A foundation pit settlement prediction method, device, equipment and storage medium

Abstract

The application discloses a kind of foundation pit settlement prediction method, device, equipment and storage medium, the method comprises: the foundation pit settlement monitoring data is input to the preset foundation pit settlement inference model, to make foundation pit settlement inference model output the foundation pit settlement data of predicted measuring point;Wherein, the training of foundation pit settlement inference model includes: analyzing historical foundation pit settlement monitoring data, according to the analysis result, a preset neural network model is trained, and the foundation pit settlement inference model is obtained;Wherein, the foundation pit settlement inference model includes spatial information processing module and representation sequence fusion module;Space information processing module is used to generate corresponding representation sequence according to the time sequence of key data;Representation sequence fusion module is used to calculate the inference value of the foundation pit settlement data of predicted measuring point.By the present application, the foundation pit settlement monitoring data can be fully analyzed and utilized, and the long-term global inference of the foundation pit settlement of the measured measuring point is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underground engineering, and in particular to a foundation pit settlement prediction method, device, equipment and storage medium. BACKGROUND

[0002] At present, the vigorous underground rail transit construction makes more and more foundation pit projects appear in the city facility dense area, which is surrounded by various underground pipelines, buildings, traffic arteries and subway tunnels and the like. Due to the tight construction site, complex working conditions, high surrounding facility environmental protection requirements, severe subway construction safety situation, the subway construction attracts close attention of the public. How to accurately and efficiently analyze and predict based on the settlement monitoring data in the whole construction process, and ensure the safety and reliability of the surrounding facilities in the foundation pit excavation and construction process has become a problem that must be concerned in the rail transit construction.

[0003] Domestic and foreign scholars have carried out a lot of research work on the deformation law of the enclosure structure and the surrounding soil and structures caused by the foundation pit excavation in the complex environment. In recent years, the data-driven machine learning model is gradually applied to the prediction of the foundation pit settlement. The deep neural network among them has attracted the attention of researchers due to its strong fitting ability, and has gradually become the key research direction of the foundation pit deformation prediction. After the deep learning model is trained, it can realize the approximate prediction and reasoning of the foundation pit settlement in a very short time. By establishing various neural network models such as multilayer perceptron, long short-term memory network and convolutional neural network, the information such as construction conditions and procedures, historical monitoring data, soil survey results and underground water level is input into the model to realize the dynamic prediction of the foundation pit settlement and reduce the risk of major safety accidents such as foundation pit collapse.

[0004] However, the existing deep learning algorithm model architecture lacks theoretical support, and the data analysis and utilization are insufficient; most of them focus on time series prediction, and few of them mine the spatial correlation of the soil body. The global deduction of space is urgently needed to be realized. SUMMARY

[0005] The present application provides a foundation pit settlement prediction method, device, equipment and storage medium to solve the technical problem that the existing deep learning algorithm data analysis and utilization are insufficient and lack of mining of the spatial correlation of the soil body.

[0006] In order to solve the above technical problems, the present application embodiment provides a foundation pit settlement prediction method, comprising:

[0007] Obtain the foundation pit settlement monitoring data of the measurement point most similar to the foundation pit settlement mode of the to-be-predicted measurement point; wherein the foundation pit settlement monitoring data comprises the settlement conditions of the ground surface and the building within the preset range of the foundation pit where the measurement point is located;

[0008] The foundation pit settlement monitoring data is input into a preset foundation pit settlement inference model so that the foundation pit settlement inference model can infer the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data and output the foundation pit settlement data of the measurement point to be predicted.

[0009] The training of the foundation pit settlement inference model includes:

[0010] Acquire historical settlement monitoring data of the foundation pit at each measurement point;

[0011] Time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis are performed on the historical foundation pit settlement monitoring data of each measurement point. Then, based on the analysis results, the measurement point with the most similar foundation pit settlement pattern to the measurement point to be predicted is determined as the key measurement point. From the historical foundation pit settlement monitoring data of the key measurement points, historical foundation pit settlement monitoring data that have a causal relationship with the historical foundation pit settlement monitoring data of the measurement point to be predicted are selected as key data.

[0012] Using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output, a preset neural network model is trained to obtain the foundation pit settlement inference model.

[0013] The foundation pit settlement inference model includes: a spatial information processing module and a representation sequence fusion module;

[0014] The spatial information processing module is used to generate a corresponding representation sequence based on the time series of the key data;

[0015] The representation sequence fusion module is used to fuse the representation sequences and calculate the inference value of the foundation pit settlement data of the measurement point to be predicted based on the fused representation sequence.

[0016] As a preferred embodiment, the step of performing time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point includes:

[0017] Based on the autocorrelation function and partial autocorrelation function, time autocorrelation analysis was performed on the historical foundation pit settlement monitoring data of each measurement point to obtain the autocorrelation between the historical foundation pit settlement monitoring data of each measurement point.

[0018] Based on dynamic time warping and clustering, spatial similarity analysis was performed on the historical foundation pit settlement monitoring data of each measurement point to obtain the similarity of foundation pit settlement patterns between different measurement points.

[0019] Based on the Granger causality test method, a spatiotemporal causality analysis was conducted on the historical foundation pit settlement monitoring data of each measurement point to obtain the causal relationships between variables in the historical foundation pit settlement monitoring data of each measurement point.

[0020] Based on the Spearman correlation coefficient matrix of the same measurement point at different times, a spatiotemporal consistency analysis was conducted on the historical foundation pit settlement monitoring data of each measurement point to determine whether the settlement pattern of the same measurement point at different times is consistent.

[0021] As a preferred approach, the corresponding representation sequence is generated according to the following formula:

[0022] r t =relu(W r ·s t +b r );

[0023] Where, r t W is the representation sequence corresponding to the time series of key data. r Let b be the weight matrix. r Let be the bias matrix, ReLU be the nonlinear activation function, and s be the bias matrix. t The value represents the settlement of the foundation pit measured at time t in history.

[0024] As a preferred approach, the inferred value of the foundation pit settlement data at the measurement point to be predicted is calculated using the following formula:

[0025] e = W e ·[r1,…,r n ]+b e ;

[0026] Where e is the inferred value of the foundation pit settlement data at the measurement point to be predicted, and W e b is the weight matrix; e Here is the bias matrix; [r1,…,r n This indicates that all representations are merged and connected.

[0027] As a preferred embodiment, the step of using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output to train a preset neural network model to obtain the foundation pit settlement inference model includes:

[0028] Get the preset hyperparameters;

[0029] Based on the hyperparameters, the key data is used as input, and the historical foundation pit settlement data corresponding to the measurement point to be predicted is used as output. A preset neural network model is trained, and the hyperparameters are optimized according to the asynchronous continuous halving algorithm and the grid search method during the training process. Finally, the foundation pit settlement inference model is obtained.

[0030] As a preferred embodiment, after the foundation pit settlement inference model infers the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data and outputs the foundation pit settlement data of the measurement point to be predicted, it further includes:

[0031] Based on the preset interpretable machine learning model, the Shapley additive interpretation analysis is performed on the foundation pit settlement data of the measurement points to be predicted output by the foundation pit settlement inference model, and the impact of changes in each input feature of the foundation pit settlement inference model on the foundation pit settlement data of the measurement points to be predicted is calculated.

[0032] Based on the impact of changes in each input feature on the foundation pit settlement data of the measurement points to be predicted, the importance of each input feature is statistically determined.

[0033] As a preferred approach, the Shapley additive interpretation analysis is performed on the foundation pit settlement data of the measurement points to be predicted, output by the foundation pit settlement inference model, according to the following method:

[0034]

[0035] Where g is the interpretable machine learning model, x′ is the simplified data input that can be mapped to the original input data points, m is the number of input features, and φ j Let φ0 be the attribute value of each input feature, and φ1 be the mean of all predicted values ​​for the samples, where {x1, ..., x0} is the value assigned to each input feature. p} represents the set of all input features, {x1,…,x...} p}\{x j} represents division by x j The possible set of all input features other than f x (S) is the prediction of the feature subset S.

[0036] Based on the above embodiments, another embodiment of the present invention provides a foundation pit settlement prediction device, including: a data acquisition module, a data prediction module, and a model training module;

[0037] The data acquisition module is used to acquire foundation pit settlement monitoring data of the measurement point that is most similar to the foundation pit settlement pattern of the measurement point to be predicted; wherein, the foundation pit settlement monitoring data includes: the settlement status of the ground surface and buildings within a preset range of the foundation pit where the measurement point is located;

[0038] The data prediction module is used to input the foundation pit settlement monitoring data into a preset foundation pit settlement inference model, so that the foundation pit settlement inference model can infer the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data, and output the foundation pit settlement data of the measurement point to be predicted.

[0039] The model training module is used to acquire historical foundation pit settlement monitoring data of each measurement point; perform time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point; then, based on the analysis results, determine the measurement point most similar to the foundation pit settlement pattern of the measurement point to be predicted as the key measurement point; and select historical foundation pit settlement monitoring data with a causal relationship with the historical foundation pit settlement monitoring data of the measurement point to be predicted from the historical foundation pit settlement monitoring data of the key measurement point as key data; using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output, train a preset neural network model to obtain the foundation pit settlement inference model; wherein, the foundation pit settlement inference model includes: a spatial information processing module and a representation sequence fusion module; the spatial information processing module is used to generate a corresponding representation sequence based on the time series of the key data; the representation sequence fusion module is used to fuse the representation sequence and calculate the inferred value of the foundation pit settlement data of the measurement point to be predicted based on the fused representation sequence.

[0040] Based on the above embodiments, another embodiment of the present invention provides an electronic device, the device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the foundation pit settlement prediction method described in the above embodiments of the invention.

[0041] Based on the above embodiments, another embodiment of the present invention provides a storage medium, the storage medium including a stored computer program, wherein, when the computer program is running, it controls the device where the storage medium is located to execute the foundation pit settlement prediction method described in the above embodiments of the invention.

[0042] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0043] This invention provides a method for predicting foundation pit settlement. The method involves acquiring foundation pit settlement monitoring data from the measurement point whose settlement pattern is most similar to that of the measurement point to be predicted; inputting the foundation pit settlement monitoring data into a preset foundation pit settlement inference model, so that the foundation pit settlement inference model infers the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data, and outputs the foundation pit settlement data of the measurement point to be predicted.

[0044] The training of the foundation pit settlement inference model includes: acquiring historical foundation pit settlement monitoring data for each measurement point; performing time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on the historical foundation pit settlement monitoring data for each measurement point; then determining the measurement point most similar to the foundation pit settlement pattern of the measurement point to be predicted based on the analysis results, as the key measurement point; selecting historical foundation pit settlement monitoring data with a causal relationship to the historical foundation pit settlement monitoring data of the measurement point to be predicted from the historical foundation pit settlement monitoring data of the key measurement point, as the key data; using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output, training a preset neural network model to obtain the foundation pit settlement inference model; wherein, the foundation pit settlement inference model includes: a spatial information processing module and a representation sequence fusion module; the spatial information processing module is used to generate a corresponding representation sequence based on the time series of the key data; the representation sequence fusion module is used to fuse the representation sequence and calculate the inference value of the foundation pit settlement data of the measurement point to be predicted based on the fused representation sequence.

[0045] This invention, when training the foundation pit settlement inference model, performs time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on historical foundation pit settlement monitoring data at each measurement point. This allows for comprehensive analysis and utilization of the foundation pit settlement monitoring data, fully exploring its inherent mechanisms and patterns. This identifies the measurement point most similar to the settlement pattern of the measurement point to be predicted, and historical foundation pit settlement monitoring data with causal relationships to the historical settlement monitoring data of the measurement point to be predicted. Furthermore, the foundation pit settlement inference model includes a spatial information processing module and a representation sequence fusion module. It generates corresponding representation sequences based on the time series of key data, enabling comprehensive extraction of spatiotemporal information about the foundation pit. These representation sequences are then fused, and the inferred value of the foundation pit settlement data for the measurement point to be predicted is calculated based on the fused representation sequences. This allows for long-term global inference of foundation pit settlement at measurement points with unknown settlement based on the monitoring data of the measurement points. Attached Figure Description

[0046] Figure 1 This is a flowchart illustrating a method for predicting foundation pit settlement according to an embodiment of the present invention.

[0047] Figure 2 This is a flowchart of the model training process for the foundation pit settlement inference model of the present invention;

[0048] Figure 3 This is a schematic diagram of the model architecture for the foundation pit settlement inference model;

[0049] Figure 4This is a schematic diagram of the hyperparameter tuning method for the foundation pit settlement inference model;

[0050] Figure 5 This is a schematic diagram of the structure of a foundation pit settlement prediction device provided in an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0053] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0054] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0055] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0056] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0057] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0058] Example 1

[0059] Please refer to Figure 1 The following is a flowchart illustrating a method for predicting foundation pit settlement according to an embodiment of the present invention, including the following specific steps:

[0060] S1. Obtain the foundation pit settlement monitoring data of the measurement point that is most similar to the foundation pit settlement pattern of the measurement point to be predicted; wherein, the foundation pit settlement monitoring data includes: the settlement status of the ground surface and buildings within the preset range of the foundation pit where the measurement point is located.

[0061] First, obtain the foundation pit settlement monitoring data of the nearest measurement point that has the most similar foundation pit settlement pattern to the measurement point to be measured; wherein, the foundation pit settlement monitoring data includes: the settlement status of the ground surface and buildings within a preset range of the foundation pit where the measurement point is located.

[0062] S2. Input the foundation pit settlement monitoring data into the preset foundation pit settlement inference model, so that the foundation pit settlement inference model can infer the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data, and output the foundation pit settlement data of the measurement point to be predicted.

[0063] The acquired foundation pit settlement monitoring data is then input into a pre-trained foundation pit settlement inference model. This model infers the foundation pit settlement data at the monitoring points based on the monitoring data and outputs the current foundation pit settlement data for each monitoring point. Specifically, for m monitoring points... Historical monitoring data for n days On day n+1, measuring point P of number k k If data loss or anomalies occur, the latest few days' (including day n+1) of foundation pit settlement monitoring data from other measurement points are input into the foundation pit settlement inference model. The output is the measurement point P obtained from the model inference. k Settlement data of the foundation pit on day n+1.

[0064] The training of the foundation pit settlement inference model includes:

[0065] Acquire historical settlement monitoring data of the foundation pit at each measurement point;

[0066] Time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis are performed on the historical foundation pit settlement monitoring data of each measurement point. Then, based on the analysis results, the measurement point with the most similar foundation pit settlement pattern to the measurement point to be predicted is determined as the key measurement point. From the historical foundation pit settlement monitoring data of the key measurement points, historical foundation pit settlement monitoring data that have a causal relationship with the historical foundation pit settlement monitoring data of the measurement point to be predicted are selected as key data.

[0067] Using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output, a preset neural network model is trained to obtain the foundation pit settlement inference model.

[0068] The foundation pit settlement inference model includes: a spatial information processing module and a representation sequence fusion module;

[0069] The spatial information processing module is used to generate a corresponding representation sequence based on the time series of the key data;

[0070] The representation sequence fusion module is used to fuse the representation sequences and calculate the inference value of the foundation pit settlement data of the measurement point to be predicted based on the fused representation sequence.

[0071] Specifically, in existing foundation pit settlement inference methods, the deep learning algorithm models used for foundation pit settlement inference lack theoretical support, data analysis and utilization are insufficient, and model interpretability is inadequate. Most can only achieve short-term inference and have high requirements for monitoring data; errors are significant when data is missing or incorrect, but data anomalies are very common in actual engineering. Most focus on temporal inference, with little exploration of the spatial correlation of soil; global spatial extrapolation urgently needs to be implemented. Based on this, this invention mainly addresses the following three problems:

[0072] (1) Establish a complete data analysis process for the spatiotemporal correlation of foundation pit settlement, which can fully explore and extract the settlement mechanism, pattern and law within the monitoring data;

[0073] (2) Establish a deep learning model for settlement inference using data from nearby measuring points to complete and detect abnormal monitoring data such as missing or erroneous data;

[0074] (3) Use advanced interpretable machine learning models to interpret the output results of deep learning models.

[0075] To address the three problems mentioned above, this invention proposes a method for predicting foundation pit settlement, which involves a series of improvements to the deep learning algorithm. The specific improvements are as follows:

[0076] (1) By conducting time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on historical foundation pit settlement monitoring data, we can fully explore the inherent mechanism and laws of foundation pit settlement data, assist in the architecture design of deep learning models, and provide a reference for the construction methods of similar foundation pit projects.

[0077] (2) The constructed deep learning model includes a spatial information processing module and a representation sequence fusion module, which can realize the comprehensive extraction of spatiotemporal information of the foundation pit. Based on the historical monitoring data of the adjacent measuring points, it can realize long-term inference of the settlement of unknown measuring points, which is fast and accurate.

[0078] (3) Unlike general deep learning methods, the algorithm proposed in this invention can also interpret the reasoning results through an interpretable machine learning model, thereby enhancing the reliability of the reasoning results and making it applicable to large-scale practical engineering.

[0079] Please refer to Figure 2 The following is a flowchart of the model training process for the foundation pit settlement inference model of the present invention. The following details the model training process of the aforementioned improvements, namely the foundation pit settlement inference model:

[0080] Preferably, the step of performing time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point includes: performing time autocorrelation analysis on the historical foundation pit settlement monitoring data of each measurement point based on autocorrelation function and partial autocorrelation function to obtain the autocorrelation between the historical foundation pit settlement monitoring data of each measurement point; performing spatial similarity analysis on the historical foundation pit settlement monitoring data of each measurement point based on dynamic time warping and clustering to obtain the similarity of foundation pit settlement patterns between different measurement points; performing spatiotemporal causality analysis on the historical foundation pit settlement monitoring data of each measurement point based on Granger causality test to obtain the causal relationship between variables in the historical foundation pit settlement monitoring data of each measurement point; and performing spatiotemporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point based on the Spearman correlation coefficient matrix of the same measurement point at different periods to determine whether the settlement pattern of the same measurement point at different periods is consistent.

[0081] (1) First step: Data analysis of foundation pit settlement monitoring data

[0082] Data analysis mainly includes four parts: time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis.

[0083] The time autocorrelation analysis section studies the autocorrelation of settlement data at each measuring point to clarify the impact of historical settlement on future development, verify the feasibility of time series forecasting, and determine the input window length for historical data. The correlation coefficient between a series and its lagged series is calculated using the Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) to evaluate the autocorrelation of the series; a higher correlation coefficient indicates stronger autocorrelation.

[0084] ACF(k) = corr(x) t ,x t-k (1)

[0085]

[0086] Where, x t and x t-k These represent the original time series and the translated time series, respectively. and x represents the value obtained from minimum mean square error (MSE) regression. t and x t-k The predicted value is denoted by corr, which represents the correlation coefficient.

[0087] ACF quantifies the linear relationship between a time series and its lagged series, while PACF reveals the relationship between a time series and a series with a specific lag order.

[0088] Spatial similarity analysis: A clustering algorithm is introduced to evaluate the similarity of settlement patterns among different measuring points, showing that measuring points with similar settlement patterns can contribute more to settlement inference. As the excavation pit is excavated from one side to the other, the surface settlement at different measuring points exhibits a non-linear lag relationship. Therefore, Dynamic Time Warping (DTW) and clustering are used to compare the similarity between two sequences. DTW minimizes the total distance by dynamically adjusting the corresponding points between two sequences along the time axis, and then clusters the sequences at different measuring points based on the distance measured by DTW. The working principle of DTW is shown in Table 1 below:

[0089]

[0090]

[0091] Table 1 DTW Working Principle

[0092] Spatiotemporal causality analysis: Historical monitoring data from other measurement points may be valuable in inference, improving the accuracy and robustness of predictions. The Granger causality test is used to analyze causal relationships in time series data, assessing whether one variable has a causal effect on another based on the predictive performance of the time series. Granger defines causality based on two principles: (1) cause precedes effect; (2) cause contains the only information about the future value of its effect. Given these two assumptions about causality, Granger proposes testing the following hypotheses to determine the causal effect of x on y:

[0093]

[0094] in, Representing probability, A is an arbitrary non-empty set. and Let represent the available information for time t in the universal set and the available information in the modified set excluding x, respectively. If the above assumption is accepted, then x has Granger causality with respect to y.

[0095] Spatiotemporal consistency analysis: Ensuring the consistency of settlement patterns among a set of measurement points between two different periods is crucial for validating the effectiveness of a model trained on earlier data in later stages. The consistency of settlement mechanisms between the two stages can be assessed by comparing the Spearman correlation coefficient matrices of the two periods. The Spearman correlation coefficient ranges from -1 (strong negative correlation) to 1 (strong positive correlation), with values ​​close to 0 indicating no correlation between variables x and y. For a sample of size n, the n original data points x are sorted... i ,y i Convert to rank(x) i ),rank(y i The Spearman correlation coefficient is calculated as follows:

[0096] r s =corr(rank(x),rank(y)) (4)

[0097] Preferably, the corresponding representation sequence is generated according to the following formula:

[0098] r t =relu(W r ·s t +b r );

[0099] Where, r t W is the representation sequence corresponding to the time series of key data. r Let b be the weight matrix.r Let be the bias matrix, ReLU be the nonlinear activation function, and s be the bias matrix. t The value represents the settlement of the foundation pit measured at time t in history.

[0100] Preferably, the inferred value of the foundation pit settlement data at the measurement point to be predicted is calculated using the following formula:

[0101] e = W e ·[r1,…,r n ]+b e ;

[0102] Where e is the inferred value of the foundation pit settlement data at the measurement point to be predicted, and W e b is the weight matrix; e Here is the bias matrix; [r1,…,r n This indicates that all representations are merged and connected.

[0103] (2) Second step: Determine the model architecture and train it.

[0104] Based on the conclusions of data analysis, appropriate data and model architecture are selected for training. Temporal autocorrelation analysis can determine the predictability of the data, thereby determining the length of the time window for the input data; spatial similarity analysis can identify the most similar measurement points to the measurement points to be predicted, thus determining the spatial source of the input data; spatiotemporal causality analysis can determine which measurement points' data are meaningful when making predictions, thereby selecting data suitable as input features; spatiotemporal consistency analysis can evaluate the feasibility of using this dataset for intelligent reasoning and determine the generalization ability of the method at different construction stages.

[0105] Due to the limited number of samples available for model training in foundation excavation projects, traditional time series processing models such as LSTM and GRU often encounter overfitting problems. To address this issue, the proposed model consists of two main modules: a spatial information processing module and a representation sequence fusion module. These two modules run sequentially; the former processes spatial information, while the latter manages sequence information, ultimately yielding an estimated surface settlement. Please refer to [reference needed]. Figure 3 This is a schematic diagram of the model architecture for the foundation pit settlement inference model.

[0106] The purpose of the spatial information processing module is to process a set of auxiliary / nearby measuring points. It extracts common settlement patterns from different stages. The working principle of this module can be abstracted as extracting common settlement patterns from surface settlement time series. Generate a representation sequence in, and The subscript t represents a specific historical moment, where s t For auxiliary measuring points Settlement value (s)1 ,…,s m The same neural network processes the data s from each time step of the auxiliary measurement points sequentially. t Thus, the represented output r is obtained. t This module uses an MLP neural network, and the computation process can be represented as follows:

[0107] r t =relu(W r ·s t +b r (5)

[0108] Among them, W r b is the weight matrix; r is the bias matrix; ReLU is the nonlinear activation function.

[0109] In obtaining the characterization sequence The next step involves fusing these representations. Given their sequential nature, one possible approach is to use recurrent neural networks (RNNs) to process the sequence of representations. However, since these representations already contain fine spatial information, each one is crucial. To mitigate problems such as vanishing gradients and forgetting common in RNN-type networks, these representations are concatenated and then passed through an MLP, outputting an estimate of the sedimentation, e:

[0110] e = W e ·[r1,…,r n ]+b e (6)

[0111] Among them W e b is the weight matrix; e Here is the bias matrix; [r1,…,r n The symbol ] indicates that all representations are joined together.

[0112] Preferably, the step of training a preset neural network model using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output to obtain the foundation pit settlement inference model includes: obtaining preset hyperparameters; training a preset neural network model using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output based on the hyperparameters, and optimizing the hyperparameters during the training process using an asynchronous continuous halving algorithm and a grid search method, and finally training to obtain the foundation pit settlement inference model.

[0113] After determining the model architecture, appropriate hyperparameters need to be selected during training. Hyperparameters are manually configured settings used during model training, and deep learning models typically exhibit sensitivity to these hyperparameters. A well-executed hyperparameter search aims to identify parameter combinations that optimize the performance of a given task, thereby improving the model's accuracy and generalization. This paper employs a combined approach of the Asynchronous Successive Halving Algorithm (ASHA) and the Grid Search Method for hyperparameter optimization. ASHA defines an approximate search space, while the Grid Search Method identifies the optimal hyperparameter combinations. Please refer to [link to relevant documentation]. Figure 4 This is a schematic diagram of the hyperparameter tuning method for the foundation pit settlement inference model.

[0114] Preferably, after the foundation pit settlement inference model infers the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data and outputs the foundation pit settlement data of the measurement point to be predicted, the method further includes: performing Shapley additive interpretation analysis on the foundation pit settlement data of the measurement point to be predicted output by the foundation pit settlement inference model according to a preset interpretable machine learning model, calculating the impact of changes in each input feature of the foundation pit settlement inference model on the inferred foundation pit settlement data of the measurement point to be predicted; and statistically determining the importance of each input feature based on the impact of changes in each input feature on the inferred foundation pit settlement data of the measurement point to be predicted.

[0115] Preferably, the settlement data of the foundation pit at the measurement points to be predicted, output by the foundation pit settlement inference model, are subjected to Shapley additive interpretation analysis in the following manner:

[0116]

[0117] Where g is the interpretable machine learning model, x′ is the simplified data input that can be mapped to the original input data points, m is the number of input features, and φ j Let φ0 be the attribute value of each input feature, and φ1 be the mean of all predicted values ​​for the samples, where {x1, ..., x0} is the value assigned to each input feature. p} represents the set of all input features, {x1,…,x...} p}\{x j} represents division by x j The possible set of all input features other than f x (S) is the prediction of the feature subset S.

[0118] (3) Third step: Use interpretable machine learning models to explain the reasoning results

[0119] To enhance the interpretability of the proposed model, Shapley additive explanations (SHAP) analysis is performed, a widely used technique in the field of interpretable machine learning. SHAP analysis originates from the concept of Shapley values ​​in cooperative game theory. First, a reference value is established, typically the average regression value in the dataset. Then, for each feature, it calculates the impact on the model's inference value when that feature is changed in the dataset. The average of these impact values ​​represents the feature's overall contribution to the regression result. SHAP analysis interprets Shapley values ​​as additive feature attributes and represents the inference value as the sum of the values ​​of each input feature attribute:

[0120]

[0121] Where g is the explanatory model; x′ is the simplified data input that can be mapped to the original input data points; m is the number of input features; φ j φ1 represents the attribute value (Shapley value) for each feature; φ0 represents the mean of the predicted values ​​for all samples; {x1,…,x p} represents the set of all input features; {x1,…,x p}\{x j} represents division by x j The possible set of all input features other than f; x (S) is the prediction of the feature subset S.

[0122] The SHAP method is used to perform an impact effect analysis on all input features, illustrating the influence of each input feature on the final estimate. Generally, most features have a negative impact on the results; that is, the smaller the feature value, the greater the negative contribution. When neighboring measurement points experience more significant settlement, the settlement of the measurement point to be inferred is usually greater. Furthermore, the SHAP method can be used to perform an importance analysis on input features, evaluating the most important parts of all features from the data perspective. For example, data collected closer to the current time is more important, as are measurement points with settlement mechanisms more similar to the measurement point to be inferred. Specifically, the SHAP method provides the impact of each feature on the result, summing the impact of all data in each feature category to compare the importance of each feature category.

[0123] Example 2

[0124] Please refer to Figure 5 This is a schematic diagram of the structure of a foundation pit settlement prediction device provided in an embodiment of the present invention. The device includes: a data acquisition module, a data prediction module, and a model training module.

[0125] The data acquisition module is used to acquire foundation pit settlement monitoring data of the measurement point that is most similar to the foundation pit settlement pattern of the measurement point to be predicted; wherein, the foundation pit settlement monitoring data includes: the settlement status of the ground surface and buildings within a preset range of the foundation pit where the measurement point is located;

[0126] The data prediction module is used to input the foundation pit settlement monitoring data into a preset foundation pit settlement inference model, so that the foundation pit settlement inference model can infer the foundation pit settlement data of the measurement point to be predicted based on the foundation pit settlement monitoring data, and output the foundation pit settlement data of the measurement point to be predicted.

[0127] The model training module is used to acquire historical foundation pit settlement monitoring data of each measurement point; perform time autocorrelation analysis, spatial similarity analysis, spatiotemporal causality analysis, and spatiotemporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point; then, based on the analysis results, determine the measurement point most similar to the foundation pit settlement pattern of the measurement point to be predicted as the key measurement point; and select historical foundation pit settlement monitoring data with a causal relationship with the historical foundation pit settlement monitoring data of the measurement point to be predicted from the historical foundation pit settlement monitoring data of the key measurement point as key data; using the key data as input and the historical foundation pit settlement data corresponding to the measurement point to be predicted as output, train a preset neural network model to obtain the foundation pit settlement inference model; wherein, the foundation pit settlement inference model includes: a spatial information processing module and a representation sequence fusion module; the spatial information processing module is used to generate a corresponding representation sequence based on the time series of the key data; the representation sequence fusion module is used to fuse the representation sequence and calculate the inferred value of the foundation pit settlement data of the measurement point to be predicted based on the fused representation sequence.

[0128] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0129] Those skilled in the art will clearly understand that, for convenience and simplicity, the specific working process of the device described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0130] Example 3

[0131] Accordingly, embodiments of the present invention provide an electronic device, the device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the foundation pit settlement prediction method described in the above embodiments of the invention.

[0132] The electronic device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The device may include, but is not limited to, a processor and a memory.

[0133] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the device, connecting various parts of the device via various interfaces and lines.

[0134] Example 4

[0135] Accordingly, embodiments of the present invention provide a storage medium, the storage medium including a stored computer program, wherein, when the computer program is running, it controls the device where the storage medium is located to execute the foundation pit settlement prediction method described in the above embodiments of the invention.

[0136] The memory can be used to store the computer program. The processor implements various functions of the device by running or executing the computer program stored in the memory and calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0137] The storage medium is a computer-readable storage medium, and the computer program is stored in the computer-readable storage medium. When executed by a processor, the computer program can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0138] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method of predicting settlement of a foundation pit, characterized by, The method comprises the following steps: obtaining the foundation pit settlement monitoring data of the measurement point most similar to the settlement mode of the to-be-predicted measurement point; wherein the foundation pit settlement monitoring data comprises the settlement conditions of the ground surface and the building within the preset range of the foundation pit where the measurement point is located; inputting the foundation pit settlement monitoring data into a preset foundation pit settlement inference model, so that the foundation pit settlement inference model infers the foundation pit settlement data of the to-be-predicted measurement point according to the foundation pit settlement monitoring data, and outputs the foundation pit settlement data of the to-be-predicted measurement point; wherein the training of the foundation pit settlement inference model comprises: obtaining the historical foundation pit settlement monitoring data of each measurement point; performing time autocorrelation analysis, spatial similarity analysis, spatio-temporal causality analysis and spatio-temporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point, then determining the measurement point most similar to the settlement mode of the to-be-predicted measurement point as a key measurement point according to the analysis results, and selecting the historical foundation pit settlement monitoring data having a causal relationship with the historical foundation pit settlement monitoring data of the to-be-predicted measurement point from the historical foundation pit settlement monitoring data of the key measurement point as key data; training a preset neural network model by taking the key data as input and the historical foundation pit settlement data corresponding to the to-be-predicted measurement point as output, to obtain the foundation pit settlement inference model; wherein the foundation pit settlement inference model comprises a spatial information processing module and a representation sequence fusion module; the spatial information processing module is used to generate a corresponding representation sequence according to the time sequence of the key data; the representation sequence fusion module is used to fuse the representation sequence, and calculate the inference value of the foundation pit settlement data of the to-be-predicted measurement point according to the fused representation sequence; wherein the corresponding representation sequence is generated according to the following formula: ； wherein, a representation sequence corresponding to the time sequence of key data, is a weight matrix, is a bias matrix, is a nonlinear activation function, is a measured value of the foundation pit settlement at the measurement point at the historical moment t.

2. The foundation settlement prediction method according to claim 1, wherein the time autocorrelation analysis, spatial similarity analysis, spatio-temporal causality analysis and spatio-temporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point comprises: performing time autocorrelation analysis on the historical foundation pit settlement monitoring data of each measurement point according to the autocorrelation function and the partial autocorrelation function, to obtain the autocorrelation between the historical foundation pit settlement monitoring data of each measurement point; performing spatial similarity analysis on the historical foundation pit settlement monitoring data of each measurement point according to dynamic time warping and clustering, to obtain the similarity of the foundation pit settlement mode between different measurement points; performing spatio-temporal causality analysis on the historical foundation pit settlement monitoring data of each measurement point according to the Granger causality test method, to obtain the causal relationship between the variables in the historical foundation pit settlement monitoring data of each measurement point; performing spatio-temporal consistency analysis on the historical foundation pit settlement monitoring data of each measurement point according to the Spearman correlation coefficient matrix of the same measurement point in different periods, to determine whether the settlement mode of the same measurement point in different periods is consistent.

3. The foundation settlement prediction method according to claim 1, wherein the training of the preset neural network model by taking the key data as input and the historical foundation pit settlement data corresponding to the to-be-predicted measurement point as output, to obtain the foundation pit settlement inference model, comprises: obtaining preset hyperparameters; According to the hyperparameters, the key data is input, the historical foundation pit settlement data corresponding to the to-be-predicted measurement point is output, a preset neural network model is trained, and the hyperparameters are optimized according to the asynchronous continuous halving algorithm and the grid search method in the training process, and finally the foundation pit settlement reasoning model is trained.

4. The foundation settlement prediction method according to claim 1, wherein After the foundation pit settlement reasoning model reasons the foundation pit settlement data of the to-be-predicted measurement point according to the foundation pit settlement monitoring data, and outputs the foundation pit settlement data of the to-be-predicted measurement point, the method further comprises: According to a preset interpretable machine learning model, the foundation pit settlement data of the to-be-predicted measurement point output by the foundation pit settlement reasoning model is subjected to Shapley additive interpretation analysis, and the influence of each input feature change on the inferred foundation pit settlement data of the to-be-predicted measurement point is calculated. According to the influence of each input feature change on the inferred foundation pit settlement data of the to-be-predicted measurement point, the importance of each input feature is statistically obtained.

5. A device for predicting settlement of a foundation pit, characterized by, Comprise: a data acquisition module, a data prediction module, and a model training module; The data acquisition module is configured to acquire foundation pit settlement monitoring data of a measurement point most similar to a foundation pit settlement mode of a to-be-predicted measurement point; wherein the foundation pit settlement monitoring data comprises a settlement condition of a ground surface and a building within a preset range of a foundation pit where the measurement point is located; The data prediction module is configured to input the foundation pit settlement monitoring data into a preset foundation pit settlement reasoning model, so that the foundation pit settlement reasoning model reasons foundation pit settlement data of a to-be-predicted measurement point according to the foundation pit settlement monitoring data, and outputs the foundation pit settlement data of the to-be-predicted measurement point; The model training module is configured to acquire historical foundation pit settlement monitoring data of each measurement point; perform time autocorrelation analysis, spatial similarity analysis, temporal and spatial causality analysis, and temporal and spatial consistency analysis on the historical foundation pit settlement monitoring data of each measurement point; then determine a measurement point most similar to a foundation pit settlement mode of a to-be-predicted measurement point as a key measurement point according to the analysis results, and select historical foundation pit settlement monitoring data having a causal relationship with historical foundation pit settlement monitoring data of the to-be-predicted measurement point from the historical foundation pit settlement monitoring data of the key measurement point as key data; input the key data, and output historical foundation pit settlement data corresponding to the to-be-predicted measurement point to train a preset neural network model, and obtain the foundation pit settlement reasoning model; wherein the foundation pit settlement reasoning model comprises a spatial information processing module and a representation sequence fusion module; the spatial information processing module is configured to generate a corresponding representation sequence according to a time sequence of the key data; and the representation sequence fusion module is configured to fuse the representation sequence, and calculate a reasoning value of the foundation pit settlement data of the to-be-predicted measurement point according to the fused representation sequence; wherein the corresponding representation sequence is generated according to the following formula: ； wherein, is a representation sequence corresponding to the time sequence of key data, is a weight matrix, is a bias matrix, is a nonlinear activation function, is a measured value of the foundation pit settlement at the measurement point at the historical moment t.

6. An electronic device, comprising: The computer program is configured to be executed by the processor, and the processor executes the computer program to implement the foundation pit settlement prediction method according to any one of claims 1 to 4.

7. A storage medium, characterized by The storage medium comprises a stored computer program, wherein the computer program controls a device where the storage medium is located to execute the foundation pit settlement prediction method according to any one of claims 1 to 4 when the computer program is running.

Images