A method and system for early warning and recommendation of shield tunneling parameters

Abstract

The application discloses a kind of shield tunneling parameter early warning and recommendation method and system, belong to shield construction technical field.It includes: by the physical and mechanical parameters of soil in geological exploration borehole data extraction, total section global expansion is realized using Kriging interpolation method, and extraction is constructed tunneling state data;Form group multidimensional geological vector, after preliminary clustering with multi-base classifier, to construct similarity matrix with spectral clustering integration;For the early warning interval of the key operating parameter of shield of each cluster corresponding stratum condition;Build automatic optimization mechanism, and filter out the similarity measurement method of adaptation;Select the cluster of highest similarity;With the median of the inner parameter of the cluster of highest similarity as the recommended value of current key parameter, synchronously collect the early warning interval of the cluster of highest similarity as monitoring threshold range.The application can provide reliable parameter recommendation and early warning for shield construction, effectively reduce construction failure risk, significantly improve construction efficiency and quality.

Description

Technical Field

[0001] This invention relates to the field of tunnel boring machine (TBM) construction technology, and in particular to a method and system for early warning and recommendation of TBM tunneling parameters. Background Technology

[0002] During tunnel boring machine (TBM) construction, the complexity and uncertainty of geological conditions present numerous challenges. Different geological conditions have different requirements for TBM construction parameters. If the construction parameters are not selected properly, various construction problems can easily occur, such as excessive wear of the cutterhead, uncontrolled ground settlement, and instability of the excavation face. These problems can not only increase construction costs and affect construction efficiency, but also threaten the safety of construction personnel.

[0003] Currently, the determination of tunnel boring machine (TBM) construction parameters mainly relies on engineering experience and simple data analysis, lacking a systematic and scientific approach. While some studies attempt to utilize geological data to guide parameter selection, shortcomings exist in geological data processing and analysis. The intrinsic relationship between geological data and construction parameters cannot be fully explored. Traditional methods for processing geological data are often limited to local areas, failing to comprehensively consider geological variations across the entire construction area. In cluster analysis, single clustering algorithms struggle to accurately reflect the complex structure of geological data, and their measurement of the similarity between the geological conditions of new tunnels and existing projects is inaccurate, resulting in an inability to provide reasonable construction parameter recommendations for new tunnels. Therefore, a new method is urgently needed to address these issues and meet the growing demands of TBM construction. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies, this invention provides a comprehensive and accurate method for early warning and recommendation of tunnel boring machine (TBM) parameters. By deeply analyzing geological data and accurately extracting key information, and employing clustering and similarity measurement methods, it provides reliable parameter early warnings and accurate recommendations for TBM construction, thereby effectively reducing construction risks, improving construction efficiency and quality, and ensuring the smooth progress of TBM construction.

[0005] This invention adopts the following technical solution: a method for early warning and recommendation of shield tunneling parameters, comprising the following steps:

[0006] Soil parameters at the excavation face are extracted and weighted by the proportion of different soil layers in each ring to obtain the stratum parameter values ​​at the excavation face. The values ​​are then extended to the entire longitudinal profile using interpolation. At the same time, valid tunneling status data are filtered out from the tunneling data using a discriminant function.

[0007] Align the excavation surface stratum parameter values ​​after the full-area expansion with the ring numbers according to the construction mileage to form... Grouped multidimensional geological vectors; after initial clustering by a multi-base classifier, spectral clustering was used for ensemble formation, and Gaussian kernel function was used to construct the final structure. The similarity matrix of the dimensions ultimately yields geological vector clustering results that reflect different stratigraphic conditions;

[0008] Based on the geological vector clustering results, the early warning range of key shield tunneling operation parameters is determined according to the geological conditions corresponding to each cluster; geological vector data of newly built tunnels are obtained, and several simulated geological vectors are randomly generated.

[0009] Build an automatic optimization mechanism to select suitable similarity measurement methods;

[0010] The similarity between the geological vector data of the newly built tunnel and the existing clusters is calculated using an appropriate similarity metric. The cluster with the highest similarity is selected. The median of the intrinsic parameters of the cluster with the highest similarity is used as the recommended value of the current key parameters. The warning interval of the cluster with the highest similarity is collected simultaneously as the monitoring threshold range.

[0011] In a further embodiment, the calculation process for the formation parameter values ​​at the excavation face is as follows:

[0012] For scenarios involving complex soil layers with multiple soil types at the excavation face, each ring of the excavation face is used as the calculation unit to obtain the soil layer type within each ring of the excavation face. And calculate each soil layer type Percentage at the current excavation face , The geological parameters at the excavation face are calculated using the following formula. :

[0013] In the formula, Soil layer type The corresponding physical and mechanical parameters.

[0014] In a further embodiment, the process of implementing the longitudinal profile global expansion of the soil parameters at the excavation face is as follows:

[0015] Determine any point within the excavation section where soil parameters need to be supplemented as the point to be estimated. Collect points to be estimated Surrounding known sample points , , The soil layer type in each ring excavation face;

[0016] Establish a system of equations for the weighting coefficients and solve them to obtain the known sample points. Corresponding weight coefficients ,

[0017] Based on the calculated weight coefficients The following formula is used to achieve full longitudinal profile expansion:

[0018] , For sample points The measured value, Point to be estimated The estimated value.

[0019] In a further embodiment, the process for filtering the valid tunneling status data is as follows:

[0020] A discriminant function is established to distinguish between tunneling and non-tunneling states. for:

[0021] ,in, For the first Total thrust at each time step For the first The tunneling speed at each time step For the first The cutter head torque at each time step For the first The rotational speed of the cutter head at each time step;

[0022] like Then it means the first If the state at a time step is non-tunneling, the corresponding tunneling parameter data will be discarded.

[0023] like Then it means the first The state at each time step is the tunneling state, and the corresponding tunneling parameter data is defined as valid tunneling state data.

[0024] In a further embodiment, the similarity matrix is ​​constructed as follows:

[0025] use and They represent the first Group of multidimensional geological vectors and the first The multidimensional geological vectors are calculated using the following formula. and multidimensional geological vectors similarity coefficient between :

[0026] ,in, For kernel parameters, It is a multidimensional geological vector and multidimensional geological vectors The Euclidean distance between them;

[0027] The calculated similarity coefficient As the first in the similarity matrix Line number Column elements, iterate through all and The combinations are calculated one by one and the coefficients at the corresponding positions are filled in, ultimately forming a dimension of The similarity matrix.

[0028] In a further embodiment, the method for determining the early warning range of the key operating parameters of the tunnel boring machine is as follows:

[0029] Obtain the shield tunneling key parameter data for all time steps of a specified tunneling segment under certain geological conditions, remove outliers from the shield tunneling key parameter data, and obtain shield tunneling optimization parameter data.

[0030] Based on the shield tunneling optimization parameter data, a box plot was drawn, and the upper quartiles were obtained. Lower quartiles median and interquartile range And calculate the upper value. and lower value ;

[0031] The relative negative offset is calculated using the following formulas. and relative positive offset :

[0032] ;

[0033] Relative negative offset based on key shield tunneling parameter data for all rings and relative positive offset Draw an offset box plot;

[0034] The upper and lower quartiles of the offset box plot are used as the upper and lower limits of the warning interval.

[0035] In a further embodiment, the similarity measurement method includes at least: cluster center similarity calculation, cluster member similarity calculation, and kernel function similarity calculation; correspondingly, the construction process of the automatic optimization mechanism is as follows:

[0036] Stability and discriminability were used as evaluation indicators, and a stability quantitative evaluation model and a discriminability quantitative evaluation model were created respectively. The stability was used to measure the degree of fluctuation of the similarity measurement method under different datasets, and the discriminability index was used to evaluate the ability of the similarity measurement method to identify the differences in the characteristics of different geological vectors.

[0037] The cluster center similarity calculation, cluster member similarity calculation, and kernel function similarity calculation were evaluated using a stability quantification evaluation model and a discriminant quantification evaluation model, respectively, and a comprehensive score was obtained. :

[0038] ,in, and All are weighting coefficients. For stability quantification, To distinguish quantified values;

[0039] Select the overall score The similarity measurement method with the highest score is the appropriate similarity measurement method.

[0040] In a further embodiment, the calculation formula of the stability quantification evaluation model is as follows:

[0041] ;

[0042] In the formula, This represents the number of similarity calculations. It is the first The similarity is calculated once. The number of calculations is The average similarity.

[0043] In a further embodiment, the calculation formula for the distinguishability metric evaluation model is as follows:

[0044] ;

[0045] In the formula, This is the set of similarity values ​​between randomly generated geological vectors and different clusters.

[0046] A shield tunneling parameter early warning and recommendation system, used to implement the shield tunneling parameter early warning and recommendation method described above, includes:

[0047] The first module is set to extract soil parameters from the excavation face, and obtain the stratum parameter values ​​of the excavation face by weighted averaging according to the proportion of different soil layers in each ring; it is then extended to the entire longitudinal profile by interpolation, and at the same time, effective tunneling status data is filtered out from the tunneling data using a discriminant function;

[0048] The second module is configured to align the excavation surface stratum parameter values ​​after global expansion with the ring number according to the construction mileage, forming... Grouped multidimensional geological vectors; after initial clustering by a multi-base classifier, spectral clustering was used for ensemble formation, and Gaussian kernel function was used to construct the final structure. The similarity matrix of the dimensions ultimately yields geological vector clustering results that reflect different stratigraphic conditions;

[0049] The third module is set to determine the early warning range of key shield tunneling operation parameters based on the geological vector clustering results and the geological conditions corresponding to each cluster; acquire geological vector data of the newly built tunnel and randomly generate several simulated geological vectors.

[0050] The fourth module is set to use an adapted similarity measurement method to calculate the similarity between the geological vector data of the newly built tunnel and the existing clusters, and select the cluster with the highest similarity. The median of the internal parameters of the cluster with the highest similarity is used as the recommended value of the current key parameters, and the warning interval of the cluster with the highest similarity is collected simultaneously as the monitoring threshold range.

[0051] The beneficial effects of this invention are as follows: In the parameter extraction and global expansion stages, this invention achieves accurate expansion of the entire cross-section through interpolation, providing a comprehensive and reliable data foundation for subsequent geological analysis.

[0052] The geological data clustering calculation process employs an ensemble clustering method, combining multiple base classifiers and spectral clustering integration techniques. This fully leverages the advantages of different clustering algorithms, enabling more accurate capture of the intrinsic structure and characteristics of geological data. This improves the accuracy and stability of clustering, providing a more reasonable geological classification basis for the analysis of shield tunneling construction parameters.

[0053] The method for calculating early warning intervals utilizes isolated forests to remove outliers and calculates early warning intervals for key parameters using box plots. This method can promptly identify abnormal parameters during tunnel boring machine (TBM) construction, providing accurate early warning information to construction personnel. This helps to take preventative measures to avoid construction failures and ensure construction safety and quality.

[0054] The three similarity measurement methods proposed in the geological vector matching and parameter recommendation section for new tunnels, combined with an automatic optimization mechanism based on stability and discriminative power, can more accurately match the geological conditions of new tunnels and recommend suitable construction parameters based on similar geological clusters. Meanwhile, the revised kernel parameter calculation method and the set similarity matching threshold further improve the accuracy and reliability of parameter recommendations, effectively increasing construction efficiency and reducing construction costs. Attached Figure Description

[0055] Figure 1 This is the overall flowchart of Example 1.

[0056] Figure 2 This is a schematic diagram of the stratigraphic clustering results. Detailed Implementation

[0057] The technical solution of the present invention will be further described in detail below through embodiments and with reference to the accompanying drawings.

[0058] To make the technical problems of the present invention clearer, embodiments of the present invention will be provided below, along with a more comprehensive description. However, the specific embodiments given herein are only for explaining the present invention and are not intended to limit the scope or application of the invention.

[0059] Example 1

[0060] like Figure 1 As shown in the figure, this embodiment discloses a method for early warning and recommendation of shield tunneling parameters, including the following steps:

[0061] Soil parameters at the excavation face are extracted and weighted by the proportion of different soil layers in each ring to obtain the stratum parameter values ​​at the excavation face. The values ​​are then extended to the entire longitudinal profile using interpolation. At the same time, valid tunneling status data are filtered out from the tunneling data using a discriminant function.

[0062] Align the excavation surface stratum parameter values ​​after the full-area expansion with the ring numbers according to the construction mileage to form... Grouped multidimensional geological vectors; after initial clustering by a multi-base classifier, spectral clustering was used for ensemble formation, and Gaussian kernel function was used to construct the final structure. The similarity matrix of the dimensions ultimately yields geological vector clustering results that reflect different stratigraphic conditions;

[0063] Based on the geological vector clustering results, the early warning range of key shield tunneling operation parameters is determined according to the geological conditions corresponding to each cluster; geological vector data of newly built tunnels are obtained, and several simulated geological vectors are randomly generated.

[0064] Build an automatic optimization mechanism to select suitable similarity measurement methods;

[0065] The similarity between the geological vector data of the newly built tunnel and the existing clusters is calculated using an appropriate similarity metric. The cluster with the highest similarity is selected. The median of the intrinsic parameters of the cluster with the highest similarity is used as the recommended value of the current key parameters. The warning interval of the cluster with the highest similarity is collected simultaneously as the monitoring threshold range.

[0066] The soil parameters at the excavation face described in this embodiment can be: in a certain shield tunneling project, based on the original data of 46 boreholes obtained from the geological exploration report, the physical and mechanical parameters of the soil, such as the thickness of the overburden, the thickness of the typical rock mass at the excavation face, the natural density of the soil, cohesion, internal friction angle, deformation modulus, Poisson's ratio, lateral pressure coefficient, permeability coefficient, liquid limit, plastic limit, and the proportion of the typical rock mass at the excavation face, are extracted.

[0067] Correspondingly, the calculation process for the excavation face stratum parameter values ​​is as follows:

[0068] For scenarios involving complex soil layers with multiple soil types at the excavation face, each ring of the excavation face is used as the calculation unit to obtain the soil layer type within each ring of the excavation face. And calculate each soil layer type Percentage at the current excavation face , The geological parameters at the excavation face are calculated using the following formula. :

[0069] In the formula, Soil layer type The corresponding physical and mechanical parameters.

[0070] This embodiment considers the different soil layer types and their proportions in each ring of the excavation face, and uses a weighted average method to calculate the stratum parameter values ​​of the excavation face, which can more accurately reflect the comprehensive physical and mechanical properties of the composite soil layer.

[0071] Compared to calculation methods that simply take parameters from a single soil layer or do not consider the proportion of different layers, this method avoids parameter deviations caused by ignoring the diversity of soil layers, providing data support that is more in line with actual geological conditions for determining subsequent shield tunneling parameters. For example, in a shield tunneling project, if both sand and clay layers exist at the excavation face, weighted calculations can accurately reflect the characteristics of the mixture, avoiding the selection of inappropriate tunneling parameters due to misjudgment of soil layer properties, and reducing problems such as cutter wear and low construction efficiency.

[0072] Therefore, the process of implementing the longitudinal profile global extension of the soil parameters at the excavation face is as follows:

[0073] Determine any point within the excavation section where soil parameters need to be supplemented as the point to be estimated. Collect points to be estimated Surrounding known sample points , , The soil layer type in each ring excavation face;

[0074] Establish a system of equations for the weighting coefficients and solve them to obtain the known sample points. Corresponding weight coefficients The weighting coefficient equations described in this embodiment are expressed in the following form:

[0075] In the formula, Sample points and sample points The semivariogram values ​​between Sample points and sample points The semivariogram values ​​between For Lagrange multipliers;

[0076] Based on the calculated weight coefficients The following formula is used to achieve full longitudinal profile expansion:

[0077] , For sample points The measured value, Point to be estimated The estimated value.

[0078] Based on the above example, using the existing 46 boreholes as known sample points, for a given point to be estimated, after calculating the weighting coefficients, the formula is used... The estimated values ​​of the geological physical and mechanical parameters at that point are calculated. Following this method, point-by-point estimations are performed throughout the entire construction area to extend the physical and mechanical parameters across the entire cross-section, thus obtaining complete distribution data of the physical and mechanical parameters in the area.

[0079] During tunnel boring machine (TBM) construction, the tunneling data collected by the TBM's PLC system includes invalid data from non-tunneling states such as shutdown, ring splicing, and equipment debugging. This type of non-tunneling state data is not directly related to the normal tunneling operation. If it is directly used for subsequent calculations of critical TBM parameter warning intervals, geological vector clustering, and parameter recommendations, it will lead to the following problems: data interference, biased analysis results, and low computational efficiency.

[0080] Therefore, this embodiment filters out tunneling status data through the following process, the specific procedure of which is as follows:

[0081] A discriminant function is established to distinguish between tunneling and non-tunneling states. for:

[0082] ,in, For the first Total thrust at each time step For the first The tunneling speed at each time step For the first The cutter head torque at each time step For the first The rotational speed of the cutter head at each time step;

[0083] like Then it means the first If the state at any time step is non-tunneling, the corresponding tunneling parameter data will be discarded; if Then it means the first The state at each time step is the tunneling state, and the corresponding tunneling parameter data is defined as valid tunneling state data.

[0084] To facilitate understanding, the obtained geological parameters from the excavation face have been organized, and the data ring numbers have been aligned according to the construction mileage to form... The geological vector is generated by combining the geological parameters corresponding to each ring number into a multidimensional geological vector. For example, the geological vector for ring 100 is:

[0085] .

[0086] It is worth noting that the multi-base classifier in this embodiment includes: K-means++ algorithm, hierarchical clustering, DBSCAN clustering, and K-medoids clustering. For example, the K-means++ algorithm selects initial cluster centers based on the distribution of the data, and after multiple iterations, divides the geological vectors into different clusters. Hierarchical clustering starts with each geological vector as a separate class, gradually merging similar classes to form a tree-like clustering structure. Then, spectral clustering is used for ensemble processing.

[0087] When constructing the similarity matrix, the Gaussian kernel similarity between each pair of vectors is calculated based on the characteristics of the geological vectors. The construction process is as follows:

[0088] use and They represent the first Group of multidimensional geological vectors and the first The multidimensional geological vectors are calculated using the following formula. and multidimensional geological vectors similarity coefficient between :

[0089] ,in, For kernel parameters, It is a multidimensional geological vector and multidimensional geological vectors The Euclidean distance between them;

[0090] The calculated similarity coefficient As the first in the similarity matrix Line number Column elements, iterate through all and The combinations are calculated one by one and the coefficients at the corresponding positions are filled in, ultimately forming a dimension of The similarity matrix was used for spectral clustering, resulting in more accurate geological clustering. Based on the data characteristics of the tunnel geological vectors, the geological vectors of the entire construction area were divided into five different clusters, each representing a specific combination of geological conditions. (See reference...) Figure 2 . Figure 2 The numbers in the table correspond to Table 1, representing the stratigraphic categories.

[0091] Suppose that in a certain tunnel boring machine (TBM) construction project, after extracting and expanding geological parameters across the entire area, and aligning the ring numbers according to the construction mileage, N=1604 sets of multidimensional geological vectors are formed, and the similarity matrix is ​​calculated as follows:

[0092]

[0093] The "safe fluctuation range" of key shield tunneling parameters differs fundamentally across different geological formations. Traditional early warning methods fail to establish a correspondence between the "ground" and the parameter range, applying a single early warning standard uniformly. For example, in soft soil, the tunneling speed needs to be controlled at a low range to avoid ground subsidence, while in hard rock, the tunneling speed can be appropriately increased to ensure efficiency. Using the same early warning range for tunneling speed can lead to missed detection of "overspeeding" risks in soft soil or misjudgment of "normal speed" as abnormal in hard rock. This "one-size-fits-all" early warning method cannot match the construction characteristics of different geological formations, making it difficult to achieve scenario-specific and precise parameter risk control, and is detrimental to balancing construction safety and efficiency.

[0094] To address the aforementioned technical problems, the method for determining the early warning range of key operating parameters of the tunnel boring machine in this embodiment is as follows:

[0095] Obtain the shield tunneling key parameter data for all time steps of a specified tunneling segment under certain geological conditions; remove outliers from the shield tunneling key parameter data to obtain shield tunneling optimization parameter data; plot a box plot based on the shield tunneling optimization parameter data and obtain the upper quartiles. Lower quartiles median and interquartile range And calculate the upper value. and lower value ;in,

[0096] .

[0097] The relative negative offset is calculated using the following formulas. and relative positive offset :

[0098] ;

[0099] Relative negative offset based on key shield tunneling parameter data for all rings and relative positive offset Draw an offset box plot; use the upper and lower quartiles of the offset box plot as the upper and lower limits of the warning interval.

[0100] Based on the shield tunneling optimization parameter data, the recommended values ​​and warning ranges of cutterhead rotation speed under different geological conditions were calculated using the above method, as shown in Table 1.

[0101] Table 1. Recommended cutter head torque results and warning range

[0102]

[0103] The calculation methods for other key tunneling parameters of the shield (such as shield advance speed, penetration depth, thrust, hydraulic cylinder thrust, etc.) are the same as those described above, and will not be repeated here.

[0104] In actual construction, when the key tunneling parameters of the tunnel boring machine exceed this warning range, the construction personnel can promptly detect and take corresponding adjustment measures, such as adjusting the tunnel boring machine's advance speed or grouting volume, to ensure construction safety.

[0105] The outlier removal formula used in this embodiment is as follows:

[0106] In the formula, It is a sample Abnormal judgment values, It is a sample Path length to the root node Expected value It is the average path length of the tree. It represents the number of training samples.

[0107] If the calculated sample of If the value is close to 1, it indicates that the parameters in the corresponding sample are abnormal and need to be removed.

[0108] For example, raw data of a certain type of tunnel boring machine (TBM) during the construction of a subway project was obtained. Based on a discriminant function, the TBM tunneling status was selected, resulting in 64,982 construction parameter data points for each status. For a specific geological condition (the geological stratum represented by a certain cluster), key parameter data for a specific tunneling segment was then analyzed. First, the above method was used to remove outliers. Multiple isolated trees were constructed, and the outlier score for each sample point was calculated. For instance, for a sample value in the average earth pressure data, if its outlier score was close to 1, the sample was identified as an outlier and removed.

[0109] In another embodiment, the similarity measurement method includes at least: cluster center similarity calculation, cluster member similarity calculation, and kernel function similarity calculation; correspondingly, the construction process of the automatic optimization mechanism is as follows:

[0110] Stability and discriminability were used as evaluation indicators, and a stability quantitative evaluation model and a discriminability quantitative evaluation model were created respectively. The stability was used to measure the degree of fluctuation of the similarity measurement method under different datasets, and the discriminability index was used to evaluate the ability of the similarity measurement method to identify the differences in the characteristics of different geological vectors.

[0111] The cluster center similarity calculation, cluster member similarity calculation, and kernel function similarity calculation were evaluated using a stability quantification evaluation model and a discriminant quantification evaluation model, respectively, and a comprehensive score was obtained. :

[0112] ,in, and All are weighting coefficients. For stability quantification, To distinguish quantified values;

[0113] Select the overall score The similarity measurement method with the highest score is the appropriate similarity measurement method.

[0114] Furthermore, the calculation formula for the stability quantification assessment model is as follows:

[0115] ;

[0116] In the formula, This represents the number of similarity calculations. It is the first The similarity is calculated once. The number of calculations is The average similarity. The formula for calculating the discrimination metric evaluation model is as follows:

[0117] ;

[0118] In the formula, This is the set of similarity values ​​between randomly generated geological vectors and different clusters.

[0119] Based on the geological characteristics of the strata of the newly constructed tunnel, simulated geological vectors (for testing) were randomly generated. The comprehensive scores of three similarity calculation methods were calculated. Based on the data characteristics of the existing tunnel's strata geological vectors, the comprehensive scores of the three similarity calculation methods were 4.28, 5.09, and 7.44, respectively. Therefore, the kernel function similarity calculation method is the preferred method.

[0120] For newly constructed tunnel projects, it is necessary to first obtain their geological attributes. Specifically, the geological and stratigraphic attribute vector is... Based on the preferred kernel function similarity calculation method, the maximum similarity value is calculated, and the most similar cluster is determined to be 1. Therefore, the recommended value for the shield cutterhead torque is set at 6456 kN·m, with upper and lower warning limits of 9.77% and 7.99%, respectively. The recommended values ​​and warning limits for other key parameters are determined using the same method.

[0121] For ease of understanding, the formula for calculating cluster center similarity in this embodiment is as follows:

[0122] In the formula, Cluster center similarity For vector dimension Geological vector of the newly built tunnel, It is the vector dimension The cluster center, , For the total dimension.

[0123] In a further embodiment, the formula for calculating cluster member similarity is as follows:

[0124] In the formula, It is the cluster member similarity value. It is the number of cluster members. It is the first The first cluster member Geological parameter values ​​in each dimension.

[0125] In a further embodiment, the formula for kernel function similarity calculation described in this embodiment is as follows:

[0126] .

[0127] The above descriptions are preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the invention should be included within the scope of protection of the invention.

Claims

1. A method for early warning and recommendation of shield tunneling parameters, characterized in that, The method comprises the following steps: extracting the excavation face soil body parameters, performing weighted averaging according to the proportion of different soil layers in each ring to obtain the excavation face stratum parameter value; extending to the longitudinal section domain through an interpolation method, and screening effective tunneling state data from the tunneling data by using a discriminant function; The extended global excavation face stratum parameter values are aligned with the ring numbers according to the construction mileage to form The multi-dimensional geological vectors are preliminarily clustered by the multi-base classifier, and are integrated by spectral clustering to construct a similarity matrix of dimensions through a Gaussian kernel function, and finally a clustering result of geological vectors reflecting different stratum conditions is obtained. The multi-dimensional geological vectors are preliminarily clustered by the multi-base classifier, and are integrated by spectral clustering to construct a similarity matrix of dimensions through a Gaussian kernel function, and finally a clustering result of geological vectors reflecting different stratum conditions is obtained. determining the early warning interval of the shield key operation parameter based on the geological vector clustering result and the stratum condition corresponding to each cluster; obtaining new tunnel geological vector data and randomly generating a plurality of simulation geological vectors; constructing an automatic optimization mechanism to screen an adaptive similarity measurement method, wherein the similarity measurement method at least comprises cluster center similarity calculation, cluster member similarity calculation and kernel function similarity calculation; calculating the similarity between the new tunnel geological vector data and the existing cluster by using the adaptive similarity measurement method, and selecting the cluster with the highest similarity; taking the median of the internal parameters of the cluster with the highest similarity as the current key parameter recommended value, and synchronously collecting the early warning interval of the cluster with the highest similarity as the monitoring threshold range; wherein the implementation process of the longitudinal section domain extension of the excavation face soil body parameters is as follows: Determine any point within the excavation section where soil parameters need to be supplemented as the point to be estimated. Collect points to be estimated Surrounding known sample points , , The soil layer type in each ring excavation face; A weight coefficient equation set is established, and known sample points are solved to obtain corresponding weight coefficients , Based on the obtained weight coefficients The following formula is used to realize the global extension of the longitudinal section: , is the measured value for the sample point , is the estimated value for the point to be estimated; The construction process of the automatic optimization mechanism is as follows: stability and discrimination degree are taken as evaluation indexes to respectively create a stability quantitative evaluation model and a discrimination quantitative evaluation model; The cluster center similarity calculation, the cluster member similarity calculation and the kernel function similarity calculation are evaluated respectively using a stability quantitative evaluation model and a discrimination quantitative evaluation model, and a comprehensive score is obtained The similarity measurement method with the highest comprehensive score is the adaptive similarity measurement method ​ the calculation formula of the stability quantitative evaluation model is as follows: ； In the formula, This represents the number of similarity calculations. It is the first The similarity is calculated once. The number of calculations is The average similarity; the calculation formula of the discrimination quantitative evaluation model is as follows: ； In the formula, is a set of similarity values between the randomly generated geological vector and each different cluster.

2. The method according to claim 1, characterized in that, the calculation process of the excavation face stratum parameter value is as follows: For the composite soil layer scene that multiple soil bodies exist in the excavation face, taking each ring excavation face as a calculation unit, the soil layer type in each ring excavation face is obtained , and the proportion of each soil layer type at the current excavation face is calculated , The following formula is used to calculate the stratum parameter value of the excavation face : wherein is the soil type corresponding physical-mechanical parameters.

3. The method according to claim 1, characterized in that, the screening process of the effective tunneling state data is as follows: A discriminant function is established for dividing the tunneling state and the non-tunneling state, and the discriminant function is: Y = 0.0001X1+ 0.0001X2+ 0.0001X3+ 0.0001X wherein, is the total thrust at the time step, is the penetration speed at the time step, is the cutterhead torque at the time step, is the cutterhead rotation speed at the time step; If , it indicates that the state of the time step is a non-excavation state, and the corresponding excavation parameter data is discarded; If , it indicates that the state of the th time step is the tunneling state, and the corresponding tunneling parameter data is defined as valid tunneling state data.

4. The method according to claim 1, characterized in that, the construction process of the similarity matrix is as follows: Using and represent the first group of multi-dimensional geologic vectors and the second group of multi-dimensional geologic vectors, the similarity coefficient between the multi-dimensional geologic vectors and is calculated using the following formula : wherein, is a kernel parameter, is a multidimensional geological vector and a multidimensional geological vector is the Euclidean distance between them; The similarity coefficient calculated As the element of the similarity matrix of the first row and the first column, all combinations of the first and the second are calculated one by one and the coefficients of the corresponding positions are filled in, and finally a similarity matrix with a dimension of is formed.

5. The method according to claim 1, characterized in that, the method for determining the early warning interval of the shield key operation parameter is as follows: obtaining all time steps of the shield key parameter data of a specified tunneling ring segment under a certain stratum condition, removing the outliers in the shield key parameter data to obtain shield optimization parameter data; drawing a box plot based on the shield optimization parameter data, and obtaining upper quartile , lower quartile , median , and interquartile range , and calculating upper whisker and lower whisker ; The relative negative offset and the relative positive offset are calculated using the following equations, respectively and : ； Relative negative and positive offset amounts based on all ring shield key parameter data and relative positive offset amounts Plot offset amount box plots; taking the upper quartile and the lower quartile of the offset box plot as the upper limit value and the lower limit value of the early warning interval.

6. The method according to claim 1, characterized in that, The stability is used to measure the fluctuation degree of the results of the similarity measurement method under different data sets, and the discrimination degree index is used to evaluate the recognition ability of the similarity measurement method to the feature difference of different geological vectors; The composite score The formula for the calculation of the composite score is as follows: wherein, and are weight coefficients, is a stability quantization value, is a discrimination quantization value.

7. A shield tunneling parameter early warning and recommendation system for implementing the shield tunneling parameter early warning and recommendation method according to any one of claims 1 to 6, characterized in that, comprise: The first module is configured to extract the excavation face soil body parameters, perform weighted averaging according to the proportion of different soil layers in each ring to obtain the excavation face stratum parameter value; extend to the longitudinal section domain through an interpolation method, and screen effective tunneling state data from the tunneling data by using a discriminant function; The second module is configured to align the global expanded excavation face stratum parameter value with the construction mileage to the ring number to form a global expanded excavation face stratum parameter value ring number matrix The multi-dimensional geological vectors are clustered by a multi-base classifier, and then integrated by spectral clustering, and a similarity matrix of dimensions is constructed by a Gaussian kernel function to obtain a clustering result of geological vectors reflecting different stratum conditions The multi-dimensional geological vectors are clustered by a multi-base classifier, and then integrated by spectral clustering, and a similarity matrix of dimensions is constructed by a Gaussian kernel function to obtain a clustering result of geological vectors reflecting different stratum conditions The third module is configured to determine the early warning interval of the shield key operation parameter based on the geological vector clustering result and the stratum condition corresponding to each cluster; obtain new tunnel geological vector data and randomly generate a plurality of simulation geological vectors; The fourth module is configured to calculate the similarity between the new tunnel geological vector data and the existing cluster by using the adaptive similarity measurement method, and select the cluster with the highest similarity; take the median of the internal parameters of the cluster with the highest similarity as the current key parameter recommended value, and synchronously collect the early warning interval of the cluster with the highest similarity as the monitoring threshold range.

Images