: The steps to extract the surface settlement development information using variational mode decomposition include: Initialize the candidate range of the modal number K to [1,2,3,…,N]; Perform variational mode decomposition to decompose the original surface settlement data into K modes IMF1, IMF2, …, IMF k ; The superposition of modal components gives ; Calculate the decomposed signal With the original signal The error between , where Euclidean norm of the signal; The optimal mode number K is determined using the elbow method; The original surface settlement data are extracted by variational mode decomposition to obtain 1 trend component and K-1 fluctuation components.

4. The method for setting uneven grouting amount of a large-diameter shield considering the settlement development process as claimed in claim 1, characterized in that: The steps of constructing a surface settlement prediction model using the extracted surface settlement development information, uneven grouting parameters and the micro-condition database include: The surface settlement development information extracted by using the modified mode decomposition is input into the GRU model to extract the time series data of surface settlement development; Extracting uneven grouting parameters, tunnel geometry parameters and geological parameters from the micro-condition database using a BP neural network model; The extracted surface settlement development time series data, uneven grouting parameters, tunnel geometric parameters and geological parameters are used as the input of the deep neural network model, and the final settlement of the shield tail ring is used as the output of the deep neural network model. The deep neural network model is trained and then a surface settlement prediction model is constructed.

5. The method for setting uneven grouting amount of a large-diameter shield considering the settlement development process as claimed in claim 1, characterized in that: The steps of optimizing the uneven grouting amount using the particle swarm algorithm based on the obtained prediction results include: Initialize the candidate set of tunneling parameters and set the uneven grouting volume x=[x 上 ,x 中 ,x 下 ], adjust the range , randomly initialize m particles within the adjustment range, and the position of each particle corresponds to an uneven grouting value , the adjustment speed of each particle is ; Calculate the individual and global optimal solutions. For each particle, calculate its fitness according to the objective function, and set the initial position to its optimal position. At the same time, record the minimum value of the optimal positions of all individuals in the group as the global optimal solution, where the objective function is ; In the above formula, The final settlement of the surface of the shield tail ring, are other grouting parameters except uneven grouting amount, is the geological parameter corresponding to the shield tail ring, are the tunnel geometry parameters; Update the speed and position of each particle according to the following formula: , ; In the above formula, is the velocity of the particle in the next time step, is the inertia weight, and is the learning factor, and is a random number between [0,1], is the position of particle i at time t, is the optimal position found between particle i and time t; is the global optimal position; Update the candidate set of uneven grouting volume, and update the candidate set of uneven grouting volume at time t according to the updated particle speed and position Updated to the candidate set of uneven grouting volume at time t+1 ; Update individual and global optimal solutions, calculate the objective function value of each element in the excavation parameter candidate set according to the updated excavation parameter candidate set, and update the individual optimal solution of each element and the global optimal solution of the element set; The steps of updating the candidate set of uneven grouting amount and updating the individual and global optimal solutions are repeated until the iteration termination condition is reached, and the optimized uneven grouting amount is output.

6. A system for setting uneven grouting volume of a large-diameter shield considering the settlement development process, characterized in that: include: A general database for storing the collected construction data of historical projects; A cluster analysis unit connected to the master database is used to obtain a matching micro-condition database from the collected construction data of historical projects using a clustering algorithm according to the current construction conditions; A settlement extraction unit, connected to the cluster analysis unit, is used to extract surface settlement development information using variational mode decomposition for the surface settlement data in the obtained micro-condition database; A parameter extraction unit, connected to the cluster analysis unit, for extracting uneven grouting parameters from the micro-condition database; A model building unit, connected to the cluster analysis unit, the settlement extraction unit and the parameter extraction unit, for building a surface settlement prediction model using the extracted surface settlement development information, the non-uniform grouting parameters and the micro-condition database; The processing unit is connected to the model building unit and is used to predict the surface settlement of the current shield tail ring using the constructed surface settlement prediction model during the actual shield tunneling process to obtain a prediction result, and optimize the uneven grouting amount using a particle swarm algorithm based on the obtained prediction result, thereby generating an uneven grouting amount for the next ring.

7. The large-diameter shield uneven grouting amount setting system considering the settlement development process as claimed in claim 6 is characterized in that: The cluster analysis unit is also used to perform clustering algorithm analysis on the collected construction data of historical projects to obtain multiple micro-condition databases.

8. The large-diameter shield uneven grouting amount setting system considering the settlement development process as claimed in claim 6 is characterized in that: The settlement extraction unit uses variational mode decomposition to extract surface settlement development information including: Initialize the candidate range of the modal number K to [1,2,3,…,N]; Perform variational mode decomposition to decompose the original surface settlement data into K modes IMF1, IMF2, …, IMF k ; The superposition of modal components gives ; Calculate the decomposed signal With the original signal The error between ,in is the Euclidean norm of the signal; The optimal mode number K is determined using the elbow method; The original surface settlement data are extracted by variational mode decomposition to obtain 1 trend component and K-1 fluctuation components.

9. The large-diameter shield uneven grouting amount setting system considering the settlement development process as claimed in claim 6, characterized in that: The model building unit includes a GRU model, a BP neural network model, a deep neural network model and a training model module; The GRU model is connected to the settlement extraction unit, and is used to receive the surface settlement development information extracted by the settlement extraction unit, and extract the surface settlement development time series data; The BP neural network model is used to extract uneven grouting parameters, tunnel geometry parameters and geological parameters from the micro-condition database; The training model module is connected to the GRU model, the BP neural network model and the deep neural network model, and is used to use the extracted surface settlement development time series data, uneven grouting parameters, tunnel geometric parameters and geological parameters as the input of the deep neural network model, and the final settlement of the shield tail ring as the output of the deep neural network model to train the deep neural network model, and then construct a surface settlement prediction model.

10. The large-diameter shield uneven grouting amount setting system considering the settlement development process as claimed in claim 6, characterized in that: The processing unit is also used for: Initialize the candidate set of tunneling parameters and set the uneven grouting volume x=[x 上 ,x 中 ,x 下 ], adjust the range , randomly initialize m particles within the adjustment range, and the position of each particle corresponds to an uneven grouting value , the adjustment speed of each particle is ; Calculate the individual and global optimal solutions. For each particle, calculate its fitness according to the objective function, and set the initial position to its optimal position. At the same time, record the minimum value of the optimal positions of all individuals in the group as the global optimal solution, where the objective function is ; In the above formula, The final settlement of the surface of the shield tail ring, are other grouting parameters except uneven grouting amount, is the geological parameter corresponding to the shield tail ring, are the tunnel geometry parameters; Update the speed and position of each particle according to the following formula: , ; In the above formula, is the velocity of the particle in the next time step, is the inertia weight, and is the learning factor, and is a random number between [0,1], is the position of particle i at time t, is the optimal position found between particle i and time t; is the global optimal position; Update the candidate set of uneven grouting volume, and update the candidate set of uneven grouting volume at time t according to the updated particle speed and position Updated to the candidate set of uneven grouting volume at time t+1 ; Update individual and global optimal solutions, calculate the objective function value of each element in the excavation parameter candidate set according to the updated excavation parameter candidate set, and update the individual optimal solution of each element and the global optimal solution of the element set; The steps of updating the candidate set of uneven grouting amount and updating the individual and global optimal solutions are repeated until the iteration termination condition is reached, and the optimized uneven grouting amount is output.

Images