Strata type prediction method, device and medium based on small-diameter TBM tunneling

Abstract

The application discloses a stratum type prediction method and device based on small-diameter TBM tunneling and a medium, collects TBM tunneling parameters, judges a TBM tunneling state, removes TBM non-tunneling state data, and obtains first sample data; electromagnetic wave detection is used in the TBM tunneling process to obtain second sample data; time sequence processing is performed on the tunneling parameters to extract time sequence data of the TBM in the tunneling state; detection data corresponding to the time sequence of the TBM in the tunneling state is extracted to determine the correlation of the first sample data and the second sample data; the weight of the first sample data and the second sample data is obtained according to the correlation; a prediction model is constructed, the prediction model is optimized based on the weight, and stratum type prediction is performed based on the optimized prediction model. The TBM tunneling state and electromagnetic wave detection data at the same time sequence are obtained, parameter analysis is performed by combining the current TBM tunneling state and the electromagnetic wave detection data, and the prediction accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of TBM technology, and specifically to a method, equipment, and medium for predicting formation types based on small-diameter TBM tunneling. Background Technology

[0002] Small-section TBM construction effectively improves the problems of low clamping efficiency, poor construction environment, and high safety risks in auxiliary caverns such as drainage tunnels, cable wells, and test tunnels under traditional drill-and-blast methods. However, comprehensive research and innovation have not yet been carried out in the construction of inclined shafts in water conveyance channels, especially in the excavation of underground powerhouses on critical routes. Therefore, promoting the standardization and normalization of pumped power station design and construction, and exploring the rapid construction of underground powerhouses through the comprehensive application of TBM technology are crucial. During TBM construction, the construction environment varies greatly, especially the complex geological types, and current theoretical research and numerical simulations cannot completely and realistically simulate the objective geological conditions, and the unknown nature of the underlying strata types further complicates the issue. In existing technologies, it is usually necessary to predict the geological type in advance to improve the geological adaptability of TBMs to cope with complex strata and thus achieve high-efficiency tunneling. However, most technologies do not combine the current tunneling status of the TBM with parameter analysis when making predictions, ignoring the important impact of TBM performance parameters on tunneling efficiency. This results in subjective prediction methods and low prediction accuracy. During the tunneling process, it is impossible to explore the correlation between TBM tunneling parameters and different lithological strata, and it is impossible to predict the subsequent geological type based on the current tunneling status. This leads to the inability to control the tunneling in a timely manner according to the geological type. When the geological type changes significantly, the TBM tunneling line is very likely to change, which in turn causes tunnel quality problems. Summary of the Invention

[0003] The technical problem this invention aims to solve is that it fails to incorporate parameter analysis based on the current tunneling state of the TBM, neglecting the significant impact of TBM performance parameters on tunneling efficiency. This results in an unobjective prediction method and low prediction accuracy. The invention provides a method, equipment, and medium for predicting geological formation types based on small-diameter TBM tunneling. By determining the TBM's tunneling state and eliminating data from non-tunneling states, it simultaneously acquires electromagnetic wave detection data from the same time series under the same TBM tunneling state. A prediction model is constructed based on the correlation and weights of these two types of data to predict geological formation types. By combining the current TBM tunneling state and electromagnetic wave detection data for parameter analysis, the significant impact of TBM performance parameters on tunneling efficiency is considered, thus improving prediction accuracy.

[0004] This invention is achieved through the following technical solution:

[0005] The first aspect of this invention provides a method for predicting geological types based on small-diameter TBM tunneling, comprising the following specific steps:

[0006] Collect TBM tunneling parameters, determine TBM tunneling status, and remove TBM non-tunneling status data to obtain the first sample data;

[0007] Electromagnetic wave detection was used during the TBM tunneling process to obtain the second sample data;

[0008] Time-series processing is performed on the tunneling parameters to extract the time-series data of the TBM during tunneling.

[0009] Extract the time-series detection data of the TBM during its tunneling process and determine the correlation between the first sample data and the second sample data.

[0010] The weights of the first and second sample data are determined based on their correlation.

[0011] A prediction model is constructed, optimized based on weights, and the stratigraphic type is predicted based on the optimized prediction model.

[0012] This invention determines the TBM tunneling status, eliminates non-tunneling TBM data, and simultaneously acquires electromagnetic wave detection data under the same time sequence of the TBM tunneling status. Based on the correlation and weight of the two types of data, a prediction model is constructed to predict the geological type. By combining the current TBM tunneling status and electromagnetic wave detection data for parameter analysis, the significant impact of TBM performance parameters on tunneling efficiency is considered, thus improving the prediction accuracy.

[0013] Furthermore, the determination of the TBM tunneling status specifically includes:

[0014] The calculation steps include: obtaining the advance speed, total thrust, and cutterhead rotation speed of the TBM; constructing a binary discriminant function to determine the TBM's state; and removing data from non-tunneling states and start-up / shutdown phases.

[0015]

[0016] D=d(v1)d(v2)d(F)

[0017] Where D represents the tunneling state, D=1 indicates the normal working state of the TBM, D=0 indicates the abnormal working state of the TBM, v1 represents the propulsion speed, v2 represents the cutterhead rotation speed, and F represents the total thrust.

[0018] Furthermore, when obtaining the first sample data, outlier handling is also performed on the data after the initial removal process.

[0019] Use the quartiles and whisker lengths of the box plot to define outliers;

[0020] Data points located outside the box's tentacles are identified as outliers and replaced with the average of the data within the nearest threshold.

[0021] Furthermore, obtaining the second sample data specifically includes:

[0022] It transmits electromagnetic wave signals into the strata, receives the reflected signals, and performs noise reduction processing on the reflected signals;

[0023] Waveform analysis is performed on the denoised signal to determine the formation type.

[0024] Furthermore, the step of performing time-series processing on the tunneling parameters to extract the time-series data corresponding to the TBM under tunneling conditions specifically includes:

[0025] Obtain the timestamps of the tunneling parameters and construct a time-series vector based on the timestamps;

[0026] Based on the time series vector, construct the corresponding transformation matrix to obtain the transformation function F(x)=αN+M, where N represents the coefficient matrix, M represents the bias matrix, and α represents the ordered pair vector;

[0027] Temporal features of tunneling parameters are extracted based on the transformation function.

[0028] Furthermore, the calculation of the correlation between the first sample data and the second sample data specifically includes:

[0029] After determining that the TBM is in a normal working state, extract the temporal features and then extract the second sample data at the same timestamp as the first sample data based on the temporal features.

[0030] Sort the first and second sample data according to their timestamps:

[0031] Construct a data acquisition window, divide the sorted data into data segments according to the direction of movement of the acquisition window, and construct multiple correlation coefficient matrices.

[0032] Furthermore, the step of obtaining the weights of the first and second sample data based on their correlation specifically includes:

[0033] Extract multiple correlation coefficients from multiple correlation coefficient matrices;

[0034] Construct a judgment matrix to verify the consistency of multiple correlation coefficients;

[0035] Calculate the difference coefficient for elements that have passed the consistency check;

[0036] The weights of the first and second sample data are determined based on the difference coefficient.

[0037] Furthermore, the optimization of the prediction model based on weights specifically includes:

[0038] Obtain the weight vector of the first sample data and the weight vector of the second sample data;

[0039] Range standardization is used to normalize the weight vectors of the first sample data and the weight vectors of the second sample data respectively.

[0040] The prediction model is trained based on the normalized weights to obtain the optimized prediction model.

[0041] A second aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a method for predicting geological types based on small-diameter TBM tunneling.

[0042] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for predicting geological types based on small-diameter TBM tunneling.

[0043] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0044] By judging the TBM tunneling status and eliminating TBM non-tunneling status data, while obtaining electromagnetic wave detection data of the same time series under the same TBM tunneling status, a prediction model is constructed based on the correlation and weight of the two types of data to predict the stratum type. By combining the current TBM tunneling status and electromagnetic wave detection data for parameter analysis, the important influence of TBM performance parameters on tunneling efficiency is considered, thus improving the prediction accuracy.

[0045] By optimizing the prediction model through weights, we can find consistency or compromise among the weights of different indicators, obtain a more scientific comprehensive optimal weight allocation coefficient, and improve the prediction accuracy of the optimized prediction model. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0047] Figure 1 This is the prediction process in an embodiment of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0049] As one possible implementation method, such as Figure 1 As shown, the first aspect of this embodiment provides a method for predicting geological types based on small-diameter TBM tunneling, including the following specific steps:

[0050] Collect TBM tunneling parameters, determine TBM tunneling status, and remove TBM non-tunneling status data to obtain the first sample data;

[0051] Electromagnetic wave detection was used during the TBM tunneling process to obtain the second sample data;

[0052] Time-series processing is performed on the tunneling parameters to extract the time-series data of the TBM during tunneling.

[0053] Extract the time-series detection data of the TBM during its tunneling process and determine the correlation between the first sample data and the second sample data.

[0054] The weights of the first and second sample data are determined based on their correlation.

[0055] A prediction model is constructed, optimized based on weights, and the stratigraphic type is predicted based on the optimized prediction model.

[0056] This embodiment determines the TBM tunneling status, eliminates non-tunneling TBM data, and simultaneously acquires electromagnetic wave detection data under the same time sequence of TBM tunneling status. Based on the correlation and weight of the two types of data, a prediction model is constructed to predict the geological type. By combining the current TBM tunneling status and electromagnetic wave detection data for parameter analysis, the significant impact of TBM performance parameters on tunneling efficiency is considered, thus improving the prediction accuracy.

[0057] In some possible embodiments, removing TBM non-tunneling state data includes removing data from the stopped state, the state about to start, the state about to stop, and the stable tunneling stage. The start-up stage is the transition stage from a static state to stable tunneling, and the stop stage refers to the transition stage from a stable tunneling state to a stopped state at the end of each tunneling cycle. During TBM tunneling, a large amount of data from non-tunneling states is accumulated due to procedures such as TBM support shoe replacement, ground support, cutterhead replacement, and shutdown inspection and maintenance. Moreover, tunneling parameters have strong randomness, and the fluctuations between adjacent data are large, which will reduce the accuracy of the analysis results. In order to eliminate the interference of non-tunneling state data on tunneling state data, it is necessary to remove TBM non-tunneling state data. Specifically, judging the TBM tunneling state includes: obtaining the TBM tunneling advance speed, total thrust, and cutterhead rotation speed; constructing a binary discriminant function to judge the TBM state; and removing TBM non-tunneling state and start-up / stop stage state data. The calculation steps include:

[0058]

[0059] D=d(v1)d(v2)d(F)

[0060] Where D represents the tunneling state, D=1 indicates the normal working state of the TBM, D=0 indicates the abnormal working state of the TBM, v1 represents the propulsion speed, v2 represents the cutterhead rotation speed, and F represents the total thrust.

[0061] In some possible embodiments, during TBM tunneling, due to the complex and variable geological conditions, the parameters collected by the TBM will vary depending on the environment—such as strong vibrations, humidity, and strong electric fields. This can sometimes result in null or outlier values ​​in the recorded data. These null or outlier values ​​can affect the accuracy of subsequent predictions, so it is necessary to remove them. In this embodiment, outliers are defined using the quartiles and whisker lengths of a box plot. Data points located outside the box whiskers are identified as outliers, and these outliers are replaced with the average of the data within the nearest threshold.

[0062] In some possible embodiments, obtaining the second sample data specifically includes: transmitting an electromagnetic wave signal to the formation, receiving the reflected signal, and denoising the reflected signal; performing waveform analysis on the denoised signal to obtain the formation type.

[0063] In some possible embodiments, this embodiment uses the following scheme to perform time-series processing on the tunneling parameters and extract the time-series data corresponding to the TBM in the tunneling state: obtain the timestamps of the tunneling parameters, construct a time-series vector based on the timestamps; construct the corresponding transformation matrix based on the time-series vector to obtain the transformation function F(x)=αN+M, where N represents the coefficient matrix, M represents the deviation matrix, and α represents the order pair vector; extract the time-series features of the tunneling parameters based on the transformation function.

[0064] In some possible embodiments, calculating the correlation between the first sample data and the second sample data specifically includes: after extracting the time-series features when the TBM is determined to be in a normal working state, extracting the second sample data at the same timestamp as the first sample data based on the time-series features; sorting the first sample data and the second sample data according to the timestamp; constructing a collection window, dividing the sorted data into data segments according to the movement direction of the collection window, and constructing multiple correlation coefficient matrices R = (r ij ) n×n .

[0065] In some possible embodiments, by collecting and processing multiple data sets and assigning weights to these data sets, the problem of subjectivity in the weighting process and the discrepancy between using only a single weighting method is resolved. The weights of the first and second sample data are obtained based on their correlation, specifically including: extracting multiple correlation coefficients r from multiple correlation coefficient matrices. ij Based on the correlation coefficient r ij Construct the judgment matrix P = (p ij ) n×n , where p ij =(u ij ,v ij ),u ij ,v ij ∈(0,1],u ij Indicates r ij The degree of preference for the i-th indicator compared to the j-th indicator, v ij Indicates r ij The degree of preference for the j-th indicator over the i-th indicator; consistency verification of multiple correlation coefficients; verification passes when the verification result is less than a set threshold; the difference coefficient D of the correlation coefficient within the judgment matrix is ​​calculated for the elements that have passed the consistency verification. ij =1-p ij Based on the difference coefficient, determine the weights of the first and second sample data.

[0066] In some possible embodiments, the prediction model is optimized based on weights, specifically including: obtaining the weight vectors of the first sample data and the second sample data; normalizing the weight vectors of the first sample data and the second sample data respectively using range standardization; training the prediction model based on the normalized weights, and continuously correcting the connection weights and thresholds between each neuron based on error backpropagation to obtain the optimized prediction model; in the optimization process, two operations are usually used: forward signal propagation and backward error propagation. During forward signal propagation, the signal enters from the input layer, passes through the activation of neuron nodes in the hidden layer, and enters the output layer. The output value is compared with the expected value, and if an error exists, the error backpropagation operation is performed. In this stage, by changing the weights and thresholds of each layer, the error is averaged among the neurons in each layer to achieve the effect of error gradient descent. Finally, when the error meets the accuracy standard, learning terminates, and the optimized prediction model is output.

[0067] In some possible implementations, the geological formations encountered during TBM tunneling may include tuff, siliceous rock, sandstone, mylonite, or breccia. Because different formations have different characteristics, their rock types and hardness vary, and their responses differ under different conditions. Therefore, after outputting the prediction results, it is necessary to re-evaluate the results to repeatedly revise the prediction model and ensure the accuracy of the predictions.

[0068] The second aspect of this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a method for predicting the stratum type based on small-diameter TBM tunneling.

[0069] The third aspect of this embodiment provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for predicting geological types based on small-diameter TBM tunneling.

[0070] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting stratigraphic types based on small-diameter TBM tunneling, characterized in that, The specific steps include the following: Collect TBM tunneling parameters, determine the TBM tunneling status, and discard data from non-tunneling states to obtain the first sample data; the determination of the TBM tunneling status specifically includes: The calculation steps include: obtaining the advance speed, total thrust, and cutterhead rotation speed of the TBM; constructing a binary discriminant function to determine the TBM's state; and removing data from non-tunneling states and start-up / shutdown phases. ； ； Where D represents the tunneling state, D=1 indicates the normal working state of the TBM, D=0 indicates the abnormal working state of the TBM, v1 represents the propulsion speed, v2 represents the cutterhead rotation speed, and F represents the total thrust. Electromagnetic wave detection was used during the TBM tunneling process to obtain the second sample data; Time-series processing is performed on the tunneling parameters to extract the time-series data of the TBM under tunneling conditions. Specifically, this includes: Obtain the timestamps of the tunneling parameters, construct a time-series vector based on the timestamps, construct the corresponding transformation matrix based on the time-series vector, and obtain the transformation function. Where N represents the coefficient matrix and M represents the deviation matrix. Represents the ordered pair vector; extracts the temporal features of the tunneling parameters based on the transformation function; Extracting the time-series detection data of the TBM during its tunneling state, and determining the correlation between the first sample data and the second sample data, specifically includes: after determining that the TBM is in normal working state, extracting the time-series features, and then extracting the second sample data at the same timestamp as the first sample data based on the time-series features; sorting the first sample data and the second sample data according to the timestamp; constructing a collection window, dividing the sorted data into data segments according to the direction of movement of the collection window, and constructing multiple correlation coefficient matrices; The weights of the first and second sample data are determined based on their correlation, specifically including: extracting multiple correlation coefficients from multiple correlation coefficient matrices; constructing a judgment matrix to verify the consistency of the multiple correlation coefficients; calculating the difference coefficients for the elements that pass the consistency verification; and determining the weights of the first and second sample data based on the difference coefficients. A prediction model is constructed and optimized based on weights. The optimized prediction model is then used to predict stratigraphic types. Specifically, the optimization of the prediction model based on weights includes: obtaining the weight vectors of the first sample data and the second sample data; normalizing the weight vectors of the first sample data and the second sample data using range standardization; training the prediction model based on the normalized weights; and continuously correcting the connection weights and thresholds between each neuron based on error backpropagation to obtain the optimized prediction model.

2. The method for predicting stratigraphic type based on small-diameter TBM tunneling according to claim 1, characterized in that, The process of obtaining the first sample data also includes outlier handling for the data after the initial data removal: Use the quartiles and whisker lengths of the box plot to define outliers; Data points located outside the box's tentacles are identified as outliers and replaced with the average of the data within the nearest threshold.

3. The method for predicting stratigraphic type based on small-diameter TBM tunneling according to claim 1, characterized in that, The specific details of obtaining the second sample data include: It transmits electromagnetic wave signals into the strata, receives the reflected signals, and performs noise reduction processing on the reflected signals; Waveform analysis is performed on the denoised signal to determine the formation type.

4. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the formation type prediction method based on small-diameter TBM tunneling as described in any one of claims 1 to 3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the formation type prediction method based on small-diameter TBM tunneling as described in any one of claims 1 to 3.

Images