Slope anchor cable prestress measuring method based on data clustering

Abstract

The invention discloses a slope anchor cable prestress measuring method based on data clustering. The slope anchor cable prestress measuring method comprises the steps that a multi-dimensional database containing environment characteristics, echo energy and counter-pulling method prestress values is constructed; a mixed distance clustering algorithm is designed, numerical features and classification features are jointly processed, and automatic data classification is achieved by iteratively updating a cluster center; the cluster quality is dynamically evaluated based on the contour coefficient, and only when the sample size is gt; 3 and an average contour coefficient gt; constructing an exclusive regression model in the cluster when the contour coefficient is a threshold value; and finally predicting the anchor cable prestress in real time through echo energy. According to the method, the prediction precision under the complex working condition is remarkably improved, the use frequency of the reverse drawing method is reduced, and safe and efficient prestress nondestructive testing is achieved.

Description

Technical Field

[0001] The invention relates to a method for measuring prestress of a slope anchor cable based on data clustering, and belongs to the technical field of prestress measurement. Background Art

[0002] Accurately measuring the prestress of slope anchor cables is crucial for long-term slope stability and safe operation. While the traditional reverse tension method offers high accuracy, it requires large tensioning equipment and clamping devices, making the process complex, time-consuming, and labor-intensive. It also carries a risk of damage to the anchor cable structure, presenting significant safety risks and implementation difficulties, particularly in high and steep slopes.

[0003] In recent years, the application of nondestructive testing technologies (such as the acoustic echo method) has gradually gained popularity. By measuring the acoustic echo characteristics of anchor cables, prestress can be indirectly measured. This method is easy to implement on-site and safe and reliable. However, nondestructive testing methods rely on empirical formulas or fitting functions to predict prestress, which generally suffer from insufficient prediction accuracy and poor applicability. This is especially true when the anchor cables are diverse and the construction environment is complex. A single fitting function is difficult to meet practical application requirements. Furthermore, constructing a fitting function requires measuring at least three anchor cables using the back-pulling method, which is labor-intensive and may damage the anchor cables.

[0004] Therefore, there is an urgent need for a measurement method that combines measurement accuracy, non-destructiveness, convenience and low workload. This method should be able to automatically adapt to different types of anchor cables and complex on-site environments to achieve accurate, efficient, fast and non-destructive anchor cable prestress measurement. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method for measuring prestress of slope anchor cables based on data clustering, so as to overcome the shortcomings of the prior art.

[0006] The technical solution of the present invention is: a method for measuring prestress of slope anchor cables based on data clustering, comprising the following steps:

[0007] Step 1: Obtain multi-dimensional data features of anchor cable prestress measurement from the basic database. The multi-dimensional data features include site environment characteristics, echo energy, and anchor cable prestress measured by the back-pull method.

[0008] Step 2: Use clustering algorithm to cluster data to obtain different clusters;

[0009] Step 3: Select at least one group of anchor cables at the measurement site, measure the echo signals and collect site characteristic data, extract discrete data at different wave reflection periods, calculate the echo energy corresponding to the anchor cables, and then use the back-pull method to measure the prestress of the anchor cables. The site environmental characteristics, echo energy, and prestress of the anchor cables measured by the back-pull method are used as multi-dimensional data features corresponding to the anchor cable prestress measurement, and the multi-dimensional data features are added to the basic database;

[0010] Step 4: Calculate the Euclidean distance between the multi-dimensional data feature corresponding to the anchor in step 3 and the cluster center in step 2, and select the cluster with the smallest Euclidean distance as the cluster to which the multi-dimensional data feature corresponding to the anchor in step 3 belongs;

[0011] Step 5: Determine the number of samples in the cluster to which the multi-dimensional data feature corresponding to the anchor in step 3 belongs. If the number of samples is greater than 3 and the average silhouette coefficient is greater than the silhouette coefficient threshold, jump to step 6; otherwise, jump to step 3.

[0012] Step 6: Based on the multi-dimensional data features within the cluster determined in step 4, the prestress values ​​and corresponding echo energy values ​​are extracted and a regression fitting model is constructed;

[0013] Step 7: Measure the echo signals of other anchor cables to be inspected and calculate the echo energy corresponding to the anchor cables. Substitute the echo energy value into the fitting model constructed in step 6 for real-time prestress prediction.

[0014] Furthermore, the echo energy is calculated as follows: Where E is the echo energy, n1 is the starting time of calculation, n2 is the end time of calculation, and x[n] is the amplitude.

[0015] Furthermore, the regression fitting model is Where F is the prestress, and a, b, and c are fitting parameters, which are calculated based on the multi-dimensional data features within the cluster.

[0016] Furthermore, the environmental characteristics include anchor cable specifications, slope gradient, soil / rock layer type, anchor cable service time, grouting method, anchor plate thickness, anchor cable length, anchor section length, construction technology, ground temperature, ground humidity and echo frequency.

[0017] Furthermore, the step 2 specifically includes the following steps:

[0018] (1) Classify environmental features into numerical features and categorical features;

[0019] (2) Set the number of clusters to be divided K and the initial balance coefficient γ;

[0020] (3) Select the initial cluster centers, each cluster center includes the numerical feature center and the classification feature center;

[0021] (4) Calculate the mixed distance d(x) between the sample point and the center of each cluster according to the following formula i ,c j ),

[0022]

[0023] Among them, x i represents the feature vector of the i-th sample point, c j represents the center of the j-th cluster, The pth subvector representing the numerical features of the i-th sample, The p-th subvector representing the numerical features of the j-th cluster center, The qth sub-vector representing the classification feature of the i-th sample, represents the qth subvector of the classification feature of the jth cluster center, δ(·) is the mismatch indicator function, which takes a value of 0 or 1; γ is the balance coefficient, which is used to adjust the weight ratio of numerical features to classification features;

[0024] (5) According to the calculation results of step (4), each sample is classified into the cluster with the smallest distance;

[0025] (6) Update the cluster center, take the average value of the samples in the cluster for numerical features, and take the mode of the samples in the cluster for classification features; (7) Repeat steps (4)-(6) until the cluster division result is stable or the change in the objective function is less than the set threshold;

[0026] (8) Output the final cluster division results and the centers of each cluster.

[0027] Furthermore, the numerical features include: slope gradient, anchor cable service time, anchor plate thickness, anchor cable length, anchor section length, ground temperature, ground humidity and echo frequency;

[0028] The classification characteristics include: anchor cable specifications, soil / rock layer type, grouting method and construction technology.

[0029] Furthermore, the calculation method of the average silhouette coefficient in step 5 is:

[0030] Step (1): Let the multi-dimensional data feature corresponding to the anchor in step 3 be sample i, and calculate the average value a(i) of the distance between sample i and all other samples in its cluster;

[0031] Step (2): Calculate the average value b(i) of the distance between sample i and all samples in the nearest non-self cluster;

[0032] Step (3): Calculate the silhouette coefficient Sil(i) of each sample using the following method:

[0033]

[0034] Step (4): Calculate the average value of the silhouette coefficient of all samples in step (1) Furthermore, the silhouette coefficient threshold value ranges from 0.5 to 0.7.

[0035] The beneficial effects of the present invention are: compared with the prior art,

[0036] 1) The present invention stores a basic database of multi-dimensional data characteristics of anchor cables (field environmental characteristics, echo energy, and prestressing values ​​measured by the back-tension method) and uses data clustering technology to perform cluster analysis on the data. This allows anchor cable data under similar working conditions to be automatically classified, thereby enabling the prestressing prediction model to automatically adjust according to the actual field environment and anchor cable status, significantly improving prediction accuracy, adaptability, and ease of on-site implementation. This effectively overcomes the shortcomings of traditional methods such as complexity, time consumption, and insufficient precision, and is particularly suitable for complex and changing field environments.

[0037] 2) The data in the basic database of the present invention can realize experience accumulation. As the data in the database increases, more sample points are used to fit the model, and thus the prediction of prestress becomes more accurate;

[0038] 3) The present invention can make full use of historical experience. When historical experience accumulates to a certain level, it can be achieved that only a small amount of samples are needed (the back-pull method is used to measure the prestress of the anchor cable). While ensuring the accuracy of prestress prediction, it also reduces the workload and reduces damage to the anchor cable (the back-pull method is destructive to the anchor cable).

[0039] 4) This invention meticulously defines the composition of environmental characteristics, including comprehensive influencing factors such as anchor cable specifications, slope gradient, soil / rock layer type, anchor cable service life, and grouting method. This makes data clustering more precise, and the clustering results more accurately reflect the actual working conditions, further enhancing the model's applicability and prediction accuracy.

[0040] 5) The present invention measures the clustering quality index by the average silhouette coefficient, ensuring that the next step of fitting modeling is entered only when the sample quality of the cluster is high enough, avoiding the error accumulation when the data clustering effect is poor, and further improving the accuracy of the prestressed stress prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is the main flow chart of the present invention;

[0042] Figure 2 It is a flow chart of the clustering algorithm of the present invention;

[0043] Figure 3 This is a flow chart of the average silhouette coefficient calculation of the present invention. DETAILED DESCRIPTION

[0044] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0045] Implementation Example 1:

[0046] This embodiment provides a method for determining the prestress of slope anchor cables based on data clustering. Figure 1 , including the following steps:

[0047] Step 1: Obtain the multi-dimensional data features of anchor cable prestress measurement from the basic database. The multi-dimensional data features include the site environment characteristics, echo energy and the back-pull method to determine the anchor cable prestress. The echo energy is calculated as follows: Where E is the echo energy, n1 is the starting time of the calculation, n2 is the ending time of the calculation, x[n] is the amplitude, and environmental characteristics include anchor cable specifications, slope gradient, soil / rock layer type, anchor cable service life, grouting method, anchor plate thickness, anchor cable length, anchor section length, construction technology, ground temperature, ground humidity, and echo frequency. The numerical characteristics of environmental characteristics include slope gradient, anchor cable service life, anchor plate thickness, anchor cable length, anchor section length, ground temperature, ground humidity, and echo frequency; the categorical characteristics of environmental characteristics include anchor cable specifications, soil / rock layer type, grouting method, and construction technology.

[0048] Step 2: Reference Figure 2 , using clustering algorithms to cluster data to obtain different clusters, specifically including the following steps:

[0049] (1) Classify environmental features into numerical features and categorical features;

[0050] (2) Set the number of clusters to be divided K and the initial balance coefficient γ;

[0051] (3) Select the initial cluster centers, each cluster center includes the numerical feature center and the classification feature center;

[0052] (4) Calculate the mixed distance d(x) between the sample point and the center of each cluster according to the following formula i ,c j ),

[0053]

[0054] Among them, x i represents the feature vector of the i-th sample point, c j represents the center of the j-th cluster, The pth subvector representing the numerical features of the i-th sample, The p-th subvector representing the numerical features of the j-th cluster center, The qth sub-vector representing the classification feature of the i-th sample, represents the qth subvector of the classification feature of the jth cluster center, δ(·) is the mismatch indicator function, which takes a value of 0 or 1; γ is the balance coefficient, which is used to adjust the weight ratio of numerical features to classification features;

[0055] (5) According to the calculation results of step (4), each sample is classified into the cluster with the smallest distance;

[0056] (6) Update the cluster center, take the average value of the samples in the cluster for numerical features, and take the mode of the samples in the cluster for classification features; (7) Repeat steps (4)-(6) until the cluster division result is stable or the change in the objective function is less than the set threshold. The specific form of the objective function is:

[0057]

[0058] Among them, i is the sample number, j is the cluster number, and J represents the objective function. represents the numerical feature vector of the i-th sample, Represents the numerical feature vector of the jth cluster center; δ(z i =j) is an indicator function, which is 1 if sample i belongs to cluster j, and 0 otherwise; Represents the classification feature subvector of the i-th sample; represents the classification feature subvector of the jth cluster center; γ represents the balance coefficient, which is used to adjust the weight ratio of numerical features to classification features;

[0059] (8) Output the final cluster division results and the centers of each cluster.

[0060] Step 3: Select at least one group of anchor cables at the measurement site, measure the echo signals and collect site characteristic data, extract discrete data at different wave reflection periods, calculate the echo energy corresponding to the anchor cables, and then use the back-pull method to measure the prestress of the anchor cables. The site environmental characteristics, echo energy, and prestress of the anchor cables measured by the back-pull method are used as multi-dimensional data features corresponding to the anchor cable prestress measurement, and the multi-dimensional data features are added to the basic database;

[0061] Step 4: Calculate the Euclidean distance between the multi-dimensional data feature corresponding to the anchor in step 3 and the cluster center in step 2, and select the cluster with the smallest Euclidean distance as the cluster to which the multi-dimensional data feature corresponding to the anchor in step 3 belongs;

[0062] Step 5: Determine the number of samples in the cluster to which the multi-dimensional data feature corresponding to the anchor in step 3 belongs. If the number of samples is greater than 3 and the average silhouette coefficient is greater than the silhouette coefficient threshold, jump to step 6; otherwise, jump to step 3. The silhouette coefficient threshold range is 0.5 to 0.7.

[0063] refer to Figure 3 , the calculation method of the average silhouette coefficient is:

[0064] Step (1): Let the multi-dimensional data feature corresponding to the anchor in step 3 be sample i, and calculate the average value a(i) of the distance between sample i and all other samples in its cluster;

[0065] Step (2): Calculate the average value b(i) of the distance between sample i and all samples in the nearest non-self cluster;

[0066] Step (3): Calculate the silhouette coefficient Sil(i) of each sample using the following method:

[0067]

[0068] Step (4): Calculate the average value of the silhouette coefficient of all samples in step (1)

[0069] Step 6: According to the multi-dimensional data features within the cluster determined in step 4, the prestress value and the corresponding echo energy value are extracted to construct a regression fitting model. The regression fitting model is Where F is the prestress, a, b and c are fitting parameters, which are calculated by multi-dimensional data features within the cluster;

[0070] Step 7: Measure the echo signals of other anchor cables to be inspected and calculate the echo energy corresponding to the anchor cables. Substitute the echo energy value into the fitting model constructed in step 6 for real-time prestress prediction.

[0071] The above is a further detailed description of the present invention in conjunction with specific preferred embodiments, and the specific implementation of the present invention should not be considered to be limited to these descriptions. For those skilled in the art of the present invention, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be considered to fall within the scope of protection of the present invention.

Claims

1. A method for determining prestress of slope anchor cables based on data clustering, characterized in that: The following steps are involved: Step 1: Obtain multi-dimensional data features of anchor cable prestress measurement from the basic database. The multi-dimensional data features include site environment characteristics, echo energy, and anchor cable prestress measured by the back-pull method. Step 2: Use clustering algorithm to cluster data to obtain different clusters; Step 3: Select at least one group of anchor cables at the measurement site, measure the echo signals and collect site characteristic data, extract discrete data at different wave reflection periods, calculate the echo energy corresponding to the anchor cables, and then use the back-pull method to measure the prestress of the anchor cables. The site environmental characteristics, echo energy, and prestress of the anchor cables measured by the back-pull method are used as multi-dimensional data features corresponding to the anchor cable prestress measurement, and the multi-dimensional data features are added to the basic database; Step 4: Calculate the Euclidean distance between the multi-dimensional data feature corresponding to the anchor in step 3 and the cluster center in step 2, and select the cluster with the smallest Euclidean distance as the cluster to which the multi-dimensional data feature corresponding to the anchor in step 3 belongs; Step 5: Determine the number of samples in the cluster to which the multi-dimensional data feature corresponding to the anchor in step 3 belongs. If the number of samples is greater than 3 and the average silhouette coefficient is greater than the silhouette coefficient threshold, jump to step 6; otherwise, jump to step 3. Step 6: Based on the multi-dimensional data features within the cluster determined in step 4, the prestress values ​​and corresponding echo energy values ​​are extracted and a regression fitting model is constructed; Step 7: Measure the echo signals of other anchor cables to be inspected and calculate the echo energy corresponding to the anchor cables. Substitute the echo energy value into the fitting model constructed in step 6 for real-time prestress prediction.

2. The method for determining prestress of slope anchor cables based on data clustering according to claim 1, characterized in that: The calculation method of the echo energy is: Where E is the echo energy, n1 is the starting time of calculation, n2 is the end time of calculation, and x[n] is the amplitude.

3. The method for determining prestress of slope anchor cables based on data clustering according to claim 2, characterized in that: The regression fitting model is Where F is the prestress, and a, b, and c are fitting parameters, which are calculated through the multi-dimensional data features within the cluster.

4. The method for determining prestress of slope anchor cables based on data clustering according to claim 1, characterized in that: The environmental characteristics include anchor cable specifications, slope gradient, soil / rock layer type, anchor cable service time, grouting method, anchor plate thickness, anchor cable length, anchor section length, construction technology, ground temperature, ground humidity and echo frequency.

5. The method for determining prestress of slope anchor cables based on data clustering according to claim 4, characterized in that: The step 2 specifically includes the following steps: (1) Classify environmental features into numerical features and categorical features; (2) Set the number of clusters to be divided K and the initial balance coefficient γ; (3) Select the initial cluster centers, each cluster center includes the numerical feature center and the classification feature center; (4) Calculate the mixed distance d(x) between the sample point and the center of each cluster according to the following formula i ,c j ), Among them, x i represents the feature vector of the i-th sample point, c j represents the center of the j-th cluster, The pth subvector representing the numerical features of the i-th sample, The p-th subvector representing the numerical features of the j-th cluster center, The qth sub-vector representing the classification feature of the i-th sample, represents the qth subvector of the classification feature of the jth cluster center, δ(·) is the mismatch indicator function, which takes a value of 0 or 1; γ is the balance coefficient, which is used to adjust the weight ratio of numerical features to classification features; (5) According to the calculation results of step (4), each sample is classified into the cluster with the smallest distance; (6) Update the cluster center, take the average value of the samples in the cluster for numerical features, and take the mode of the samples in the cluster for classification features; (7) Repeat steps (4)-(6) until the clustering result is stable or the change of the objective function is less than the set threshold; (8) Output the final cluster division results and the centers of each cluster.

6. The method for determining prestress of slope anchor cables based on data clustering according to claim 5, characterized in that: The numerical characteristics include: slope gradient, anchor cable service time, anchor plate thickness, anchor cable length, anchor section length, ground temperature, ground humidity and echo frequency; The classification characteristics include: anchor cable specifications, soil / rock layer type, grouting method and construction technology.

7. The method for determining prestress of slope anchor cables based on data clustering according to claim 1, characterized in that: The calculation method of the average silhouette coefficient in step 5 is: Step (1): Let the multi-dimensional data feature corresponding to the anchor in step 3 be sample i, and calculate the average value a(i) of the distance between sample i and all other samples in its cluster; Step (2): Calculate the average value b(i) of the distance between sample i and all samples in the nearest non-self cluster; Step (3): Calculate the silhouette coefficient Sil(i) of each sample using the following method: Step (4): Calculate the average value of the silhouette coefficient of all samples in step (1) 8. The method for determining prestress of slope anchor cables based on data clustering according to claim 1, characterized in that: The silhouette coefficient threshold value ranges from 0.5 to 0.7.

Images