A foundation pit displacement prediction method, device, equipment, medium and product

Abstract

The present application discloses a method, device, equipment, medium and product for predicting foundation pit displacement, and relates to the technical field of foundation pit displacement monitoring and early warning. The method for predicting foundation pit displacement includes: using a variational modal decomposition method to decompose the foundation pit displacement data of the current time period, and obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period; respectively inputting the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period into the Elman model, and obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; fusing the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period, and obtain the foundation pit displacement data of the prediction time period. The present application has important application value in the foundation pit monitoring and early warning system, can effectively reduce the occurrence of foundation pit disaster accidents, and provides an efficient and accurate technical support for foundation pit construction and geological disaster prevention and control.

Description

Technical Field

[0001] The present application relates to the technical field of foundation pit displacement monitoring and early warning, and in particular to a method, device, equipment, medium and product for predicting foundation pit displacement based on neural network model fusion error and variational mode decomposition. Background Art

[0002] Whether it is the construction of ground buildings or the construction of subway stations, it is inseparable from the excavation and support of foundation pits. As the excavation depth of foundation pits increases, foundation pit disasters occur frequently. Therefore, it is of great significance to accurately predict the displacement of foundation pits.

[0003] The generation of foundation pit displacement is a complex process of coupling of multiple factors. The causes include: when the mechanical properties of the soil are uneven or weakened, the soil around the foundation pit will settle and deform; when the groundwater pressure increases or cannot be discharged, the stability of the soil is destroyed; when the foundation pit is excavated, the soil is affected by shear and horizontal extrusion forces, causing lateral displacement and sinking of the soil; strong earthquakes or explosion loads will cause huge dynamic effects on the surrounding soil, causing displacement and deformation of the soil. Summary of the invention

[0004] The purpose of this application is to provide a foundation pit displacement prediction method, device, equipment, medium and product, which can dynamically predict the changes in foundation pit displacement in real time.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides a method for predicting foundation pit displacement, comprising:

[0007] The foundation pit displacement is monitored to obtain the foundation pit displacement data of the current time period; the current time period includes the current moment and the first preset number of moments before the current moment.

[0008] The variational mode decomposition method is used to decompose the foundation pit displacement data of the current time period to obtain the trend term displacement, period term displacement, random term displacement and decomposition error of the current time period.

[0009] The trend item displacement, period item displacement, random item displacement and decomposition error of the current time period are respectively input into the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model.

[0010] The trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period are fused to obtain the foundation pit displacement data of the prediction time period.

[0011] Optionally, training the Elman model to obtain the trend term displacement prediction model, the period term displacement prediction model, the random term displacement prediction model and the decomposition error prediction model specifically includes the following steps:

[0012] The historical foundation pit displacement data are divided in different ways to obtain training samples corresponding to different division methods.

[0013] The variational mode decomposition method is used to decompose each training sample corresponding to different division methods, and the trend item displacement samples, period item displacement samples, random item displacement samples and decomposition error samples corresponding to different division methods are obtained.

[0014] The Elman model is trained using trend item displacement samples, period item displacement samples and random item displacement samples corresponding to different division methods, and the trained first Elman model, second Elman model and third Elman model corresponding to different division methods are obtained; among which, the first Elman model is used to predict the trend item displacement, the second Elman model is used to predict the period item displacement, and the third Elman model is used to predict the random item displacement.

[0015] Calculate the cumulative displacement error of the Elman model for predicting the trend term displacement, period term displacement and random term displacement that have been trained corresponding to the same division method.

[0016] The division method with the smallest cumulative displacement error is taken as the target division method.

[0017] The Elman model trained for predicting trend item displacement, period item displacement and random item displacement corresponding to the target division method is used as the trend item displacement prediction model, the period item displacement prediction model and the random item displacement prediction model.

[0018] The Elman model is trained using the decomposition error samples corresponding to the target division method to obtain the trained fourth Elman model as the decomposition error prediction model.

[0019] Optionally, the trend item displacement prediction model, the periodic item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model. Specifically, the trend item displacement prediction model, the periodic item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all constructed based on the Elman model, and the Elman model is optimized using the thinking evolution algorithm.

[0020] Optionally, the variational mode decomposition method is used to decompose the foundation pit displacement data of the current time period to obtain the trend term displacement, period term displacement, random term displacement and decomposition error of the current time period, specifically including:

[0021] By using the variational mode decomposition method, the number of decomposed modes is selected in increasing order. k The foundation pit displacement data of the current time period are decomposed several times respectively, and the center frequency corresponding to each decomposition of the foundation pit displacement data of the current time period is determined respectively; wherein the decomposition modal number k is a natural number greater than 1.

[0022] When the number of decomposition modes in the several decomposition processes is k +1, the center frequency and decomposition mode number corresponding to the decomposition processing are k If the difference of the center frequencies corresponding to the time decomposition processing is less than the preset threshold, it is determined k is the optimal decomposition mode number.

[0023] According to the optimal decomposition modal number k The foundation pit displacement data of the current time period is decomposed and processed to generate k The eigenmode function with similar changing trend to the foundation pit displacement is used as the trend term displacement, the eigenmode function with certain periodicity is used as the periodic term displacement, and the eigenmode function with higher frequency and no obvious changing rule is used as the random term displacement.

[0024] All intrinsic mode functions under the optimal decomposition mode number are used to reconstruct the foundation pit displacement data of the current time period and accumulate them to obtain the reconstructed foundation pit displacement of the current time period; the difference between the reconstructed foundation pit displacement of the current time period and the foundation pit displacement data of the current time period is used as the decomposition error.

[0025] In a second aspect, the present application provides a foundation pit displacement prediction device, comprising:

[0026] The data acquisition module is used to monitor the displacement of the foundation pit and obtain the displacement data of the foundation pit in the current time period; the current time period includes the current moment and the first preset number of moments before the current moment.

[0027] The data decomposition module is used to decompose the foundation pit displacement data of the current time period using the variational mode decomposition method to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period.

[0028] A model building module is used to input the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period into the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model respectively, to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model.

[0029] The data prediction module is used to fuse the trend item displacement, period item displacement, random item displacement and decomposition error in the prediction time period to obtain the foundation pit displacement data in the prediction time period.

[0030] In a third aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned foundation pit displacement prediction method.

[0031] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which implements the above-mentioned foundation pit displacement prediction method when executed by a processor.

[0032] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of foundation pit displacement prediction described in any one of the above.

[0033] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0034] The present application provides a method, device, equipment, medium and product for predicting foundation pit displacement, which predicts foundation pit displacement based on neural network model fusion error and variational modal decomposition. The variational modal decomposition method is used to decompose the foundation pit displacement data of the current time period to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period, which fully considers the complex influence of complex soil conditions and various external factors (such as weather changes, construction activities, groundwater level fluctuations, etc.) on the foundation pit displacement, so that the intrinsic characteristics of the displacement data can be analyzed more finely, providing a more accurate basis for subsequent predictions. The trend item displacement, period item displacement, random item displacement and decomposition error of the current time period are respectively input into the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period. The trend item displacement, periodic item displacement, random item displacement and decomposition error of the prediction time period are fused to obtain the foundation pit displacement data of the prediction time period, which realizes the accurate capture and prediction of the dynamic change process of the foundation pit displacement, improves the prediction accuracy of real-time dynamic prediction of foundation pit displacement, and can provide efficient and accurate technical support for foundation pit construction and geological disaster prevention and control. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 This is an application environment diagram of a foundation pit displacement prediction method in one embodiment of the present application;

[0037] Figure 2 This is a schematic diagram of the overall process of a foundation pit displacement prediction method in one embodiment of the present application;

[0038] Figure 3 A distribution diagram of displacement monitoring points provided in an embodiment of the present application;

[0039] Figure 4 It is a horizontal displacement diagram of the intrinsic mode function in one embodiment of the present application;

[0040] Figure 5 A vertical displacement diagram of an eigenmode function in an embodiment of the present application;

[0041] Figure 6 This is a diagram of displacement prediction results of horizontal trend items in different division methods in one embodiment of the present application;

[0042] Figure 7 This is a diagram of displacement prediction results of vertical trend items in different division methods in one embodiment of the present application;

[0043] Figure 8 This is a diagram showing the prediction results of horizontal periodic term displacements in different division methods in one embodiment of the present application;

[0044] Fig. 9 This is a diagram of vertical periodic displacement prediction results of different division methods in one embodiment of the present application;

[0045] Fig.10 This is a diagram of horizontal random item displacement prediction results in different division methods in an embodiment of the present application;

[0046] Fig.11 This is a diagram of vertical random item displacement prediction results of different division methods in one embodiment of the present application;

[0047] Fig.12 This is a diagram of horizontal cumulative displacement prediction results of different division methods in an embodiment of the present application;

[0048] Fig.13 This is a diagram of vertical cumulative displacement prediction results of different division methods in one embodiment of the present application;

[0049] Fig.14 This is a diagram of the horizontal displacement decomposition error prediction result of the target division method in one embodiment of the present application;

[0050] Fig.15 This is a diagram of the vertical displacement decomposition error prediction result of the target division method in one embodiment of the present application;

[0051] Fig.16 A comparison diagram of the measured horizontal displacement, the uncorrected horizontal displacement prediction result and the corrected horizontal displacement prediction result in one embodiment of the present application;

[0052] Fig.17 It is a comparison diagram of the measured vertical displacement, the uncorrected vertical displacement prediction result and the corrected vertical displacement prediction result in one embodiment of the present application;

[0053] Fig.18 A comparison diagram of the measured horizontal displacement, vertical displacement and the predicted horizontal displacement, vertical displacement of the test set in one embodiment of the present application;

[0054] Fig.19 A schematic diagram of functional modules of a foundation pit displacement prediction device provided in one embodiment of the present application;

[0055] Fig. 20 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0057] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0058] The relevant technologies for predicting foundation pit displacement mainly include numerical simulation methods, analytical methods and machine learning models. Among them, numerical simulation methods mainly include finite element method (FEM), boundary element method (BEM), finite difference method (FDM) and discrete element method (DEM), all of which can simulate complex structures and scenarios, including nonlinear and non-uniform materials, and complex boundary conditions. Analytical methods mainly include elastic foundation analytical method and elastoplastic foundation analytical method. The elastic foundation analytical method assumes that the soil and foundation pit are elastic bodies and uses analytical formulas to calculate the foundation pit displacement; the elastoplastic foundation analytical method takes into account the elastic-plastic properties of the soil and uses analytical or semi-analytical methods to analyze the foundation pit displacement. With the rise of artificial intelligence, machine learning models such as support vector machine (SVM), random forest (RF), genetic algorithm (GA), and artificial neural network (ANN) have been introduced into geotechnical engineering. They all establish foundation pit displacement prediction models by learning the input features and output values ​​of training samples.

[0059] Numerical simulation methods are widely used in foundation pit displacement prediction, but the reliability of their prediction results is highly dependent on the accuracy of model parameters. Analytical methods have limitations in considering the actual complex conditions of soil and foundation pits, and cannot consider complex soil conditions and various complex factors. Overfitting is a common problem in machine learning model training data. When the amount of data is insufficient, the model may over-learn the noise or subtle features of the training data, resulting in a decrease in the generalization performance of new data.

[0060] In view of this, the embodiments of the present application provide a method, device, equipment, medium and product for predicting foundation pit displacement. To address the problem that the generalization ability of a machine learning model decreases due to insufficient data, a variational mode decomposition method (VMD) is introduced to decompose the displacement time series data, and the Elman model is used to predict and analyze the foundation pit displacement. Finally, an accurate foundation pit displacement prediction result is obtained by fusing the predicted displacement and the prediction error.

[0061] The foundation pit displacement prediction method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the foundation pit displacement data of the current time period to the server 104, and the current time period includes the current moment and the first preset number of moments before the current moment. After receiving the foundation pit displacement data of the current time period, the server 104 uses the variational mode decomposition method to decompose the foundation pit displacement data of the current time period, and obtains the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period; the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period are respectively input into the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model; the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period are fused to obtain the foundation pit displacement data of the prediction time period. The server 104 can feed back the obtained foundation pit displacement data of the prediction time period to the terminal 102. In addition, in some embodiments, the foundation pit displacement prediction method can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly decompose the foundation pit displacement data of the current time period using the variational modal decomposition method, or the server 104 can obtain the foundation pit displacement data of the current time period from the data storage system and decompose the foundation pit displacement data of the current time period using the variational modal decomposition method.

[0062] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.

[0063] In an exemplary embodiment, Figure 2 As shown, a foundation pit displacement prediction method is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate, and specifically includes:

[0064] Step 1: monitor the displacement of the foundation pit and obtain the displacement data of the foundation pit in the current time period; the current time period includes the current moment and the first preset number of moments before the current moment.

[0065] The excavation depth of a foundation pit is about 16.35~23.35m. The soil layer structure is mainly composed of plain fill, strongly weathered granite, moderately weathered granite, slightly weathered granite, etc. It is a typical upper soil and lower rock and soil layer structure. A total of 27 slope top horizontal displacement monitoring points and 27 slope top vertical displacement monitoring points are arranged in the foundation pit, coded as BX1~BX27. The slope top horizontal displacement monitoring point and the slope top vertical displacement monitoring point are the same point. The monitoring point layout is as follows: Figure 3 shown.

[0066] The measured data of monitoring point BX19 was used as the research object. The monitoring time was from August 27, 2020 to February 10, 2022, with a total of 130 measurements. When the foundation pit was excavated in January 2021, a total of 50 measurements were measured, and the foundation pit was in the support stage from the 51st measurement. The first 100 measurements of the foundation pit monitoring data were used as the training set to train the model, the data of 101 to 120 measurements were used as the validation set, and the data of 121 to 130 measurements were used as the test set. The calculation process of the method of this application was implemented through Matlab programming.

[0067] Considering the timeliness of the data, if the number of tests is too small, there is a lack of representative data, and the model cannot generalize the new data well, resulting in a large prediction error; if the number of tests is too large, the timeliness of the historical data will be reduced, affecting the prediction results of the model. In order to determine the impact of different rolling prediction methods on the prediction results, four different division methods are constructed to train the model: predicting 1 test every 3 tests, predicting 1 test every 6 tests, predicting 1 test every 9 tests, and predicting 1 test every 12 tests, as shown in Table 1.

[0068] Table 1 Different rolling prediction methods

[0069]

[0070] Step 2: Use the variational mode decomposition method to decompose the foundation pit displacement data of the current time period to obtain the trend term displacement, periodic term displacement, random term displacement and decomposition error of the current time period.

[0071] The monitoring data has a lot of interference and noise. The variational mode decomposition method can reduce the impact of noise and interference on the prediction accuracy. The variational mode decomposition method uses an iterative search method to find the optimal solution of the variational model, thereby determining the center frequency and bandwidth of each decomposition component. Its core idea is to construct and solve the variational problem. The center frequency refers to the main frequency component of each modal component, which can be understood as the main oscillation frequency or main frequency of the signal. In signal processing, the center frequency helps to describe the periodicity or frequency characteristics of the signal. The bandwidth represents the width of the signal in the frequency domain, that is, the spectrum range of the signal. In VMD, the bandwidth of each component determines the range or width of its spectrum. The narrower the bandwidth, the more concentrated the signal is in the frequency domain, and the spectrum is mainly concentrated near the center frequency; the wider the bandwidth, the wider the spectrum distribution. The bandwidth is presented by the penalty factor in the code, and is set to 3950 according to the relevant literature. This method can not only effectively extract the modal components of the signal, but also optimize it under constraints to achieve accurate decomposition of the original signal. The formula is:

[0072] .

[0073] in, For the k An intrinsic mode function represents a time series component after the signal is decomposed; For the k The center frequency corresponding to the eigenmode function, in Hz; δ(t) is the Dirac delta function; is a complex term inversely proportional to time, where j is an imaginary unit used to represent complex numbers; π is the ratio of a circle to a circle; t is the time period; For time period t Partial derivative of, unit: mm / s; is a complex exponential function, which is used to convert the eigenmode function Convert to frequency domain; is the Euclidean norm or L2 norm, which is used to measure the length of a function; min is the minimization operator, which means to find the minimum value of the objective function. and ; Indicates the sum of all intrinsic mode functions; st is the abbreviation of "subject to", which means optimization is performed under certain constraints; f It is the foundation pit displacement data of the current time period, unit: mm.

[0074] By introducing the augmented Lagrangian function and repeatedly updating the functional until the corresponding iteration conditions are met, the constrained variation problem in the above formula is transformed into an unconstrained variation problem, solving the constrained optimization problem in the above formula.

[0075] .

[0076] in, is the loss function, λ is the constraint parameter, and α is the weight coefficient, which is used to balance the relative importance of different items; λ( t )for t Constraint parameters corresponding to the time period; f( t )for t The foundation pit displacement data corresponding to the time period.

[0077] By using the variational mode decomposition method, the number of decomposed modes is selected in increasing order. k The foundation pit displacement data of the current time period are decomposed several times respectively, and the center frequency corresponding to each decomposition process of the foundation pit displacement data of the current time period is determined respectively, as shown in Table 2 and Table 3.

[0078] When the number of decomposition modes in the several decomposition processes is k +1, the center frequency and decomposition mode number corresponding to the decomposition processing are k If the difference of the center frequencies corresponding to the time decomposition processing is less than the preset threshold, it is determined k is the optimal decomposition mode number. According to Table 2 and Table 3, we can know that: k =4, the maximum center frequency of the intrinsic mode function IMF4 is 23, and when k When ≥4, the maximum value of the center frequency tends to be stable, so the best decomposition mode k =4.

[0079] Table 2 Center frequencies of each IMF component of horizontal displacement

[0080]

[0081] Table 3 Center frequencies of each IMF component of vertical displacement

[0082]

[0083] Therefore, the number of decompositions of the intrinsic mode function (IMF) is: 4 horizontal displacements and 4 vertical displacements, respectively. Figure 4 and Figure 5 shown.

[0084] The center frequency of the IMF can be obtained by performing a Hilbert transform on the IMF signal. The Hilbert transform converts the time domain signal into an analytical signal, and then the center frequency of the IMF is obtained by solving the instantaneous frequency of the analytical signal.

[0085] The intrinsic mode functions IMF1 and IMF2 with low center frequencies are used as trend term displacements because they are similar to the trend of foundation pit displacement changes; the intrinsic mode function IMF3 with medium center frequencies and a certain periodicity is used as periodic term displacement; at the same time, the intrinsic mode function IMF4 with high center frequencies and no obvious change pattern is used as random term displacement. The foundation pit displacement data of the current time period is reconstructed using the intrinsic mode functions IMF1, IMF2, IMF3, and IMF4 and accumulated to obtain the reconstructed foundation pit displacement of the current time period; the difference between the reconstructed foundation pit displacement of the current time period and the foundation pit displacement data of the current time period is used as the decomposition error.

[0086] Step three: input the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period into the trend item displacement prediction model, period item displacement prediction model, random item displacement prediction model and decomposition error prediction model respectively to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model.

[0087] The construction of the Elman neural network model includes:

[0088] The nonlinear state space expression of the Elman neural network is:

[0089] .

[0090] In the formula, y( i ) indicates that the Elman neural network is i The output of each time period corresponds to the output foundation pit displacement data, unit: mm; g represents the activation function, which maps the input of the Elman neural network to the output, that is, the transfer function from the hidden layer to the output layer; w 3Represents the weight connecting the hidden layer and the output layer; x ( i ) indicates that the Elman neural network is i The internal state of each time period, that is, the displacement data of the foundation pit output by the hidden layer in the current time period, unit: mm; F represents the transfer function of the intermediate layer neurons, that is, the transfer function from the input layer to the hidden layer; w 1 is the weight connecting the hidden layer and the receiving layer; w 2 is the weight connecting the input layer and the hidden layer; Represents the feedback state vector, which is the foundation pit displacement output by the hidden layer in the previous time period, that is, x ( i -1), unit: mm; U ( i -1) indicates that the input layer is in i -1 time period input is the foundation pit displacement of the previous time period, unit: mm.

[0091] Furthermore, the trend term displacement prediction model, the period term displacement prediction model, the random term displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model, specifically by training the Elman model using a mind evolution algorithm.

[0092] The thinking evolution algorithm is used to optimize the weights and thresholds of the Elman neural network, including:

[0093] Constructing the objective function: The weights and thresholds of the Elman neural network are used as the variables to be optimized, and the objective function of the optimization problem is constructed. The objective function can be the error of the network on the training set, or other performance indicators, such as classification accuracy or regression accuracy.

[0094] Initialize the population: MEA (Mind Evolutionary Algorithm) randomly generates a certain number of individuals (solutions) at the beginning, each of which corresponds to the weight and threshold settings of an Elman neural network.

[0095] Fitness evaluation: Calculate the fitness of each individual (weights and threshold settings of the Elman network) under the objective function. Fitness evaluation can be achieved by training the Elman network and calculating its performance indicators.

[0096] Evolutionary operations: Use the operations of evolutionary algorithms (such as selection, crossover, and mutation) to update the population. In MEA, evolutionary operations may introduce new mind evolution mechanisms, such as mind communication or knowledge sharing between individuals to promote global search and local optimization.

[0097] Iterative optimization: Repeat fitness evaluation and evolution operations until a stopping condition is reached (such as reaching the maximum number of iterations or meeting the accuracy requirement).

[0098] Optimal solution extraction: According to the results of the optimization process, the individual with the best fitness is selected, that is, the optimal weight and threshold setting of the Elman network.

[0099] Through multiple debugging of the above steps, the parameters of the thinking evolution algorithm are determined as follows: population size 500, the number of winning sub-populations and temporary sub-populations are both 5, and the number of iterations is 10. The parameters of the Elman model are: the number of hidden nodes is 40, the total number of training rounds is 1000, and the learning rate is 0.001.

[0100] Substituting the trend item displacement sample into the Elman model, the horizontal trend item displacement prediction result is as follows: Figure 6 As shown in the figure, the vertical trend displacement prediction results are as follows: Figure 7 Table 4 shows the displacement evaluation index values ​​of horizontal and vertical trend items corresponding to the four different division methods. The absolute percentage error (MAPE), mean square error (MSE) and root mean square error (RMSE) of horizontal displacement model 3 are all minimized, and the correlation coefficient R 2 The maximum value is 0.95; similarly, the vertical displacement model 3 also reaches the minimum, and the correlation coefficient R 2 The maximum value is 0.96. Models 1 and 2 may lack training measurements, resulting in poor prediction accuracy. Model 4 uses too many measurements, which affects the timeliness of displacement data and leads to large prediction deviation. Therefore, it is believed that model 3 is more suitable for trend item displacement prediction.

[0101] Table 4 Evaluation index values ​​of horizontal and vertical trend item displacement corresponding to different division methods

[0102]

[0103] Substituting the periodic displacement samples into the Elman model, the horizontal periodic displacement prediction results are as follows: Figure 8 The vertical periodic displacement prediction results are shown in Fig. 9 As shown in Table 5, the horizontal and vertical periodic displacement evaluation index values ​​corresponding to the four different division methods are shown in Table 5. The absolute percentage error (MAPE), mean square error (MSE) and root mean square error (RMSE) of the horizontal displacement model 4 are all minimized, and the correlation coefficient R 2 The maximum value is 0.98; similarly, the vertical displacement model 2 reaches the minimum, and the correlation coefficient R 2The maximum value is 0.99. Overall, the horizontal and vertical evaluation indicators are close, and different prediction methods have little effect on the displacement prediction of the periodic term. It can be seen that since the displacement of the periodic term has a certain periodicity, the neural network model can better understand its change law, and the four division methods all show high prediction accuracy.

[0104] Table 5 Evaluation index values ​​of horizontal and vertical periodic displacement corresponding to different division methods

[0105]

[0106] Substitute the random displacement samples into the Elman model. The random displacement is affected by many factors. The change rules of the random displacement of horizontal and vertical displacements are not obvious. Since the proportion of the total displacement is small, only 0.1%, the impact on the prediction result is low. The prediction results of horizontal random displacement are as follows: Fig.10 As shown in the figure, the prediction results of vertical random displacement are as follows: Fig.11 As shown in Table 6, the horizontal and vertical random item displacement evaluation index values ​​corresponding to the four different division methods are shown.

[0107] Table 6 Evaluation index values ​​of horizontal and vertical random displacement corresponding to different division methods

[0108]

[0109] The prediction results and prediction accuracy of trend term displacement, periodic term displacement and random term displacement under different division methods are different, which leads to the difference in the prediction results of cumulative displacement of different models. The prediction results of the cumulative displacement of the four models are as follows: Fig.12 As shown in the figure, the prediction results of vertical cumulative displacement of the four models are as follows Fig.13 Table 7 shows the evaluation index values ​​of horizontal and vertical cumulative displacements corresponding to different models. The absolute percentage error (MAPE), mean square error (MSE) and root mean square error (RMSE) of the three horizontal displacements of the model are all the smallest, and the correlation coefficient R 2 The maximum is 0.92. Similarly, the absolute percentage error (MAPE), mean square error (MSE) and root mean square error (RMSE) of the vertical displacement of model three are also the smallest, and the correlation coefficient R 2 The maximum is 0.96.

[0110] Table 7 Comparison of prediction accuracy of horizontal and vertical cumulative displacements corresponding to different models

[0111]

[0112] In general, the prediction accuracy of model 3 after superposition of trend term displacement, period term displacement and random term displacement is higher than that of other models. Therefore, model 3 with the division method of predicting 1 measurement for every 9 measurements is selected as the prediction model for displacement and decomposition error.

[0113] The Elman model is trained using the decomposition error samples corresponding to the division method of predicting 1 measurement for every 9 measurements. The prediction results of the decomposition error of the horizontal displacement are as follows: Fig.14 As shown in the figure, the prediction results of the decomposition error of the vertical displacement are as follows Fig.15 As shown. Fig.14 and Fig.15 It can be seen that the decomposition error is predicted using the same method as the displacement prediction method, and the decomposition error prediction results of the horizontal displacement and vertical displacement are consistent with the actual decomposition error.

[0114] Step 4: Fusion the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period to obtain the foundation pit displacement data of the prediction time period.

[0115] The decomposed error prediction result is combined with the prediction result of the trend term displacement, period term displacement and random term displacement to obtain the final displacement prediction result. Fig.16 and Fig.17 It can be seen that there is a large error between the predicted results of horizontal displacement and vertical displacement without correction by fusion error and the measured horizontal displacement and vertical displacement results. The main reason is that the decomposition error caused by variational mode decomposition reconstruction accounts for a large proportion, of which horizontal displacement is 72% and vertical displacement is 67%; the prediction error accounts for a small proportion, of which horizontal displacement is 28% and vertical displacement is 33%. After correction by error fusion, part of the error caused by variational mode decomposition is effectively removed, and the predicted results of the corrected horizontal displacement and vertical displacement are in good agreement with the measured horizontal displacement and vertical displacement results.

[0116] According to the foundation pit displacement prediction method of this application, the results of the horizontal displacement and vertical displacement of 121 to 130 times are predicted by using the division method of predicting 1 time for every 9 times of measurement and the results of the measured horizontal displacement and vertical displacement are as follows: Fig.18 The predicted result is close to the actual displacement value, and the prediction accuracy is high, which further verifies the reliability of the foundation pit displacement prediction method of this application.

[0117] The present application also provides an application scenario, which applies the above-mentioned foundation pit displacement prediction method. Specifically: the foundation pit displacement prediction method provided in this embodiment can be applied to the foundation pit monitoring and safety assessment scenarios in civil engineering construction. The foundation pit displacement prediction method provided in this embodiment belongs to the risk prevention and monitoring link of the civil engineering construction management scenario. Specifically, after obtaining the predicted foundation pit displacement based on the above method, the development trend and potential risks of the foundation pit displacement can be further analyzed, and corresponding early warning measures and emergency plans can be formulated based on the prediction results. The main purpose of doing this is to take timely intervention measures to ensure the stability of the foundation pit, so as to prevent safety accidents caused by excessive foundation pit displacement, ensure the safety of personnel and equipment on the construction site, and at the same time reduce the economic losses and construction delays caused by foundation pit instability.

[0118] Based on the same inventive concept, the embodiment of the present application also provides a foundation pit displacement prediction device for implementing the foundation pit displacement prediction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more foundation pit displacement prediction device embodiments provided below can refer to the limitations of the foundation pit displacement prediction method above, and will not be repeated here.

[0119] In an exemplary embodiment, Fig.19 As shown, a foundation pit displacement prediction device is provided, comprising:

[0120] The data acquisition module 201 is used to monitor the displacement of the foundation pit and obtain the displacement data of the foundation pit in the current time period; the current time period includes the current moment and the first preset number of moments before the current moment.

[0121] The data decomposition module 202 is used to decompose the foundation pit displacement data of the current time period using a variational mode decomposition method to obtain the trend term displacement, period term displacement, random term displacement and decomposition error of the current time period.

[0122] The model building module 203 is used to input the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period into the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model respectively, to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model.

[0123] The data prediction module 204 is used to fuse the trend item displacement, period item displacement, random item displacement and decomposition error in the prediction time period to obtain the foundation pit displacement data in the prediction time period.

[0124] As an optional implementation, training the Elman model to obtain the trend term displacement prediction model, the period term displacement prediction model, the random term displacement prediction model and the decomposition error prediction model specifically includes the following steps:

[0125] The historical foundation pit displacement data are divided in different ways to obtain training samples corresponding to different division methods.

[0126] The variational mode decomposition method is used to decompose each training sample corresponding to different division methods, and the trend item displacement samples, periodic item displacement samples, random item displacement samples and decomposition error samples corresponding to different division methods are obtained.

[0127] The Elman model is trained using trend item displacement samples, period item displacement samples and random item displacement samples corresponding to different division methods, and the trained first Elman model, second Elman model and third Elman model corresponding to different division methods are obtained; among which, the first Elman model is used to predict the trend item displacement, the second Elman model is used to predict the period item displacement, and the third Elman model is used to predict the random item displacement.

[0128] Calculate the cumulative displacement error of the Elman model for predicting the trend term displacement, period term displacement and random term displacement that have been trained corresponding to the same division method.

[0129] The division method with the smallest cumulative displacement error is taken as the target division method.

[0130] The trained Elman model for predicting trend item displacement, period item displacement and random item displacement corresponding to the target division method is used as the trend item displacement prediction model, the period item displacement prediction model and the random item displacement prediction model.

[0131] The Elman model is trained using the decomposition error samples corresponding to the target division method to obtain the trained fourth Elman model as the decomposition error prediction model.

[0132] Optionally, the trend item displacement prediction model, the periodic item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model. Specifically, the trend item displacement prediction model, the periodic item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all constructed based on the Elman model, and the Elman model is optimized using the thinking evolution algorithm.

[0133] The variational mode decomposition method is used to decompose the foundation pit displacement data of the current time period to obtain the trend term displacement, period term displacement, random term displacement and decomposition error of the current time period, specifically including:

[0134] By using the variational mode decomposition method, the number of decomposed modes is selected in increasing order. k The foundation pit displacement data of the current time period are decomposed several times respectively, and the center frequency corresponding to each decomposition of the foundation pit displacement data of the current time period is determined respectively; wherein the decomposition modal number k is a natural number greater than 1.

[0135] When the number of decomposition modes in the several decomposition processes is k +1, the center frequency corresponding to the decomposition processing is k When the center frequency corresponding to the time decomposition processing decreases, it is determined k is the optimal decomposition mode number.

[0136] According to the optimal decomposition modal number k The foundation pit displacement data of the current time period is decomposed and processed to generate k The eigenmode function with similar changing trend to the foundation pit displacement is used as the trend term displacement, the eigenmode function with certain periodicity is used as the periodic term displacement, and the eigenmode function with higher frequency and no obvious changing rule is used as the random term displacement.

[0137] All intrinsic mode functions under the optimal decomposition mode number are used to reconstruct the foundation pit displacement data of the current time period and accumulate them to obtain the reconstructed foundation pit displacement of the current time period; the difference between the reconstructed foundation pit displacement of the current time period and the foundation pit displacement data of the current time period is used as the decomposition error.

[0138] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig. 20As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video tag processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a foundation pit displacement prediction method is implemented.

[0139] Those skilled in the art will understand that Fig. 20 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0140] In an exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps in the above-mentioned foundation pit displacement prediction method embodiment are implemented.

[0141] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the steps in the above-mentioned foundation pit displacement prediction method embodiment.

[0142] In an exemplary embodiment, a computer program product is provided, including a computer program, which implements the steps in the above-mentioned foundation pit displacement prediction method embodiment when executed by a processor.

[0143] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0144] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0145] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0146] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for predicting foundation pit displacement, characterized in that: The foundation pit displacement prediction method comprises: Monitor the displacement of the foundation pit to obtain the displacement data of the foundation pit in the current time period; the current time period includes the current moment and the first preset number of moments before the current moment; The variational mode decomposition method is used to decompose the foundation pit displacement data of the current time period to obtain the trend term displacement, period term displacement, random term displacement and decomposition error of the current time period, including: By using the variational modal decomposition method, a decomposition modal number k that increases in sequence is selected to perform several decomposition processes on the foundation pit displacement data of the current time period, and the central frequency corresponding to each decomposition process of the foundation pit displacement data of the current time period is determined respectively; wherein the decomposition modal number k is a natural number greater than 1; When the difference between the center frequency corresponding to the decomposition process when the decomposition mode number is k+1 and the center frequency corresponding to the decomposition process when the decomposition mode number is k in the several decomposition processes is less than a preset threshold, k is determined to be the optimal decomposition mode number; Decomposing the foundation pit displacement data of the current time period according to the optimal decomposition modal number k, generating k intrinsic modal functions; using an intrinsic modal function similar to the foundation pit displacement change trend as the trend term displacement, using an intrinsic modal function with a certain periodicity as the periodic term displacement, and using an intrinsic modal function with a high frequency and no obvious change pattern as the random term displacement; Reconstruct the foundation pit displacement data of the current time period using all the intrinsic mode functions under the optimal decomposition mode number and accumulate them to obtain the reconstructed foundation pit displacement of the current time period; the difference between the reconstructed foundation pit displacement of the current time period and the foundation pit displacement data of the current time period is taken as the decomposition error; The trend item displacement, period item displacement, random item displacement and decomposition error of the current time period are respectively input into the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model to obtain the trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model; The trend item displacement, period item displacement, random item displacement and decomposition error of the prediction time period are fused to obtain the foundation pit displacement data of the prediction time period.

2. The foundation pit displacement prediction method according to claim 1, characterized in that: Training the Elman model to obtain the trend term displacement prediction model, the period term displacement prediction model, the random term displacement prediction model and the decomposition error prediction model specifically includes the following steps: Divide the historical foundation pit displacement data in different ways to obtain training samples corresponding to different ways of division; The variational mode decomposition method is used to decompose each training sample corresponding to different division methods, and the trend item displacement samples, period item displacement samples, random item displacement samples and decomposition error samples corresponding to different division methods are obtained; The Elman model is trained using the trend item displacement samples, period item displacement samples and random item displacement samples corresponding to different division methods respectively, and the trained first Elman model, second Elman model and third Elman model corresponding to different division methods are obtained; wherein the first Elman model is used to predict the trend item displacement, the second Elman model is used to predict the period item displacement, and the third Elman model is used to predict the random item displacement; Calculate the cumulative displacement error of the Elman model for trend item displacement, period item displacement and random item displacement prediction corresponding to the same division method; The division method with the smallest cumulative displacement error is used as the target division method; The Elman model trained for trend item displacement, period item displacement and random item displacement prediction corresponding to the target division method is used as the trend item displacement prediction model, the period item displacement prediction model and the random item displacement prediction model; The Elman model is trained using the decomposition error samples corresponding to the target division method to obtain the trained fourth Elman model as the decomposition error prediction model.

3. The foundation pit displacement prediction method according to claim 1, characterized in that: The trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model. Specifically, the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all constructed based on the Elman model, and the thinking evolution algorithm is used to optimize the Elman model.

4. A foundation pit displacement prediction device, characterized in that: The foundation pit displacement prediction device applies the foundation pit displacement prediction method according to any one of claims 1 to 3, and the foundation pit displacement prediction device comprises: A data acquisition module is used to monitor the displacement of the foundation pit and obtain the displacement data of the foundation pit in the current time period; the current time period includes the current moment and the first preset number of moments before the current moment; The data decomposition module is used to decompose the foundation pit displacement data of the current time period by using the variational mode decomposition method to obtain the trend term displacement, period term displacement, random term displacement and decomposition error of the current time period; The model building module is used to input the trend item displacement, period item displacement, random item displacement and decomposition error of the current time period into the trend item displacement prediction model, period item displacement prediction model, random item displacement prediction model and decomposition error respectively. Prediction model, obtaining trend item displacement, period item displacement, random item displacement and decomposition error in a prediction time period; the prediction time period includes the second preset number of moments after the current moment; the trend item displacement prediction model, the period item displacement prediction model, the random item displacement prediction model and the decomposition error prediction model are all obtained by training the Elman model; The data prediction module is used to fuse the trend item displacement, period item displacement, random item displacement and decomposition error in the prediction time period to obtain the foundation pit displacement data in the prediction time period.

5. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the foundation pit displacement prediction method described in any one of claims 1 to 3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the foundation pit displacement prediction method described in any one of claims 1 to 3 is implemented.

7. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the foundation pit displacement prediction method described in any one of claims 1 to 3 is implemented.

Images