A method for monitoring and warning the settlement of tunnel vault

Abstract

The present invention provides a method for monitoring and warning the settlement of the tunnel vault, which is applicable to the field of tunnel construction monitoring, including point layout, data acquisition, data analysis, and achieving the purpose of warning. The method for monitoring and warning the settlement of the tunnel vault proposed by the present invention has the advantages of high accuracy and strong comprehensiveness, and can be widely applied to the field of tunnel construction monitoring.

Description

Technical Field

[0001] The present invention relates to a method for monitoring and warning the settlement of the tunnel vault, which is applicable to the field of tunnel construction monitoring. Background Art

[0002] Tunnel construction monitoring is an important link to ensure the safety, quality and progress of tunnel projects. During tunnel construction, the settlement of the vault is a key monitoring index, as it is directly related to the stability of the tunnel structure and construction safety. The research and application of methods for monitoring and warning the settlement of the tunnel vault are of great significance for preventing accidents during tunnel construction, protecting personnel safety and extending the service life of the tunnel.

[0003] At present, with the continuous development of computer technology and the continuous extension of the cross - application of multiple disciplines, in the analysis of tunnel vault settlement, using advanced algorithms for analysis and prediction is a hot topic in current industry research. Currently, industry insiders mainly consider the tunnel vault settlement monitoring data as time - series data, and use some algorithms for time - series data to analyze or build models, etc., and finally obtain predicted values to guide construction.

[0004] However, in the current research status, the fit between the change situation of tunnel vault settlement and the prediction algorithm adopted is not fully considered, resulting in the prediction results not conforming to the actual construction, and even hindering the normal progress of construction. The main reason is that the fit between the change situation of tunnel vault settlement and the prediction algorithm adopted is not considered. The change rate of tunnel vault settlement varies. For example, the monitoring results of the vault settlement in sections with shallow burial depth and poor surrounding rock conditions change relatively fast, while the monitoring results of the vault settlement in sections with relatively good surrounding rock conditions change relatively slowly. At the same time, there are many algorithms for time - series data, some of which are good at analyzing rapidly changing data, and some are conducive to capturing and analyzing slowly changing data. Therefore, there is an urgent need for a method for monitoring and warning the settlement of the tunnel vault that fully considers the fit between the change situation of tunnel vault settlement and the prediction algorithm adopted. Summary of the Invention

[0005] The purpose of the present invention is to solve the problem that there is an urgent need for a method for monitoring and warning the settlement of the tunnel vault that fully considers the fit between the change situation of tunnel vault settlement and the prediction algorithm adopted, and a method for monitoring and warning the settlement of the tunnel vault is proposed.

[0006] The purpose of the present invention can be achieved by adopting the following technical solutions:

[0007] A method for monitoring and warning the settlement of the tunnel vault, characterized in that the steps of the method for monitoring and warning the settlement of the tunnel vault are as follows:

[0008] S101, Point layout;

[0009] The point layout includes arranging settlement monitoring points at the crown according to the excavation method adopted for the tunnel. The number of settlement monitoring points at the crown is i, denoted as M i , and the distance between the settlement monitoring points at the crown is adjusted according to the actual excavation situation;

[0010] S102, Data acquisition;

[0011] The data acquisition includes classifying the crown monitoring points according to the different cross-sections they are located in and dividing them into two categories: high-deformation-expectation crown settlement monitoring points H j and low-deformation-expectation crown settlement monitoring points L k , where j + k = i. The classification method is as follows: a) Explore and collect data on the burial depth and topographic and geological conditions of the cross-sections where the crown settlement monitoring points are located. b) When the burial depth of the cross-section is greater than or equal to x meters and the surrounding rock grade of the cross-section is at levels Ⅰ to Ⅳ, the crown settlement monitoring points of this cross-section are low-deformation-expectation crown settlement monitoring points at this time. c) When the burial depth of the cross-section is less than x meters or the surrounding rock grade is less than or equal to grade Ⅴ, then the crown settlement monitoring points of this cross-section are high-deformation-expectation crown settlement monitoring points at this time;

[0012] The data acquisition also includes obtaining datasets containing monitoring timestamps by monitoring the high-deformation-expectation crown settlement monitoring points H j and low-deformation-expectation crown settlement monitoring points L k at the frequencies required by relevant specifications, which are respectively the high-deformation-expectation dataset C j and the low-deformation-expectation dataset D k ;

[0013] S103, Data analysis;

[0014] The data analysis includes performing classified and directional data analysis on the obtained datasets containing monitoring timestamps to obtain predicted values P corresponding to each crown settlement monitoring point. The steps of the classified and directional data analysis are as follows:

[0015] a) Organize the datasets containing monitoring timestamps. The organization includes organizing the datasets containing monitoring timestamps into datasets with consistent time intervals,

[0016] b) When the datasets containing monitoring timestamps are high-deformation-expectation datasets C j , use a long short-term memory neural network that can capture rapid changes to construct a prediction model to obtain predicted values,

[0017] c) When the datasets containing monitoring timestamps are low-deformation-expectation datasets Dk When it is, the prediction value is obtained by using the linear regression method capable of capturing slow changes;

[0018] S104, achieving the warning purpose;

[0019] The achievement of the warning purpose includes judging the stability state according to the obtained prediction value P in combination with the corresponding vault settlement monitoring warning standard, and achieving the warning purpose. The corresponding vault settlement monitoring warning standard is comprehensively formulated according to the section topography and geology conditions and excavation conditions of the monitoring point where the prediction value P is obtained;

[0020] Further, in the above S103, in the data analysis, the time interval is 0.5 days or 1 day;

[0021] Further, in the above S103, the steps of constructing a prediction model by using the long short-term memory neural network to obtain a prediction value are: a) data normalization processing, b) dividing the data set into a training set and a test set, c) constructing a long short-term memory neural network prediction model, d) model training, e) model evaluation, f) model deployment and application, and obtaining a prediction value;

[0022] Further, in the above construction of the long short-term memory neural network prediction model, the smooth rectified linear unit function is used as the activation function, and the expression of the smooth rectified linear unit function is formula (1),

[0023] f(x) = ln(1 + e x ) (1)

[0024] In the formula, x is the input and f(x) is the output;

[0025] Further, in the above S103, the steps of obtaining a prediction value by using the linear regression method are: a) data preprocessing, including data cleaning and conversion, b) data division, c) establishing a univariate linear regression model, e) evaluating model parameters by using the least squares method, f) evaluating model performance by using the test set, g) model optimization, h) model prediction to obtain a prediction value.

[0026] The beneficial effects of the present invention are:

[0027] The beneficial effects of the present invention are: The compatibility between the change of the tunnel vault settlement and the adopted prediction algorithm is comprehensively considered, so that the prediction value obtained by using this method can more effectively guide the construction. Description of the Drawings

[0028] Figure 1 : Flow chart of a method for monitoring and warning tunnel vault settlement of the present invention;

[0029] Figure 2: The measured value and predicted value curve diagram of section number TJ-12 of the embodiment of the present invention;

[0030] Figure 3 : Measured value and predicted value curve diagram of section number TJ-12 of an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings; it should be understood that the specific embodiments given here are only used to illustrate and explain the present invention and cannot be used to limit the present invention.

[0032] The following is a specific embodiment of a tunnel vault settlement monitoring and early warning method.

[0033] like Figure 1 FIG. 1 is a flow chart of a tunnel vault settlement monitoring and early warning method according to the present invention.

[0034] A tunnel vault settlement monitoring and early warning method, characterized in that the tunnel vault settlement monitoring and early warning method comprises the following steps:

[0035] S101, point arrangement;

[0036] The point arrangement includes arranging the arch settlement monitoring points in a targeted manner according to the excavation method adopted by the tunnel. The number of the arch settlement monitoring points is i, represented by M i , the distance between the vault settlement monitoring points is adjusted according to the actual excavation situation;

[0037] In this embodiment, the name of the project in the embodiment is the Eighth Sub - division of the General Manager's Department of the China Communications Construction Company's East Coast Railway Project in Malaysia, namely the Gombak No. 1 Tunnel. The total length of this tunnel is 935 m. Among them, the mileage at the entrance is CH516 + 955, and the mileage at the exit is CH517 + 890. The gradient from the starting point of the tunnel to CH517 + 000 is - 3‰, and the gradient from CH517 + 000 to the end of the tunnel is 5.8‰. The maximum buried depth is 74 m, and the minimum buried depth is 9 m. Among them, the section from CH517 + 160 to CH517 + 260 is the shallow - buried section of the tunnel body, with the shallowest buried depth of 10.5 m. The tunnel is designed with a composite lining structure, the standard line spacing is 4.2 m, the entrance portal is of the end - wall type, and the exit portal is of the bevel - cut type. The geological lithology of the tunnel is mainly medium - weathered granite, and the rock mass is relatively fractured. The surface vegetation of the Gombak Tunnel is well - developed, all of which are mature tree forests. The entrance end is located under the high - voltage line corridor, and high - voltage line relocation is required; the traffic conditions at the exit end are convenient, close to the E8 highway and adjacent to the local existing road. The entrance, exit and shallow - buried sections of the Gombak Tunnel are mainly composed of residual diluvial clay, fine sand and granite from completely weathered to medium - weathered, with the rock mass broken to extremely broken. The predicted normal seepage volume is 2361.18 m³ / day; the deep - buried section of the tunnel is mainly medium - weathered granite, with the rock mass relatively fractured. The predicted normal seepage volume is 2361.18 m³ / day, and the maximum seepage volume is 9131.62 m³ / day.

[0038] In this embodiment, the Gombak No. 1 Tunnel is arranged with vault settlement monitoring points according to the specification requirements.

[0039] S102, Data acquisition;

[0040] The data acquisition includes classifying the vault monitoring points according to the different sections they are in and dividing them into two categories: high - deformation - expectation vault settlement monitoring points H j and low - deformation - expectation vault settlement monitoring points L k , where j + k = i. The classification method is as follows: a) Explore and collect data on the buried depth and topographic and geological conditions of the section where the vault settlement monitoring points are located. b) When the buried depth of the section is greater than or equal to x meters and the surrounding rock grade of the section is at levels Ⅰ - Ⅳ, then the vault settlement monitoring points of this section are low - deformation - expectation vault settlement monitoring points. c) When the buried depth of the section is less than x meters or the surrounding rock grade is less than or equal to grade Ⅴ, then the vault settlement monitoring points of this section are high - deformation - expectation vault settlement monitoring points;

[0041] The data acquisition also includes obtaining datasets containing monitoring timestamps according to the high - deformation - expectation vault settlement monitoring points H j and low - deformation - expectation vault settlement monitoring points L k at the frequencies required by relevant specifications, which are respectively the high - deformation - expectation dataset C j and the low - deformation - expectation dataset D k;

[0042] In this embodiment, the vault settlement monitoring points of the tunnel are classified into high-deformation-expectation vault settlement monitoring points and low-deformation-expectation vault settlement monitoring points according to requirements. After classification according to requirements, the typical point numbers of the high-deformation-expectation vault settlement monitoring points are selected as TJ-12, and the typical monitoring point of the low-deformation-expectation vault settlement monitoring points is TJ-35. Monitoring is carried out according to relevant specifications to obtain their monitoring data sets;

[0043] S103, data analysis;

[0044] The data analysis includes performing classified directional data analysis on the obtained data set containing monitoring timestamps to obtain the predicted value P corresponding to each vault settlement monitoring point. The steps of the classified directional data analysis are as follows:

[0045] a) Organize the data set containing monitoring timestamps. The organization includes organizing the data set containing monitoring timestamps into a data set with a consistent time interval.

[0046] b) When the data set containing monitoring timestamps is a high-deformation-expectation data set C j At this time, a prediction model is constructed using a long short-term memory neural network capable of capturing rapid changes to obtain the predicted value.

[0047] c) When the data set containing monitoring timestamps is a low-deformation-expectation data set D k At this time, the linear regression method capable of capturing slow changes is used to obtain the predicted value;

[0048] Furthermore, in the above S103, in the data analysis, the time interval is 0.5 days or 1 day;

[0049] Furthermore, in the above S103, the steps for the long short-term memory neural network to construct a prediction model to obtain the predicted value are as follows: a) data normalization processing, b) dividing the data set into a training set and a test set, c) constructing a long short-term memory neural network prediction model, d) model training, e) model evaluation, f) model deployment and application to obtain the predicted value;

[0050] Furthermore, the construction of the long short-term memory neural network prediction model includes using the smooth rectified linear unit function as the activation function. The expression of the smooth rectified linear unit function is formula (1).

[0051] f(x) = ln(1 + e x ) (1)

[0052] In the formula, x is the input and f(x) is the output;

[0053] Further, in the above S103, the steps of obtaining the predicted value by the linear regression method are as follows: a) Data preprocessing, including data cleaning and transformation; b) Data division; c) Establishing a univariate linear regression model; e) Evaluating the model parameters using the least squares method; f) Evaluating the model performance using the test set; g) Model optimization; h) Obtaining the predicted value through model prediction.

[0054] In this embodiment, for the typical section numbers TJ-12 and TJ-35, the section number TJ-12 is the high point of the deformation expectation. A prediction model is constructed using a long short-term memory neural network that can capture rapid changes to obtain the predicted value. The curve of the measured value and the predicted value of the section number TJ-12 is as Figure 2 shown. The section number TJ-35 is the low point of the deformation expectation. The predicted value is obtained using the linear regression method that can capture slow changes. The curve of the measured value and the predicted value of the section number TJ-12 is as Figure 3 shown;

[0055] S104, Achieving the warning purpose;

[0056] The achievement of the warning purpose includes judging the stability state according to the obtained predicted value P in combination with the corresponding arch crown settlement monitoring warning standard, so as to achieve the warning purpose. The corresponding arch crown settlement monitoring warning standard is formulated comprehensively according to the section topography and geology conditions and the excavation conditions of the monitoring point where the predicted value P is obtained;

[0057] In this embodiment, the arch crown settlement monitoring standard is shown in Table 1. The predicted values of the section numbers TJ-12 and TJ-35 do not exceed the arch crown settlement monitoring standard, and it is judged that they are in a stable state.

[0058] Table 1 Arch Crown Settlement Monitoring Standard

[0059]

[0060] In the above embodiment, the present invention discloses a method for monitoring and warning the arch crown settlement of a tunnel, including point layout, data acquisition, data analysis, and achieving the warning purpose. The method for monitoring and warning the arch crown settlement of a tunnel proposed by the present invention has the advantages of high accuracy and strong comprehensiveness, and can be widely applied to the field of tunnel construction monitoring.

[0061] The above is the preferred embodiment of the present invention, and it does not limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A tunnel vault settlement monitoring and early warning method, characterized in that: The steps of the tunnel vault settlement monitoring and early warning method are as follows: 1) Point arrangement; 2) Data acquisition; 3) Data analysis; 4) Achievement of early warning objectives; The point arrangement includes arranging the arch settlement monitoring points in a targeted manner according to the excavation method adopted by the tunnel. The number of the arch settlement monitoring points is i, represented by M i , the distance between the vault settlement monitoring points is adjusted according to the actual excavation situation; The data acquisition includes classifying the vault settlement monitoring points according to the different sections where they are located and dividing them into two categories: deformation expectation high vault settlement monitoring points H j and deformation expectation low arch settlement monitoring point L k , where j+k=i, the classification method is: a) exploring and collecting data on the buried depth and topographic geological conditions of the section where the arch settlement monitoring point is located, b) when the buried depth of the section is greater than or equal to x meters, and the surrounding rock grade of the section is level I to IV, the arch settlement monitoring point of the section is a low-expectation arch settlement monitoring point, c) when the buried depth of the section is less than x meters or the surrounding rock grade is less than or equal to level V, the arch settlement monitoring point of the section is a high-expectation arch settlement monitoring point; The data acquisition also includes the following steps: j and deformation expectation low arch settlement monitoring point L k The monitoring frequency required by the relevant specifications is used to obtain the data sets containing monitoring timestamps, which are the deformation expectation high data sets C j and the low deformation expectation dataset D k ; The data analysis includes performing classified directional data analysis on the obtained data set containing the monitoring timestamp to obtain a predicted value P corresponding to each vault settlement monitoring point. The steps of the classified directional data analysis are: a) arranging the data set containing the monitoring timestamps, wherein the arranging includes arranging the data set containing the monitoring timestamps into data sets with consistent time intervals, b) When the data set containing the monitoring timestamp is a high deformation expectation data set C j When the prediction value is obtained, a prediction model is constructed using a long short-term memory neural network that can capture fast changes. c) When the data set containing the monitoring timestamp is a low deformation expectation data set D k When , the linear regression method capable of capturing slow changes is used to obtain the predicted value; The achievement of the early warning purpose includes judging the stable state according to the obtained prediction value P in combination with its corresponding arch settlement monitoring and early warning standard to achieve the purpose of early warning. The corresponding arch settlement monitoring and early warning standard is comprehensively formulated based on the topographic and geological conditions and excavation conditions of the section where the monitoring point where the prediction value P is obtained is located.

2. A tunnel vault settlement monitoring and early warning method according to claim 1, characterized in that: In the data analysis, the time interval is 0.5 day or 1 day.

3. A tunnel vault settlement monitoring and early warning method according to claim 1, characterized in that: The steps of constructing a prediction model of the long short-term memory neural network to obtain a prediction value are: a) data normalization processing, b) dividing the data set into a training set and a test set, c) constructing a long short-term memory neural network prediction model, d) model training, e) model evaluation, and f) model deployment and application to obtain a prediction value.

4. A tunnel vault settlement monitoring and early warning method according to claim 1, characterized in that: The steps of obtaining the predicted value by the linear regression method are: a) data preprocessing, including data cleaning and conversion, b) data partitioning, c) establishing a univariate linear regression model, e) evaluating model parameters using the least squares method, f) evaluating model performance using a test set, g) model optimization, and h) model prediction to obtain the predicted value.

5. A tunnel vault settlement monitoring and early warning method according to claim 3, characterized in that: The construction of the long short-term memory neural network prediction model includes using a smoothed corrected linear unit function as an activation function, and the expression of the smoothed corrected linear unit function is formula (1): f(x)=ln(1+e x )(1) Where x is the input and f(x) is the output.

Images