A large water conservancy hub sluice deep foundation pit construction safety control method and system

Abstract

This invention provides a method and system for safety control during the construction of deep foundation pits for spillway gates in large-scale water conservancy projects. Applicable to the field of deep foundation pit construction, the method includes: developing a monitoring plan for the construction of deep foundation pits for spillway gates in large-scale water conservancy projects; arranging monitoring points; obtaining monitoring data; grouping monitoring points; processing monitoring data; constructing a prediction model; training and optimizing the prediction model; and obtaining predicted values ​​and early warnings. This method, by grouping monitoring points according to their types and employing different data processing methods, obtains more accurate predicted values, improves the overall response capability of the monitoring system, and can be widely applied in the field of deep foundation pit construction.

Description

TECHNICAL FIELD

[0001] The present application relates to a large water conservancy hub sluice deep foundation pit construction safety control method and system, which is suitable for the field of deep foundation pit construction. BACKGROUND

[0002] With the continuous development of large-scale water conservancy projects, especially in the construction of water conservancy hubs and sluices, the safety of deep foundation pit construction is increasingly valued. Deep foundation pit construction involves complex geological conditions and dynamic hydrological environment, so during the construction process, various safety hazards are prone to occur, such as foundation pit deformation, soil slip, abnormal water level, etc. If effective monitoring and control are not carried out, serious accidents may occur, causing huge economic losses to the project. Under this background, traditional monitoring methods have been difficult to meet the high requirements of modern engineering construction for safety management. The rapid development of modern technology, especially in the fields of data acquisition, data processing and artificial intelligence, has provided new means and ideas for the safety monitoring and early warning of foundation pit construction. Existing monitoring methods often rely on manual experience and periodic inspection, lack of real-time and systematicness, and are easy to miss key change information. In recent years, the concept of introducing intelligent monitoring systems in deep foundation pit construction has gradually been valued. Through the combination of data acquisition module, data transmission module, data processing module and early warning module, a real-time and accurate monitoring and early warning system can be built. The system can obtain various environmental data of the foundation pit construction in real time by arranging necessary monitoring points, and intelligently evaluate the construction state and predict potential risks by using advanced data processing and analysis technology, thereby providing strong guarantee for construction safety.

[0003] In the aspect of monitoring data processing, the following problems exist in the current safety control system and method for deep foundation pit construction of large-scale water control sluice: 1) Due to the variety of monitoring points, the monitoring data obtained by different types of monitoring points have different data variation characteristics. For example, some points have obvious random variation trend, such as foundation uplift (rebound) monitoring points and soil nail internal force monitoring points, which are mainly affected by various uncertain factors. Some points have obvious periodic variation trend, such as water level monitoring, air temperature and water temperature monitoring, which are mainly affected by periodic factors (seasonal change and daily operation cycle). Some points have obvious increasing or decreasing trend, such as foundation deformation monitoring points and anchor internal force monitoring points. The current research results cannot well classify and analyze various monitoring points. 2) There is lack of effective classification standard. The current research mainly focuses on the variation analysis of single monitoring point, and lacks unified classification standard and analysis method for different monitoring points. Especially for the monitoring points with random, periodic or trend variation, the current tools and technologies cannot effectively group and process them, which limits the response ability of the system when facing complex real-time changes.

[0004] In order to overcome the above problems, it is urgent to develop a more flexible and intelligent safety control system and method for deep foundation pit construction of large-scale water control sluice, which can identify and process data variation characteristics for different types of monitoring points. At the same time, advanced technologies such as deep learning should be combined to explore a multi-level monitoring data processing scheme based on feature extraction and prediction model. SUMMARY

[0005] The purpose of the present application is to provide a safety control method and system for deep foundation pit construction of large-scale water control sluice.

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

[0007] A safety control method for deep foundation pit construction of large-scale water control sluice, comprising the following steps:

[0008] S101, a monitoring scheme for deep foundation pit construction of large-scale water control sluice is established.

[0009] The monitoring scheme for deep foundation pit construction of large-scale water control sluice is established, including establishing a monitoring scheme for deep foundation pit construction of large-scale water control sluice according to the engineering situation and survey results. The monitoring scheme for deep foundation pit construction of large-scale water control sluice includes monitoring content, monitoring project, monitoring equipment and instrument, monitoring point arrangement, monitoring frequency, pre-alarm value setting and monitoring data processing and analysis.

[0010] S102 arranging monitoring points;

[0011] The arrangement of the monitoring points includes arranging m deep foundation construction monitoring points, marked as D1-D m The types of the deep foundation monitoring points include foundation deformation monitoring points, stress monitoring points, water level monitoring points, and foundation site heave monitoring points. The types of the deep foundation monitoring points are divided into n types, denoted as J1 type-J n type;

[0012] S103 obtaining monitoring data;

[0013] The obtaining of the monitoring data includes obtaining corresponding monitoring data Data1-Data m for the arranged deep foundation construction monitoring points D1-D n The monitoring data includes a timestamp of the obtained monitoring data and a specific monitoring value;

[0014] S104 grouping of the monitoring points;

[0015] The grouping of the monitoring points includes grouping the monitoring point types J1 type-J n type using an expert judgment method, into three groups, denoted as Group I, Group II, and Group III. According to the grouping results, all deep foundation construction monitoring points D1-D m are labeled with label A or label B or label C. Finally, the monitoring points labeled with label A are marked as monitoring points G1-G x , the monitoring points labeled with label B are marked as H1-H y , and the monitoring points labeled with label C are marked as K1-K z , where x+y+z=m;

[0016] S105 monitoring data processing;

[0017] The monitoring data processing includes checking and supplementing missing values of the monitoring data of the monitoring points G1-G x , H1-H y , and K1-K z . The missing values of the monitoring data of the monitoring points G1-G x are supplemented using a periodic moving average method. The missing values of the monitoring data of the monitoring points H1-H y are supplemented using a linear interpolation method. The missing values of the monitoring data of the monitoring points K1-K z are supplemented using a multiple interpolation method;

[0018] S106 constructing a prediction model;

[0019] The construction prediction model comprises constructing a recurrent neural network prediction model M x for the monitoring data of monitoring points G1-G G , constructing a recurrent neural network prediction model M y for the monitoring data of monitoring points H1-H H , and constructing a recurrent neural network prediction model M z for the monitoring data of monitoring points K1-K K .

[0020] S107 prediction model training and optimization adjustment;

[0021] The prediction model training and optimization adjustment comprises using model M x to carry out model training and optimization for the monitoring data of monitoring points G1-G G , ultimately obtaining an optimized and adjusted model M G1 , using model M y to carry out model training and optimization for the monitoring data of monitoring points H1-H H , ultimately obtaining an optimized and adjusted model M H1 , using model M z to carry out model training and optimization for the monitoring data of monitoring points K1-K K , and ultimately obtaining an optimized and adjusted model M K1 .

[0022] S108 obtaining prediction values and early warning;

[0023] The obtaining prediction values and early warning comprises analyzing the latest obtained data according to the obtained optimized and adjusted models M G1 , M H1 , and M K1 to obtain prediction values for a future period of time, and comparing the prediction values with early warning values in the large-scale water conservancy hub sluice deep foundation pit construction monitoring scheme. If the prediction values exceed the early warning values, early warning is issued.

[0024] Further, in the grouping step in S104, the steps are as follows:

[0025] a) obtaining the types J i of monitoring points, i=1-n;

[0026] b) using an expert evaluation method to evaluate the data change trend of the data of the J i types of monitoring points, and the evaluation result is that the data change trend of the J i types of monitoring points is one of periodic change trend, tendency change trend, and random change trend;

[0027] c) obtaining J ithe data change trend of the monitoring point G i the data change trend of the monitoring point G i the data change trend of the monitoring point G i the data change trend of the monitoring point G

[0028] d) repeating steps a-c until all the monitoring points G

[0029] Further, in the step S105, the step of filling the missing values by using the periodic moving average method is:

[0030] a) obtaining the monitoring data and the checking situation of the missing values in the monitoring points G i i=1-x;

[0031] b) selecting the period length T according to the periodicity of the data and the position of the missing values;

[0032] c) calculating the average value of a certain number N of data points before and after the position of each missing value as the filling value according to the selected period length T, wherein the number N is obtained by formula (1);

[0033]

[0034] e) repeating steps a-c until all the monitoring points G i complete the filling of the missing values.

[0035] Further, in the step S105, the step of filling the missing values by using the linear interpolation method is:

[0036] a) obtaining the monitoring data and the checking situation of the missing values in the monitoring points H j j=1-y;

[0037] b) obtaining the previous value of each missing value as y 前 and the next value as y 后 ;

[0038] c) calculating the linear interpolation y as the filling value of the missing value, and the calculation formula is shown in formula (2),

[0039]

[0040] wherein ω1 is the weight of the previous value, ω2 is the weight of the next value, x is the time stamp of the missing value, x 前 is the time stamp of the previous value, and x后 the timestamp of the later value;

[0041] d) repeating steps a-c until all monitoring points H j complete the missing value supplement.

[0042] Further, in the step S105, the step of supplementing the missing values by using the multiple imputation method is:

[0043] a) obtaining monitoring points K q , monitoring data q = 1 ~ z and the inspection of missing values;

[0044] b) selecting an imputation model according to the distribution and characteristics of the monitoring data, the imputation model including linear regression, logistic regression and random forest;

[0045] c) using the imputation model to impute each missing value to generate n complete data sets, n being 5-10;

[0046] d) using the Rubin rule to combine multiple imputed values to supplement the missing values;

[0047] e) repeating steps a-d until all monitoring points K q complete the missing value supplement.

[0048] Further, in the step S106, the activation function of the prediction model M G is a combination of hyperbolic tangent function and linear rectified function, the expression of the combination of hyperbolic tangent function and linear rectified function being formula (3), and the combination of hyperbolic tangent function and linear rectified function can better target periodic data. Periodic time series data usually has certain periodicity and regularity, and the combination of hyperbolic tangent function and linear rectified function can learn both the periodic pattern in the data and other potential nonlinear features, so that the model can more accurately capture the trend of change when making predictions.

[0049]

[0050] where ω is a combination coefficient, and the optimal value is determined by learning data in model training.

[0051] Further, in the step S106, the prediction model M HThe activation function of the prediction model M adopts a polynomial and exponential function combination activation function, the expression of the polynomial and exponential function combination activation function is formula (4), the combination of the polynomial and the exponential function is used, the network can have flexible function approximation capability, both processing smooth change of data (polynomial) and capturing rapid growth behavior (exponential), the polynomial and exponential function combination activation function can process the change of different characteristics in the prediction process, and the expression capability of the network can be increased, so that the recurrent neural network can capture various potential dynamic characteristics when facing complex time series data, and the prediction accuracy and robustness of the model are improved.

[0052] f(x) = (az 2 +bz+c)exp(-dz 2 ) (4)

[0053] In the formula, a, b, c and d are coefficients.

[0054] Further, in the step S106, the prediction model M K The activation function of the prediction model M adopts a sine amplification unit function, the expression of the sine amplification unit function is formula (5), the sine amplification unit function can produce output in the negative region and the positive region, which makes the model be able to flexibly respond to different input conditions when processing random data with positive and negative fluctuations. Compared with traditional activation functions (such as ReLU or sigmoid), it can provide more output dynamic range,

[0055] f(x) = x sin(x) (5)

[0056] In the formula, sin(x) is the sine function of variable x.

[0057] A large-scale water conservancy hub sluice deep foundation construction safety control system, specifically includes data acquisition module, data transmission module, data processing module, early warning module, the data acquisition module is used for obtaining monitoring data, the data transmission module is used for transmitting monitoring data to the data processing module, the transmission module is also used for transmitting the early warning result of the data processing module to the early warning module, the data processing module is used for processing monitoring data to obtain early warning result, the early warning module is used for obtaining early warning result and making corresponding operation according to early warning result.

[0058] The present application has the following beneficial effects: 1) intelligent classification analysis: through the classification (periodicity, tendency, randomness, etc.) of the data change trend of different monitoring points, more targeted analysis and processing can be realized, and this intelligent classification method can make the system more flexible and adaptive when facing complex construction environment, and improve the response capability of the whole monitoring system; 2) diversified activation function design: different activation functions (such as the combination of hyperbolic tangent function and linear rectification function, and the combination of polynomial and exponential function) are applied in the prediction model according to the data classification result, so as to improve the targeted fitting capability of the model to the data. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 A flow chart of a large-scale water conservancy hub sluice deep foundation pit construction safety control method of the present application;

[0060] Figure 2 A typical diagram of monitoring point data change of periodic change trend of an embodiment of the present application;

[0061] Figure 3 A typical diagram of monitoring point data change of tendency change trend of an embodiment of the present application;

[0062] Figure 4 A typical diagram of monitoring point data change of random change trend of an embodiment of the present application. DETAILED DESCRIPTION

[0063] The specific embodiments of the present application will be described in detail below with reference to the accompanying drawings; it should be understood that the specific embodiments given here are only for the purpose of description and explanation of the present application, and cannot be used to limit the present application.

[0064] The following is a specific embodiment of a large-scale water conservancy hub sluice deep foundation pit construction safety control method and system.

[0065] A large-scale water conservancy hub sluice deep foundation pit construction safety control method, characterized in that the large-scale water conservancy hub sluice deep foundation pit construction safety control method comprises the following steps:

[0066] S101 Formulate a large-scale water conservancy hub sluice deep foundation pit construction monitoring scheme

[0067] The large-scale water conservancy hub sluice deep foundation pit construction monitoring scheme includes monitoring content, monitoring project, monitoring equipment and instrument, monitoring point arrangement, monitoring frequency, pre-alarm value setting and monitoring data processing and analysis.

[0068] S102 Arrange monitoring points;

[0069] The arrangement monitoring points, including arranging deep foundation construction monitoring points m, marked as D1~D m The types of deep foundation monitoring points include foundation deformation monitoring points, stress monitoring points, water level monitoring points, and foundation site uplift monitoring points. The types of deep foundation monitoring points are divided into n types, denoted as J1~J n ;

[0070] S103 obtaining monitoring data;

[0071] The obtained monitoring data includes corresponding monitoring data Data1~Data m , obtained for the arranged deep foundation construction monitoring points D1~D n , including the timestamp of obtaining monitoring data and the specific monitoring value;

[0072] S104 grouping of monitoring points

[0073] The monitoring point grouping includes grouping the monitoring point types J1~J n using expert evaluation method, divided into three groups, denoted as group I, group II and group III, and according to the grouping result, all deep foundation construction monitoring points D1~D m are labeled A or B or C, and finally the monitoring points labeled A are marked as monitoring points G1~G x , the monitoring points labeled B are marked as H1~H y and the monitoring points labeled C are marked as K1~K z , where x+y+z=m;

[0074] Further in the above step S104, the grouping step is:

[0075] a) obtaining the types of monitoring points J i , i=1~n;

[0076] b) using expert evaluation method to distinguish the data change trend of J i type monitoring point data, and the distinguishing result is that the data change trend of J i type monitoring point is one of periodic change trend, tendency change trend and random change trend;

[0077] c) obtaining the data change trend distinguishing result of J i type monitoring point, when the data change trend distinguishing result of J i type monitoring point is periodic change trend, it is divided into group I, when the data change trend distinguishing result of J i type monitoring point is tendency change trend, it is divided into group II, and when the data change trend distinguishing result of Ji When the data change trend of a monitoring point is determined to be a random change trend, it is classified into Group III;

[0078] d) Repeat steps a to c until all monitoring points are grouped by type.

[0079] In this embodiment, the expert evaluation method is applied to J. i The data change trend of the monitoring points is judged, and the judgment result is the J i The data change trend of monitoring points can be one of the following: periodic trend, trend-based trend, or random trend. A typical graph of data change at monitoring points with periodic trends is shown below. Figure 2 As shown in the figure, a typical graph of the changes in monitoring point data with a tendency to change is shown below. Figure 3 As shown in the figure, a typical graph of the changes in monitoring point data with random variation trends is as follows: Figure 4 As shown.

[0080] S105 monitoring data processing;

[0081] The monitoring data processing includes processing data for monitoring points G1 to G2. x H1~H y and K1~K z The monitoring data is checked for missing values ​​and supplemented with missing values. The monitoring points G1 to G2 are... x Missing values ​​in the monitoring data were supplemented using a periodic moving average method. The monitoring points H1 to H2 were used for this purpose. y For monitoring data with missing values, linear interpolation was used to fill in the missing values. The monitoring points K1 to K2 were used for this purpose. z Missing values ​​in the monitoring data were filled using multiple interpolation.

[0082] S106 Construct a prediction model;

[0083] The construction of the prediction model includes targeting monitoring points G1 to G2. x The monitoring data is used to construct a recurrent neural network prediction model M G For monitoring points H1 to H y The monitoring data is used to construct a recurrent neural network prediction model M H For monitoring points K1 to K z The monitoring data is used to construct a recurrent neural network prediction model M K ;

[0084] S107 Prediction Model Training and Optimization Adjustment;

[0085] The training and optimization of the prediction model includes adjustments for monitoring points G1 to G2. x The monitoring data uses model MG Model training and optimization are carried out, and finally the optimized and adjusted model M G1 The monitoring data of monitoring points H1-H y The monitoring data of monitoring points H1-H H Model training and optimization are carried out, and finally the optimized and adjusted model M H1 The monitoring data of monitoring points K1-K z The monitoring data of monitoring points K1-K K Model training and optimization are carried out, and finally the optimized and adjusted model M K1 ;

[0086] S108 obtains the prediction value and early warning;

[0087] The obtained prediction value and early warning include analyzing the latest obtained data according to the obtained optimized and adjusted models M G1 , M H1 and M K1 to obtain the prediction value in the future period of time, and comparing the prediction value with the early warning value in the large-scale water conservancy hub sluice deep foundation pit construction monitoring scheme. If it exceeds the early warning value, an early warning is issued.

[0088] Further, in the step S105, the step of supplementing the missing value by using the periodic moving average method is:

[0089] a) Obtain the monitoring data of monitoring points G i , i=1-x and the inspection of missing values;

[0090] b) Select the period length T according to the periodicity of the data and the position of the missing value;

[0091] c) According to the selected period length T, for the position of each missing value, calculate the average value of a certain number N of data points before and after it as the supplement value, and the number N is obtained by formula (1);

[0092]

[0093] e) Repeat steps a-c until all monitoring points G i complete the supplement of missing values.

[0094] Further, in the step S105, the step of supplementing the missing value by using the linear interpolation method is:

[0095] a) Obtain the monitoring data of monitoring points H j , j=1-y and the inspection of missing values;

[0096] b) For each missing value, obtain its previous value represented as y前 , the obtained post-value is represented as y 后 ;

[0097] c) calculating a linear interpolation value y as a supplementary value of the missing value, according to formula (2), wherein ω1 is the weight of the pre-value, ω2 is the weight of the post-value, x is the time stamp of the missing value, x 前 is the time stamp of the pre-value, and x 后 is the time stamp of the post-value;

[0098]

[0099] In the formula, ω1 is the weight of the pre-value, ω2 is the weight of the post-value, x is the time stamp of the missing value, x 前 is the time stamp of the pre-value, and x 后 is the time stamp of the post-value;

[0100] d) repeating steps a-c until all monitoring points H j complete the supplement of the missing values.

[0101] Further, in the step S105, the step of supplementing the missing values by using the multiple imputation method is:

[0102] a) obtaining monitoring data and inspection conditions of missing values in monitoring points K q , q = 1-z;

[0103] b) selecting an imputation model according to the distribution and characteristics of the monitoring data, wherein the imputation model includes linear regression, logistic regression, and random forest;

[0104] c) using the imputation model to impute each missing value to generate n complete data sets, wherein n is 5-10;

[0105] d) using the Rubin rule to merge multiple imputation values to supplement the missing values;

[0106] e) repeating steps a-d until all monitoring points K q complete the supplement of the missing values.

[0107] Further, in the step S106, the activation function of the prediction model M G uses a combination of hyperbolic tangent function and linear rectified function as the activation function, and the expression of the combination of hyperbolic tangent function and linear rectified function is formula (3). The combination of hyperbolic tangent function and linear rectified function can better deal with periodically changing data. Periodic time series data usually has certain periodicity and regularity. The combination of hyperbolic tangent function and linear rectified function can learn the periodic pattern in the data and other potential nonlinear features at the same time, so that the model can more accurately capture the change trend when making predictions.

[0108]

[0109] where ω is a combination coefficient, and an optimal value is determined by learning data in model training.

[0110] Further, in the step S106, the prediction model M H uses a polynomial and exponential function combination activation function, and the polynomial and exponential function combination activation function has an expression of formula (4).

[0111] f(x)=(az 2 +bz+c)exp(-dz 2 ) (4)

[0112] where a, b, c, and d are coefficients.

[0113] Further, in the step S106, the prediction model M K uses a sine amplification unit function, and the sine amplification unit function has an expression of formula (5).

[0114] f(x)=xsin(x) (5)

[0115] where sin(x) is a sine function of the variable x.

[0116] A large water conservancy hub sluice deep foundation pit construction safety control system, specifically includes data acquisition module, data transmission module, data processing module, early warning module, the data acquisition module is used for obtaining monitoring data, the data transmission module is used for transmitting monitoring data to the data processing module, the transmission module is also used for transmitting the early warning result of the data processing module to the early warning module, the data processing module is used for processing monitoring data to obtain early warning result, the early warning module is used for obtaining early warning result and making corresponding operation according to early warning result.

[0117] In the above embodiment, the application discloses a large-scale water conservancy hub sluice deep foundation pit construction safety control method and system, a large-scale water conservancy hub sluice deep foundation pit construction safety control system includes a data acquisition module, a data transmission module, a data processing module, an early warning module, a large-scale water conservancy hub sluice deep foundation pit construction safety control method, including formulating a large-scale water conservancy hub sluice deep foundation pit construction monitoring scheme, arranging monitoring points, obtaining monitoring data, grouping monitoring points, monitoring data processing, constructing a prediction model, training and optimizing the prediction model, and obtaining a prediction value and early warning; the method considers the grouping of the types of monitoring points, uses different data processing methods, obtains more accurate prediction values, improves the response capability of the overall monitoring system, and can be widely applied to the field of deep foundation pit construction.

[0118] The above is the preferred embodiment of the application, which does not limit the application, and any modification, equivalent replacement, improvement, etc. within the spirit and principle of the application should be included in the protection scope of the application.

Claims

1. A method for safety control during the construction of deep foundation pits for spillway gates in large-scale water conservancy projects, characterized in that, The steps of the safety control method for deep foundation pit construction of a large-scale water conservancy hub spillway gate are as follows: S101 developed a monitoring plan for the construction of deep foundation pits for spillway gates of large-scale water conservancy projects; S102 is used to set up monitoring points; S103 obtains monitoring data; Grouping of monitoring points S104; S105 monitoring data processing; S106 Construct a prediction model; S107 Prediction Model Training and Optimization Adjustment; S108 obtains predicted values ​​and early warnings; The aforementioned development of a monitoring plan for the deep foundation pit construction of a large-scale water conservancy hub spillway includes, based on the project conditions and survey results, the development of a monitoring plan for the deep foundation pit construction of a large-scale water conservancy hub spillway, which includes monitoring content, monitoring items, monitoring equipment and instruments, layout of monitoring points, monitoring frequency, setting of early warning values, and processing and analysis of monitoring data. The monitoring points to be set up include a total of m monitoring points for deep foundation pit construction, marked as D1 to D2. m The types of monitoring points for deep foundation pit construction include, but are not limited to, foundation pit deformation monitoring points, stress monitoring points, water level monitoring points, and foundation pit base heave monitoring points. These monitoring points are categorized into n types, referred to as J1 to J2. n kind; The obtained monitoring data includes data from monitoring points D1 to D2 of the deep foundation pit construction site. m Obtain the corresponding monitoring data Data1~Data n The monitoring data includes the timestamp of the acquisition of the monitoring data and the specific monitoring value; The monitoring points are grouped, including the monitoring points of categories J1 to J2, which are determined by expert evaluation. n The data were grouped into three groups, denoted as Group I, Group II, and Group III. Based on the grouping results, all deep foundation pit construction monitoring points D1 to D2 were grouped. m Apply label A, label B, or label C, and finally, mark the monitoring points with label A as monitoring points G1 to G2. x The monitoring points for tag B are marked as H1 to H2. y The monitoring points marked with tag C are K1 to K1. z Where x + y + z = m; The monitoring data processing includes processing data for monitoring points G1 to G2. x H1~H y and K1~K z The monitoring data is checked for missing values ​​and supplemented with missing values. The monitoring points G1 to G2 are... x Missing values ​​in the monitoring data were supplemented using a periodic moving average method. The monitoring points H1 to H2 were used for this purpose. y Missing values ​​in the monitoring data were supplemented using linear interpolation. The monitoring points K1 to K2 were used for this purpose. z Missing values ​​in the monitoring data were filled using multiple interpolation. The construction of the prediction model includes targeting monitoring points G1 to G2. x The monitoring data is used to construct a recurrent neural network prediction model M G For monitoring points H1 to H y The monitoring data is used to construct a recurrent neural network prediction model M H For monitoring points K1 to K z The monitoring data is used to construct a recurrent neural network prediction model M K ; The training and optimization of the prediction model includes adjustments for monitoring points G1 to G2. x The monitoring data uses model M G The model is trained and optimized to obtain the optimized model M. G1 For monitoring points H1 to H y The monitoring data uses model M H The model is trained and optimized to obtain the optimized model M. H1 For monitoring points K1 to K z The monitoring data uses model M K The model is trained and optimized to obtain the optimized model M. K1 ; The acquisition of predicted values ​​and early warnings includes obtaining the optimized and adjusted model M. G1 M H1 and M K1 The latest data is analyzed to obtain predicted values ​​for a future period. These predicted values ​​are then compared with the early warning values ​​in the monitoring plan for the deep foundation pit construction of large-scale water conservancy hub spillway gates. If the predicted values ​​are exceeded, an early warning is issued.

2. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S104, the monitoring points are classified into categories J1 to J2 using an expert evaluation method. n The specific steps for grouping classes are as follows: a) Types of monitoring points obtained J i Class, i = 1 to n; b) Using expert evaluation method to evaluate J i The data change trend of the monitoring points is judged, and the judgment result is the J i The data change trend of the monitoring points is one of the following: periodic change trend, trend-based change trend, and random change trend; c) Obtain J i The data change trend judgment results of monitoring points of type J, when J i When the data change trend of a monitoring point is determined to be periodic, it is assigned to Group I. When J... i When the data change trend of a monitoring point is determined to be a trend, it is assigned to Group II. i When the data change trend of a monitoring point is determined to be a random change trend, it is classified into Group III; d) Repeat steps a to c until all monitoring points are grouped by type.

3. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S105, the step of using a periodic moving average method to fill in missing values ​​is as follows: a) Obtain monitoring point G i The monitoring data in i = 1 to x and the status of missing value checks; b) Select the period length T based on the periodicity of the data and the location of the missing values; c) Based on the selected period length T, for each missing value, calculate the average of a certain number N data points before and after it as a supplementary value. The number N is calculated by formula (1). e) Repeat steps a through c until all monitoring points G are reached. i All missing values ​​were filled in.

4. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S105, the step of using linear interpolation to fill in missing values ​​is as follows: a) Obtain monitoring point H j The monitoring data in j=1 to y and the status of missing value checks; b) For each missing value, obtain its preceding value, represented as y. 前 The value obtained is represented as y. 后 ; c) Calculate the linear interpolation y as a supplementary value for the missing values. The calculation formula is shown in equation (2). In the formula, ω1 is the weight of the previous value, ω2 is the weight of the subsequent value, and x is the timestamp of the missing value. 前 The timestamp of the previous value, x 后 The timestamp is the value after the timestamp; d) Repeat steps a to c until all monitoring points H are reached. j All missing values ​​were filled in.

5. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S105, the step of using multiple interpolation to fill in missing values ​​is as follows: a) Obtain monitoring point K q The monitoring data in q = 1 to z and the status of missing value checks; b) Select an interpolation model based on the distribution and characteristics of the monitoring data. The interpolation model includes linear regression, logistic regression, and random forest. c) Use an imputation model to impute each missing value and generate n complete datasets, where n ranges from 5 to 10; d) Use Rubin's rule to merge multiple imputed values ​​to fill in missing values; e) Repeat steps a through d until all monitoring points K are reached. q All missing values ​​were filled in.

6. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S106, the activation function of the prediction model MG is a combination activation function of hyperbolic tangent function and linear rectified function, and the expression of the combination activation function of hyperbolic tangent function and linear rectified function is Equation (3). In the formula, ω is the combination coefficient, the optimal value of which is determined by learning data during model training.

7. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S106, the activation function of the prediction model MH is a combination of polynomial and exponential activation functions, and the expression of the combination of polynomial and exponential activation functions is Equation (4). f(x)=(az 2 +bz+c)exp(-dz 2 ) (4) In the formula, a, b, c, and d are coefficients.

8. The method for safety control during the construction of a deep foundation pit for a spillway gate of a large-scale water conservancy project according to claim 1, characterized in that, In step S106, the activation function of the prediction model MK is a sinusoidal amplification unit function, and the expression of the sinusoidal amplification unit function is equation (5). f(x)=xsin(x) (5) In the formula, sin(x) is the sine function of the variable x.

9. A safety control system for the construction of deep foundation pits for spillway gates of large-scale water conservancy projects, used in the safety control method for the construction of deep foundation pits for spillway gates of large-scale water conservancy projects as described in any one of claims 1 to 8, characterized in that: Specifically, it includes a data acquisition module, a data transmission module, a data processing module, and an early warning module. The data acquisition module is used to acquire monitoring data, the data transmission module is used to transmit the monitoring data to the data processing module, the data transmission module is also used to transmit the early warning results of the data processing module to the early warning module, the data processing module is used to process the monitoring data to obtain the early warning results, and the early warning module is used to perform corresponding operations based on the early warning results after obtaining them.

Images