A ship lock main body deep foundation pit construction water level monitoring system and method

Abstract

The application provides a ship lock main body deep foundation pit construction water level monitoring system and method, which is suitable for the field of deep foundation pit construction, and a large water conservancy hub sluice deep foundation pit construction safety control system, which comprises a water level monitoring module, a water level monitoring data transmission and receiving module, a water level monitoring data processing module and a water level monitoring early warning module; a ship lock main body deep foundation pit construction water level monitoring method, which comprises arranging water level monitoring points, obtaining water level monitoring data, preliminary processing of monitoring data, data analysis of single water level monitoring points, data analysis of overall water level distribution and realization of water level early warning; the method can better predict the water level change of the ship lock main body deep foundation pit by analyzing and processing the monitoring data of the upstream water level of the ship lock main body deep foundation pit, thereby ensuring the construction safety, and can be widely applied to the field of deep foundation pit construction.

Description

TECHNICAL FIELD

[0001] The present application relates to a ship lock main body deep foundation pit construction water level monitoring system and method, which is suitable for the field of deep foundation pit construction. BACKGROUND

[0002] In modern engineering construction, deep foundation pit construction is a highly difficult operation, especially in the construction process of water conservancy projects such as ship lock main body, the change of water level has an important influence on the stability and safety of the construction of the foundation pit. Therefore, it is particularly necessary to establish an efficient water level monitoring system. The traditional water level monitoring method relies on manual measurement, which not only consumes time and effort, but also is difficult to realize real-time monitoring, and is easy to cause missed measurement and data lag. With the progress of sensor technology and communication technology, water level monitoring systems based on Internet of Things technology have emerged. This system can realize automatic water level monitoring, provide real-time data and conduct data analysis, thereby reducing construction risks and improving construction efficiency.

[0003] In the process of deep foundation pit construction, water level monitoring not only requires high-precision sensors, but also needs effective data processing and analysis methods to identify potential safety hazards in a timely manner. Based on this, a water level monitoring module is used to obtain real-time water level data by arranging multiple water level monitoring points, and the data is transmitted to the processing center through the data transmission and reception module. The data processing module will preliminarily process the monitoring data, including outlier identification and supplementation, to ensure the accuracy and reliability of the data.

[0004] At the same time, there are many problems in the field of ship lock main body deep foundation pit construction water level monitoring, the main problems are: 1) The data of the water level monitoring points arranged near the deep foundation pit is distorted; the reason for this phenomenon is that the deep foundation pit will have frequent pumping and water releasing operations during the construction process, thus causing irregular changes in the water level around the deep foundation pit, which also leads to very abnormal water level monitoring data near the deep foundation pit, making analysis very difficult; 2) The change of the water level of the ship lock main body deep foundation pit is closely related to the water level of its upstream, so in order to obtain the real change of the water level of the deep foundation pit, monitoring points can be arranged upstream for data analysis and processing, however, the data of the upstream water level monitoring points presents two obvious characteristics, characteristic 1 is that the water level monitoring data of a single monitoring point presents obvious periodic changes, the main reason for this is the influence of periodic factors such as tides and day-night changes, characteristic 2 is that at the same monitoring time, the monitoring data obtained by continuous monitoring point arrangement often presents more complex nonlinear conditions, the main reason is that the influencing factors are too complex, mainly including the influence of rainfall, reservoir scheduling, hydrological and meteorological factors, etc., in data processing, reasonable processing methods should be selected according to these factors to better utilize the monitoring data for construction control.

[0005] In summary, in order to overcome the above problems, it is urgent to develop a more comprehensive and more intelligent ship lock main body deep foundation construction water level monitoring system and method. SUMMARY

[0006] The purpose of the present application is to address the urgent need for a more comprehensive and more intelligent ship lock main body deep foundation construction water level monitoring system and method, a ship lock main body deep foundation construction water level monitoring system and method are proposed.

[0007] The purpose of the present application can be achieved by adopting the following technical solutions:

[0008] A ship lock main body deep foundation construction water level monitoring method, characterized in that the ship lock main body deep foundation construction water level monitoring method comprises the following steps:

[0009] S101 arranging water level monitoring points;

[0010] The arrangement of water level monitoring points comprises arranging water level monitoring points in a certain interval distance L1 from upstream to downstream of the water level of the ship lock main body deep foundation construction, a total of n, marked as J i , i = 1 ~ n, and J i The distance L i from the deep foundation pit is L i , which is obtained by formula (1);

[0011] L i = L1×(n+1-j) (1)

[0012] S102 obtaining water level monitoring data;

[0013] The water level monitoring data comprises obtaining monitoring data at fixed time intervals for the arranged water level monitoring points J i , assuming that m times of monitoring data have been obtained, the time of obtaining monitoring data is T j , j = 1 ~ m, and the time T j The monitoring data obtained for the water level monitoring point J i is represented as J ij , and the monitoring data set obtained for the water level monitoring point J i is C i = {J ij |j = 1 ~ m};

[0014] S103 preliminary processing of monitoring data;

[0015] The preliminary processing of monitoring data comprises preliminary processing of the obtained data set C i , which comprises identification and supplement of missing values and identification and supplement of abnormal values;

[0016] Data analysis of a single water level monitoring point S104;

[0017] The data analysis of the single water level monitoring point includes data from water level monitoring point J. i The obtained monitoring data set is C i ={J ij |j=1~m}, data analysis is performed to obtain the water level monitoring point J. i Future Time T j+1 ~T j+x The predicted value J i(j+1) ~J i(j+x) ;

[0018] Data analysis of the overall water level distribution in S105;

[0019] The data analysis of the overall water level distribution includes obtaining future time T y The set of monitoring data from all monitoring points in the range y = (j+1) to (j+x) is C. y ={J iy |i=1~n}, construct the future time T y The analysis model of the monitoring data and the acquisition of future time T y Predicted water level J at deep foundation pit (n+1)y ;

[0020] The implementation of S106 water level early warning;

[0021] The implementation of the water level early warning includes combining the obtained future time T y Predicted water level J at deep foundation pit (n+1)y The system analyzes the early warning criteria to obtain early warning results, and then decides whether to issue an early warning based on these results.

[0022] Furthermore, in step S103 above, the identification and supplementation of outliers uses the moving average method. The step of identifying and supplementing outliers using the moving average method is as follows:

[0023] a) Calculate the moving average M;

[0024] b) Examine the actual value S and the moving average M, and determine whether it is an outlier according to formula (2);

[0025]

[0026] c) If an outlier is identified, delete it and replace it with a moving average M.

[0027] Furthermore, in step S104 above, the step of performing data analysis to obtain the water level monitoring point J... ifuture time T j+1 ~T j+x predicted value J i(j+1) ~J i(j+x) The steps are:

[0028] a) preprocessing the monitoring data set C i ={J i |j = 1 ~ m} of the monitoring point J ij , the preprocessing includes data normalization, and the preprocessing includes dividing the monitoring data set into a training set and a test set;

[0029] b) constructing a recurrent neural network prediction model, the recurrent neural network prediction model adopts a hyperbolic tangent scaling linear function, and the expression of the hyperbolic tangent scaling linear function is formula (3),

[0030]

[0031] In the formula, a is a parameter, and the value range is 0.01~1;

[0032] In order to better fit the periodic characteristics of individual monitoring data changes, the recurrent neural network is selected to better capture the time dependence in the data, and the hyperbolic tangent scaling linear function is selected due to its typical nonlinear characteristics and smoothness and scale adjustment characteristics. The typical nonlinear characteristics can better capture the periodically changing data, and the parameter a provides the scale adjustment characteristics of the hyperbolic tangent scaling linear function, which can be obtained by cross-validation and other methods in actual use.

[0033] c) training the recurrent neural network prediction model using the training set data, testing and optimizing the recurrent neural network prediction model using the test set data, and finally obtaining an applicable model;

[0034] d) using the applicable model to obtain the predicted value J j+1 ~J j+x of the future time T i(j+1) ~T i(j+x) .

[0035] Further, in the above step S105, the step of constructing a monitoring data analysis model of the future time T y and obtaining the water level monitoring prediction value J y of the deep foundation pit site of the future time T (n+1)y is:

[0036] a) for the future time T y , the monitoring data set C y ={J iy |i = 1 ~ n} of all monitoring points at this time is obtained;

[0037] b) constructing an input-output data set, the input data being the monitoring point J i distance L from the deep foundation pit i , the corresponding output data being J iy ;

[0038] c) constructing a BP neural network model M y , the model M y adopting an enhanced Gaussian activation function, the expression of the enhanced Gaussian activation function being formula (4);

[0039]

[0040] In the formula, a is an enhanced response coefficient, and a≥1.01 at the same time;

[0041] In order to better fit the data distribution characteristics (complex nonlinearity) of the overall monitoring point data at one time, the enhanced Gaussian activation function is selected, which can better fit the change characteristics of the data, and the enhanced response coefficient a is introduced at the same time, and a≥1.01 is required, which increases the nonlinearity of the function and makes it have better performance in complex nonlinearity problems;

[0042] d) placing the output data set into the BP neural network model M y for model training and optimization, and obtaining a final optimized model;

[0043] e) inputting L i =0 into the final optimized model to obtain a water level monitoring prediction value J y of the deep foundation pit at a future time T (n+1)y .

[0044] Further, in the above step S106, the warning standard includes a time threshold T 阈值 and a monitoring data threshold J 阈值 , and the warning analysis to obtain a warning result is to perform warning analysis and obtain a warning result by using formula (2),

[0045]

[0046] In the formula, T 阈值 is a time threshold, and J 阈值 is a monitoring data threshold;

[0047] Here, the actual construction is mainly considered. On the one hand, the influence of the water level needs to be considered, and an alarm needs to be given when the set threshold is exceeded. However, in actual problems, some monitoring points are far away, and a long period of time is needed to reach the vicinity of the deep foundation pit. Therefore, it is more reasonable to determine the warning standard by comprehensively considering the influence of time and water level monitoring value.

[0048] A ship lock main body deep foundation construction water level monitoring system, specifically includes water level monitoring module, water level monitoring data transmission and receiving module, water level monitoring data processing module and water level monitoring early warning module, the water level monitoring module is used for monitoring water level and obtaining water level monitoring data, the water level monitoring data transmission and receiving module is used for receiving water level monitoring data and transmitting to water level monitoring data processing module, the water level monitoring data processing module is used for processing the obtained water level monitoring data and obtaining early warning result, and the early warning result is transmitted to the water level monitoring early warning module, and the water level monitoring early warning module makes corresponding operation according to the received early warning result.

[0049] The present application has the following beneficial effects: 1) for the problem of irregular data distortion of the water level monitoring data around the ship lock deep foundation, the method is arranged upstream of the monitoring point, the upstream monitoring point data is analyzed and studied, and then the predicted value of the water level around the deep foundation is obtained; 2) for the obvious characteristics of the data of the upstream water level monitoring point, the activation function more consistent with the data change rule is selected in the data processing, and the fitting generalization ability of the model is improved; 3) for the obtained prediction data, and combined with the present situation of the ship lock main body deep foundation construction, a more practical early warning method is developed. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 The flowchart of the ship lock main body deep foundation construction water level monitoring method of the present application is shown in the figure. DETAILED DESCRIPTION

[0051] 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 illustration and explanation of the present application, and cannot be used to limit the present application.

[0052] The following is a specific embodiment of a ship lock main body deep foundation construction water level monitoring system and method.

[0053] A ship lock main body deep foundation construction water level monitoring method, characterized in that the ship lock main body deep foundation construction water level monitoring method comprises the following steps:

[0054] S101 arranging water level monitoring points;

[0055] The arrangement of water level monitoring points comprises arranging water level monitoring points in the upstream of the water level of the ship lock main body deep foundation according to a certain interval distance L1 from far to near, a total of n, marked as J i , i=1~n, and J i The distance L i from the deep foundation is L i , which is obtained by formula (1).

[0056] L i = L1×(n+1-j) (1)

[0057] S102 obtaining water level monitoring data;

[0058] The obtaining of the water level monitoring data comprises obtaining monitoring data at fixed time intervals for each of the arranged water level monitoring points J i , where m monitoring data have been obtained, the time of obtaining the monitoring data is T j , j = 1 ~ m, and the time T j corresponds to the water level monitoring point J i The obtained monitoring data is represented as J ij , and the obtained monitoring data set for the water level monitoring point J i is C i = {J ij |j = 1 ~ m};

[0059] S103 preliminary processing of the monitoring data;

[0060] The preliminary processing of the monitoring data comprises preliminary processing of the obtained data set C i , and the data preliminary processing comprises identification and supplement of missing values and identification and supplement of abnormal values.

[0061] S104 data analysis of a single water level monitoring point;

[0062] The data analysis of the single water level monitoring point comprises obtaining the monitoring data set C i = {J i |j = 1 ~ m} for the water level monitoring point J ij , performing data analysis to obtain the predicted value J i ~ J j+1 of the water level monitoring point J j+x at the future time T i(j+1) ~ T i(j+x) .

[0063] S105 data analysis of overall water level distribution;

[0064] The data analysis of the overall water level distribution comprises obtaining the monitoring data set C y = {J y |i = 1 ~ n} of all monitoring points at the future time T iy , y = (j+1) ~ (j+x), constructing an analysis model of the monitoring data at the future time T y and obtaining the water level monitoring prediction value J y of the deep foundation site at the future time T (n+1)y .

[0065] The implementation of S106 water level early warning;

[0066] The implementation of the water level early warning includes combining the obtained future time T y Predicted water level J at deep foundation pit (n+1)y The system analyzes the early warning criteria to obtain early warning results, and then decides whether to issue an early warning based on these results.

[0067] Furthermore, in step S103 above, the identification and supplementation of outliers uses the moving average method. The step of identifying and supplementing outliers using the moving average method is as follows:

[0068] a) Calculate the moving average M;

[0069] b) Examine the actual value S and the moving average M, and determine whether it is an outlier according to formula (2);

[0070]

[0071] c) If an outlier is identified, delete it and replace it with a moving average M.

[0072] Furthermore, in step S104 above, the data analysis is performed to obtain the monitoring point J. i Future Time T j+1 ~T j+x The predicted value J i(j+1) ~J i(j+x) The steps are as follows:

[0073] a) Monitoring point J i Monitoring data set C i ={J ij The data set is preprocessed according to the range |j=1~m}, including data normalization and dividing the monitoring data set into training set and test set;

[0074] b) Construct a recurrent neural network prediction model, wherein the recurrent neural network prediction model adopts a hyperbolic tangent scaling linear function, and the expression of the hyperbolic tangent scaling linear function is Equation (3).

[0075]

[0076] In the formula, a is a parameter, and its value ranges from 0.01 to 1;

[0077] In order to better fit the periodic characteristics of individual monitoring data changes, the recurrent neural network is selected to better capture the time-dependent relationship in the data, and the hyperbolic tangent scaling linear function is selected due to its typical nonlinear characteristics and smoothness and scale adjustment characteristics. The typical nonlinear characteristics can better capture the periodically changing data, and the parameter a provides the scale adjustment characteristics of the hyperbolic tangent scaling linear function, which can be obtained through cross-validation and other methods in actual use.

[0078] c) using the training set data to train the recurrent neural network prediction model, using the test set data to test and optimize the recurrent neural network prediction model, and finally obtaining an applicable model;

[0079] d) using the applicable model to obtain the prediction value J j+1 ~J j+x of the future time T i(j+1) ~T i(j+x) .

[0080] Further, in the step S105, the step of constructing an analysis model of the monitoring data of the future time T y and obtaining the water level monitoring prediction value J y of the deep foundation pit site at the future time T (n+1)y is:

[0081] a) for the future time T y , obtain the monitoring data set of all monitoring points at this time as C y ={J iy |i=1~n};

[0082] b) construct an input-output data group, the input data is the distance L i of the monitoring point J i from the deep foundation pit, and the corresponding output data is J iy ;

[0083] c) construct a BP neural network model M y , the model M y uses an enhanced Gaussian activation function, and the expression of the enhanced Gaussian activation function is formula (4);

[0084]

[0085] In the formula, α is an enhanced response coefficient, and α≥1.01;

[0086] In order to better fit the data distribution characteristics of the whole monitoring point data at a time (complex nonlinearity), the enhanced Gaussian activation function is selected, which can better fit the change characteristics of the data, and the enhanced corresponding coefficient a is introduced, and a is required to be greater than or equal to 1.01, which increases the nonlinearity of the function and makes it have better performance in complex nonlinear problems;

[0087] d) placing the output data set into the BP neural network model M y to carry out model training and optimization, and obtaining a final optimized model;

[0088] e) inputting L i =0 into the final optimized model to obtain a water level monitoring prediction value J y of a deep foundation pit at a future time T (n+1)y .

[0089] Further, in the step S106, the warning criteria include a time threshold T 阈值 and a monitoring data threshold J 阈值 , and the warning analysis to obtain a warning result is to use formula (2) to perform warning analysis and obtain a warning result,

[0090]

[0091] In the formula, T 阈值 is the time threshold, and J 阈值 is the monitoring data threshold;

[0092] Here, the actual construction is mainly considered. On the one hand, the influence of water level is considered, and alarm is reported when the set threshold is exceeded. However, in actual problems, some monitoring points are far away, and a long time is needed to reach the vicinity of the deep foundation pit. Therefore, it is more reasonable to determine the warning criteria by comprehensively considering the influence of time and water level monitoring value.

[0093] A ship lock main body deep foundation pit construction water level monitoring system, specifically comprising a water level monitoring module, a water level monitoring data transmission and receiving module, a water level monitoring data processing module, and a water level monitoring warning module. The water level monitoring module is used for monitoring the water level and obtaining water level monitoring data. The water level monitoring data transmission and receiving module is used for receiving water level monitoring data and transmitting it to the water level monitoring data processing module. The water level monitoring data processing module is used for processing the obtained water level monitoring data and obtaining a warning result, and transmitting the warning result to the water level monitoring warning module. The water level monitoring warning module makes corresponding operation according to the received warning result.

[0094] In the above embodiment, a large water conservancy hub sluice deep foundation pit construction safety control system includes a water level monitoring module, a water level monitoring data transmission and receiving module, a water level monitoring data processing module, and a water level monitoring early warning module; a ship lock main body deep foundation pit construction water level monitoring method includes arranging water level monitoring points, obtaining water level monitoring data, preliminary processing of monitoring data, data analysis of a single water level monitoring point, data analysis of overall water level distribution, and realization of water level warning; the method can better predict the water level change of the ship lock main body deep foundation pit by analyzing and processing the monitoring data of the upstream water level of the ship lock main body deep foundation pit, thereby ensuring construction safety, and can be widely applied in the field of deep foundation pit construction.

[0095] The above is the preferred embodiment of the present application, which does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A ship lock main body deep foundation pit construction water level monitoring method, characterized in that, The monitoring method comprises the following steps: S101 arranging a water level monitoring point; S102 obtaining water level monitoring data; S103 preliminary processing of the monitoring data; S104 data analysis of a single water level monitoring point; S105 data analysis of overall water level distribution; S106 implementation of water level early warning; The arrangement of the water level monitoring points comprises arranging water level monitoring points in a certain interval distance L1 from far to near in the water level upstream of the main body deep foundation pit construction of the ship lock, and a total of n water level monitoring points are marked as J i , i=1~n, while J i The distance L i from the deep foundation pit is L i , which is obtained by formula (1). （1） Wherein, j is the serial number of the water level monitoring point; The water level monitoring data is obtained, including obtaining water level monitoring data of water level monitoring points J i arranged at fixed time intervals, assuming that m times of monitoring data have been obtained, the time of obtaining the monitoring data is T j , j=1~m, the time T j corresponds to the water level monitoring point J i The obtained monitoring data is represented as J ij The obtained monitoring data set of the water level monitoring point J i is C i ={J ij |j=1~m}. The preliminary processing of the monitoring data includes the identification and completion of missing values, the identification and completion of outliers, and the identification and completion of missing values in the obtained data set C i The preliminary processing of the data includes the identification and completion of missing values, the identification and completion of outliers, and the identification and completion of missing values in the obtained data set C The data analysis of the single water level monitoring point includes data analysis on water level monitoring points J i The obtained monitoring data set C i ={J ij |j=1~m} is analyzed to obtain the predicted value J i of the future time T j+1 ~T j+x of the water level monitoring point J i(j+1) ~J i(j+x) ; Data analysis of the overall water level distribution includes obtaining future time T y , the monitoring data set of all monitoring points of y=(j+1)~(j+x) is C y ={J iy |i=1~n}, constructing an analysis model of monitoring data of future time T y and obtaining the water level monitoring prediction value J y of the deep foundation pit site of future time T （n+1）y ; The water level warning is realized by combining the water level monitoring prediction value J （n+1）y with the warning standard to perform warning analysis and obtain a warning result, and whether to issue a warning is determined according to the warning result.

2. The water level monitoring method for the deep foundation pit construction of a ship lock body according to claim 1, characterized in that, In step S104, the water level monitoring point J i future time T j+1 ~T j+x predicted value J i(j+1) ~J i(j+x) the steps are: a) pre-processing the monitoring data set C i = {J ij |j = 1 ~ m} of the monitoring point J i , the pre-processing including data normalization, the pre-processing including dividing the monitoring data set into a training set and a test set; b) constructing a recurrent neural network prediction model, the recurrent neural network prediction model adopts a hyperbolic tangent scaling linear function, the expression of the hyperbolic tangent scaling linear function is formula (2), （2） In the formula, is a parameter, and has a value ranging from 0.01 to 1; c) training the recurrent neural network prediction model using training set data, testing and optimizing the recurrent neural network prediction model using test set data, and finally obtaining an applicable model; d) obtaining a predicted value J of the future time T using the applicable model j+1 ~T j+x i(j+1) ~J i(j+x)。 ​ 3. The water level monitoring method for the deep foundation pit construction of a ship lock body according to claim 1, characterized in that, In step S105, the step of constructing an analysis model of the monitoring data at future time Ty and obtaining a water level monitoring prediction value J(n+1)y of the deep foundation pit part at future time Ty is: a) obtaining a set of monitoring data C y for all monitoring points at the future time T y ={J iy |i=1~n} b) constructing an input-output data set, the input data being monitoring points J i distance L from the deep foundation pit i , the corresponding output data being J iy ; c) constructing a BP neural network model M y , the M y An enhanced Gaussian activation function is adopted, and the expression of the enhanced Gaussian activation function is formula (3). （3） In the formula, To enhance the response coefficient, while GELU(x) is a Gaussian error linear unit activation function. d) placing the output data set into the BP neural network model M y model training and optimization, and obtaining a final optimized model; e) inputting L i =0 into the final optimized model to obtain a future time T y a water level monitoring prediction value J （n+1）y .

4. The water level monitoring method for the deep foundation pit construction of a ship lock body according to claim 1, characterized in that, In step S106, the pre-warning criterion includes a time threshold T 阈值 and a monitoring data threshold J 阈值 The pre-warning analysis obtains a pre-warning result by using formula (4) to perform pre-warning analysis and obtain a pre-warning result. （4） In the formula, T 阈值 is a time threshold, J 阈值 is a monitoring data threshold, T m is the time of the mth monitoring data.

5. A ship lock main body deep foundation pit construction water level monitoring system for the ship lock main body deep foundation pit construction water level monitoring method according to any one of claims 1 to 4, characterized in that: Specifically comprises a water level monitoring module, a water level monitoring data transmission and receiving module, a water level monitoring data processing module and a water level monitoring early warning module, the water level monitoring module is used for monitoring water level and obtaining water level monitoring data, the water level monitoring data transmission and receiving module is used for receiving water level monitoring data and transmitting to the water level monitoring data processing module, the water level monitoring data processing module is used for processing the obtained water level monitoring data and obtaining early warning results, and the early warning results are transmitted to the water level monitoring early warning module, the water level monitoring early warning module makes corresponding operation according to the received early warning results.

Images