CN114996830A - Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel - Google Patents

Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel Download PDF

Info

Publication number
CN114996830A
CN114996830A CN202210928018.2A CN202210928018A CN114996830A CN 114996830 A CN114996830 A CN 114996830A CN 202210928018 A CN202210928018 A CN 202210928018A CN 114996830 A CN114996830 A CN 114996830A
Authority
CN
China
Prior art keywords
model
tunnel
gcn
existing tunnel
prediction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202210928018.2A
Other languages
Chinese (zh)
Other versions
CN114996830B (en
Inventor
张立茂
林鹏辉
吴贤国
覃亚伟
徐文胜
张军
姚春桥
王金峰
曾铁梅
陶文涛
熊朝辉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN202210928018.2A priority Critical patent/CN114996830B/en
Publication of CN114996830A publication Critical patent/CN114996830A/en
Application granted granted Critical
Publication of CN114996830B publication Critical patent/CN114996830B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/08—Construction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/10—Internal combustion engine [ICE] based vehicles
    • Y02T10/40—Engine management systems

Abstract

The invention belongs to the technical field of tunnel construction, and particularly discloses a visual safety assessment method and equipment for a shield tunnel to pass through an existing tunnel. The method comprises the following steps: constructing a tunnel construction parameter index system and acquiring the existing data; constructing a GCN prediction model of the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface based on the GCN regression model and the existing data, and explaining the GCN prediction model by adopting an SHAP model to obtain the sensitivity of the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface to different characteristics; simulating uncertainty in soil parameters by using Monte Carlo, and linearly regressing the predicted value and the actual value of accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface to obtain a confidence interval of a prediction result when the uncertainty of the soil and the model is considered at the same time, so as to calculate the safety risk; and visualizing the interaction and the safety risk of the operation process. The invention has accurate calculation of safety risk and ensures the reliability of construction.

Description

Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel
Technical Field
The invention belongs to the technical field of tunnel construction, and particularly relates to a visual safety assessment method and equipment for a shield tunnel to pass through an existing tunnel downwards, in particular to a visual safety assessment method and equipment for a large-diameter shield tunnel to pass through the existing tunnel downwards based on interactive interpretable artificial intelligence.
Background
Due to the progress of shield tunnel technology, the number of large-diameter shield tunnels is increasing, and in consideration of the complexity of underground tunnel networks in the current city, designers are inevitably confronted with the situation that a newly-built tunnel penetrates through an existing tunnel. In this case, it is necessary to consider the existence of the existing tunnel at the initial stage of design to find a suitable position for the construction of the new tunnel. Due to the rapid development of artificial intelligence algorithms in recent years, such as Artificial Neural Networks (ANN), Gaussian Process Regression (GPR), gradient enhanced decision trees (GBDT), GCN (graph convolution network), existing datasets can now be used to predict tunnel excavation damage. Aiming at the problem of tunnel construction, the machine learning method can fully and reasonably simulate the damage caused by construction, but the application of a machine learning algorithm needs considerable background knowledge and is difficult to be directly applied to a tunnel engineer. Therefore, the enhancement of interactivity and interpretability of artificial intelligence is especially important for the engineering field. Therefore, by applying interpretable artificial intelligence algorithms such as SHAP and the like and combining the algorithms with BIM software commonly used in the engineering field, a designer can adjust the algorithms through interfaces in the BIM software and realize the visualization of results in a 3D model. However, simple application of artificial intelligence algorithms to make predictions of the tunnel construction process lacks consideration of the uncertainty of the variables. First, the soil quality within the construction area is not completely consistent with the parameters of the design, and if the differences in soil properties are not properly considered, the reaction may exceed the design capability, resulting in huge losses. Meanwhile, because the error of the proxy model is unavoidable, the model also generates a certain degree of uncertainty in prediction.
Based on the above drawbacks and deficiencies, there is a need in the art to predict and explain the damage caused by the excavation of the existing tunnel, and to quantify the safety risk reliability coefficient of the new tunnel, which is currently designed, so as to obtain a more reliable prediction result.
Disclosure of Invention
Aiming at the defects or improvement requirements in the prior art, the invention provides a visual safety assessment method and equipment for a large-diameter shield tunnel to pass through an existing tunnel downwards based on interactive interpretable artificial intelligence, wherein the ground surface accumulated settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation surface supporting force (EFP) are predicted through a GCN regression model; explaining the proxy model through SHAP technology to obtain the sensitivity of the target to different characteristics; obtaining a confidence interval of a prediction result when the soil property and the model uncertainty are considered simultaneously by using Monte Carlo simulation and linear regression, thereby calculating the system security risk; machine learning codes are embedded into BIM software through a parametric modeling tool, so that interaction of an artificial intelligence operation process and visualization of results in a 3D model are realized. The method can predict the excavation damage of the existing tunnel, simultaneously consider various types of uncertainty such as geological conditions and meta-models, analyze the safety risk of current design to ensure the reliability, and enable the machine learning process to be interactive and the result to be visualized.
In order to achieve the above object, according to an aspect of the present invention, a visual safety assessment method for a shield tunnel passing through an existing tunnel is provided, which includes the following steps:
step one, constructing a tunnel construction parameter index system, wherein the index system comprises stratum parameters, soil parameters and shield tunneling parameters, and acquiring the existing data;
step two, constructing a GCN prediction model of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface based on the GCN regression model and the existing data;
thirdly, explaining a GCN prediction model by using a SHAP model to obtain the sensitivity of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface to different characteristics;
simulating uncertainty in soil parameters by using Monte Carlo, and linearly regressing the predicted values and actual values of accumulated surface settlement, existing tunnel deformation and excavation face supporting force to obtain a confidence interval of a prediction result when the uncertainty of the soil and the model is considered at the same time, so that the safety risk of the large-diameter shield in the construction process of passing through the existing tunnel is calculated;
and step five, visualizing the interaction and the safety risk of the operation process.
As a further preferred, step two specifically comprises the following steps:
(1) dividing the existing data into a training set and a testing set according to a specified proportion;
(2) constructing a GCN regression model, taking the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface as outputs based on a training set, and training the GCN regression model to obtain a GCN prediction model of the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface;
(3) testing the GCN prediction model by adopting a test set;
(4) and (4) carrying out prediction precision analysis on the GCN prediction model, and evaluating the GCN prediction model by adopting a mean square error and a root mean square error.
As a further preferred, step three specifically comprises the following steps:
through the SHAP model, the black box model of the GCN prediction model is simplified, and the sensitivity of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface to each characteristic is obtained:
Figure 413016DEST_PATH_IMAGE001
wherein the content of the first and second substances,
Figure 656915DEST_PATH_IMAGE002
the number of features that represent the input is,
Figure 846588DEST_PATH_IMAGE003
representation feature
Figure 90619DEST_PATH_IMAGE004
The simplified characteristic value of (2) is,
Figure 453467DEST_PATH_IMAGE005
in order to simplify the reference values of the model,
Figure 958398DEST_PATH_IMAGE006
is characterized in that
Figure 194338DEST_PATH_IMAGE004
The value of the SHAP of (1),
Figure 112615DEST_PATH_IMAGE007
is an importance value of a feature;
the characteristics
Figure 951258DEST_PATH_IMAGE004
The calculation model of the SHAP value of (1) is as follows:
Figure 717220DEST_PATH_IMAGE008
wherein the content of the first and second substances,
Figure 311013DEST_PATH_IMAGE009
for the output values of the original model, S is a non-0 subset of the features used in the model and N is the set of all input features.
As a further preferred, in step four, the uncertainty in the soil parameter simulation using monte carlo includes: given the probability distribution of soil parameters, generating according to the distributionnAnd (3) grouping samples, wherein the sample mean value obeys normal distribution, and after the Monte Carlo sample mean value is taken as an output predicted value after the uncertainty of the soil parameters is considered, the confidence interval of the mean value can be calculated according to the following formula:
Figure 654269DEST_PATH_IMAGE010
wherein the content of the first and second substances,
Figure 109653DEST_PATH_IMAGE011
is the predicted value mean of the Monte Carlo samples,
Figure 713809DEST_PATH_IMAGE012
the standard deviation of the predicted values for the monte carlo samples,
Figure 416186DEST_PATH_IMAGE013
is at a significant level of
Figure 59788DEST_PATH_IMAGE014
The value of the standard normal distribution of time,
Figure 364867DEST_PATH_IMAGE015
is the confidence interval.
As a further preferred, in step four, the quantifying the model uncertainty with the prediction interval of the regression process includes: taking data of an original data set as
Figure 230055DEST_PATH_IMAGE016
The predicted value of the GCN prediction model is used as
Figure 916383DEST_PATH_IMAGE017
By using the least square method, a linear regression model with the following format can be trained:
Figure 93286DEST_PATH_IMAGE018
at a given sample capacity
Figure 77423DEST_PATH_IMAGE019
Then, the corresponding prediction interval can be calculated by the following formula:
Figure 610166DEST_PATH_IMAGE020
wherein, the first and the second end of the pipe are connected with each other,
Figure 778979DEST_PATH_IMAGE021
is based on
Figure 52966DEST_PATH_IMAGE022
The predicted value of the regression model is determined,
Figure 653843DEST_PATH_IMAGE023
is composed of
Figure 352677DEST_PATH_IMAGE024
Degree of freedom
Figure 380807DEST_PATH_IMAGE025
The value of t at the significant level is,
Figure 673248DEST_PATH_IMAGE026
in order to be the standard error of the residual error,
Figure 858242DEST_PATH_IMAGE027
and
Figure 365578DEST_PATH_IMAGE028
are respectively as
Figure 751560DEST_PATH_IMAGE029
The mean and the variance of (a) is,PIis a prediction interval.
Preferably, in the fourth step, the influence of uncertainty of the soil and the model on the result is considered at the same time, and the upper and lower bounds of the prediction interval are substituted into the calculation of the confidence interval, so as to obtain the new upper and lower bounds of the confidence interval as follows:
Figure 124772DEST_PATH_IMAGE030
wherein the content of the first and second substances,
Figure 90889DEST_PATH_IMAGE031
and
Figure 639682DEST_PATH_IMAGE032
respectively are the average values of the upper and lower bounds of the prediction interval of the Monte Carlo sample,
Figure 55620DEST_PATH_IMAGE033
and
Figure 870123DEST_PATH_IMAGE034
respectively an upper bound and a lower bound of a confidence interval after simultaneously considering the uncertainty of the soil and the model.
Preferably, in the fourth step, for performing security risk analysis, the new confidence interval is compared with the retrievable range of the target value, and the percentage of the intersection in the confidence interval is the reliability coefficient, and the operation mode is as follows: if the confidence interval is lower bound
Figure 803444DEST_PATH_IMAGE035
Not greater thanUAnd upper bound of confidence interval
Figure 65798DEST_PATH_IMAGE036
Not less thanLThen r =1, otherwise one proceeds as follows: if it is used
Figure 606632DEST_PATH_IMAGE037
And is
Figure 95383DEST_PATH_IMAGE038
Then, then
Figure 894711DEST_PATH_IMAGE039
Otherwise, if
Figure 27883DEST_PATH_IMAGE040
And is
Figure 988886DEST_PATH_IMAGE041
Then, then
Figure 699353DEST_PATH_IMAGE042
Otherwise, if
Figure 849843DEST_PATH_IMAGE043
And is provided with
Figure 696576DEST_PATH_IMAGE044
Then, then
Figure 94060DEST_PATH_IMAGE045
And if not, the step (B),
Figure 901610DEST_PATH_IMAGE046
wherein, in the step (A),U、Lthe upper and lower bounds of the target value range are,rfor the reliability coefficient, the closer the coefficient is to 1, the smaller the system security risk. More specifically, the operation method is as follows:
Figure 980424DEST_PATH_IMAGE047
preferably, in the fifth step, a GCN model for predicting surface accumulated settlement (GSS), Existing Tunnel Deformation (ETD) and excavation face supporting force (EFP), SHAP interpretability analysis of the model, and a parameterized modeling process for embedding three parts of machine learning codes for safety risk analysis in the construction process of a large-diameter shield-penetrating existing tunnel into Grasshopper and performing 3D model rendering and calculation result presentation are performed through a Grasshopper plug-in the Rhino software, so that interaction of an artificial intelligence operation process and visualization of the result in the 3D model are realized.
According to another aspect of the present invention, there is also provided a visual safety assessment system for a shield tunnel passing through an existing tunnel, including: the first main module is used for constructing a tunnel construction parameter index system, wherein the index system comprises stratum parameters, soil parameters and shield tunneling parameters and acquires the existing data; the second main module is used for constructing a GCN prediction model of accumulated earth surface settlement, existing tunnel deformation and excavation face supporting force based on the GCN regression model and the existing data; the third main module is used for explaining the GCN regression model by adopting an SHAP model to obtain the sensitivity of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface to different characteristics; the fourth main module is used for simulating uncertainty in soil parameters by using Monte Carlo, and simultaneously performing linear regression on the predicted value and the actual value of accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface to obtain a confidence interval of a prediction result when the uncertainty of the soil and the model is considered at the same time, so that the safety risk in the construction process of passing the existing tunnel under the large-diameter shield is calculated; and the fifth main module is used for visualizing the interaction and the safety risk of the operation process.
According to another aspect of the present invention, there is also provided an electronic apparatus including:
at least one processor, at least one memory, and a communication interface; wherein, the first and the second end of the pipe are connected with each other,
the processor, the memory and the communication interface are communicated with each other;
the storage stores program instructions which can be executed by the processor, and the processor calls the program instructions to execute a visual safety assessment method for the shield tunnel to pass through the existing tunnel.
According to another aspect of the present invention, there is also provided a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform a method for visual security assessment of shield tunneling through an existing tunnel.
Generally, compared with the prior art, the above technical solution conceived by the present invention mainly has the following technical advantages:
1. the method provided by the invention considers various types of uncertainty such as geological conditions and meta-models while predicting the excavation damage of the existing tunnel passing through the tunnel, analyzes the currently designed safety risk to ensure the reliability, and enables the machine learning process to be interactive and the result to be visualized. The invention considers the uncertainty of the soil and the meta-model, so that the result is more reliable, and a more conservative reference is provided for decision making.
2. According to the method, the GCN regression model is used for predicting the excavation damage of the existing tunnel penetrating through the large-diameter shield tunnel, and compared with the traditional neural network, the accuracy of algorithms such as integrated learning is higher, so that the prediction result is more accurate, and the reliability is higher.
3. According to the method, the SHAP model is used for explaining the GCN regression model, the accumulated settlement of the earth surface, the deformation of the existing tunnel and the sensitivity of the supporting force of the excavation surface to different characteristics are obtained, the black box model is visualized, the understanding of a user on the model in the prediction process is further ensured, and the potential risk of blindly using the black box model is avoided.
3. According to the method, the Monte Carlo simulation and linear regression technology are utilized to obtain the confidence interval of the prediction result under the condition that the uncertainty of the soil and the model is considered at the same time, so that a more reliable prediction result is provided for a user, and potential safety hazards caused by incomplete consideration of the uncertainty factor are avoided. Meanwhile, the problem that the user is over-conservative in design due to insufficient understanding of the uncertain conditions, and economic and resource waste is caused is avoided.
Drawings
Fig. 1 is a flowchart of a visual security assessment method for a shield tunnel passing through an existing tunnel according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a visual safety assessment system for a shield tunnel passing through an existing tunnel according to an embodiment of the present invention;
fig. 3 is a schematic physical structure diagram of an electronic device according to an embodiment of the present invention;
FIG. 4 shows 16 factor features of a tunnel excavation project according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating the results of MAE, RMSE, and R2 of training data and test data corresponding to EFP, ETD, and GSS, according to an embodiment of the present invention;
fig. 6 is a schematic diagram of the calculation of the security risk in 4 scenes, where the variation range is ± 1m and the angle variation range is ± 8 ° according to the embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
As shown in fig. 1, a visual safety assessment method for a shield tunnel passing through an existing tunnel provided by an embodiment of the present invention includes: constructing a tunnel construction parameter index system, wherein the index system comprises stratum parameters, soil parameters and shield tunneling parameters, and acquiring the existing data; according to the existing data, a GCN regression model is constructed to predict the accumulated surface settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation face supporting force (EFP); explaining and predicting a proxy model of the surface accumulated settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation face supporting force (EFP) through a SHAP technology to obtain the sensibility of the surface accumulated settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation face supporting force (EFP) to different characteristics so as to simplify and explain the black box model; simulating uncertainty in soil parameters by using Monte Carlo, and performing linear regression on predicted values and sample actual values of surface accumulated settlement (GSS), Existing Tunnel Deformation (ETD) and excavation face supporting force (EFP) to obtain a confidence interval of a prediction result when soil quality and model uncertainty are considered simultaneously, so as to calculate a safety risk in the construction process of passing the large-diameter shield through the existing tunnel; by means of a Grasshopper plug-in the Rhino software, a parameterized modeling process of embedding machine learning codes of three parts including a GCN prediction model for predicting surface accumulated settlement (GSS), Existing Tunnel Deformation (ETD) and excavation face supporting force (EFP), SHAP interpretability analysis of the model and safety risk analysis in the construction process of passing a large-diameter shield through an existing tunnel into the Grasshopper is carried out, 3D model rendering and calculation result presentation are carried out, and interaction of an artificial intelligence operation process and visualization of results in the 3D model are achieved.
More specifically, the shield construction parameter index system comprises: stratum parameters, soil parameters, shield tunneling parameters, and cumulative Ground Surface Subsidence (GSS), Existing Tunnel Deformation (ETD), and excavation face support force (EFP). In the above-mentioned existing data, the electronic sensor monitoring of tunnel construction includes the accumulated surface settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation face supporting force (EFP) caused in the tunneling process. In the invention, the accumulated surface settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation face supporting force (EFP) are used as the output of the GCN regression model, the GCN regression model is trained and tested through the existing data to obtain the trained GCN regression model, and the regression model is used as a prediction model. Furthermore, in the invention, soil parameters in a construction site are recorded simultaneously, and a corresponding GCN regression model is trained by taking relevant stratum parameters, soil parameters and shield tunneling parameters as input values. More specifically, the existing data samples are collected and divided, 80% of the samples are randomly selected as a training set, the remaining 20% of the samples are used as a test set, the GCN regression model is subjected to prediction accuracy analysis, and the accuracy of the GCN regression model prediction is evaluated by Mean Square Error (MSE) and Root Mean Square Error (RMSE).
More specifically, in order to simulate the underground environment of the tunnel passing through the existing tunnel under the large-diameter shield tunnel, six soil layers are considered, and 16 characteristics under the conditions of soil properties and tunnel positions are selected as input variables, as shown in fig. 4, the 16 characteristics are as follows: the method comprises the following steps of embedding depth of a newly-built tunnel, included angle of a new tunnel and an old tunnel, Young modulus of a soil layer I, Poisson ratio of a soil layer I, Young modulus of a soil layer II, Poisson ratio of a soil layer II, Young modulus of a soil layer III, Poisson ratio of a soil layer III, Young modulus of a soil layer IV, Poisson ratio of a soil layer IV, Young modulus of a soil layer V, Poisson ratio of a soil layer V, Young modulus of a soil layer VI, Poisson ratio of a soil layer VI, friction angle of a soil layer VI and cohesion of a soil layer VI. When the ensemble learning model is trained, 240 groups of samples are used, 48 samples are randomly selected from the 240 groups of samples to serve as test data, and the rest 192 samples are used as a training data set to establish a GCN regression model. After the GCN regression model is trained, evaluating the GCN regression model by using a decision coefficient, a root mean square error and an average absolute error, wherein the calculation models of the decision coefficient, the root mean square error and the average absolute error are respectively as follows:
Figure 9560DEST_PATH_IMAGE048
in the formula (I), the compound is shown in the specification,
Figure 390994DEST_PATH_IMAGE049
in order to determine the coefficients, the coefficients are,
Figure 810474DEST_PATH_IMAGE050
is the root mean square error (rms) of the signal,
Figure 489717DEST_PATH_IMAGE051
n is the total number of data in the sample data set;
Figure 186409DEST_PATH_IMAGE052
respectively representing a model predicted value and an actual observed value;
Figure 863378DEST_PATH_IMAGE053
the average actual value is indicated.
The training process of the GCN regression model comprises the following steps: (1) dividing the existing data into a training set and a testing set according to a specified proportion; (2) constructing a GCN regression model, taking the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface as outputs based on a training set, and training the GCN regression model to obtain a GCN prediction model of the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface; (3) testing the GCN prediction model by adopting a test set; (4) and (4) carrying out prediction accuracy analysis on the GCN prediction model, and evaluating the GCN prediction model by adopting a mean square error and a root mean square error.
In the present application, the GCN prediction model is compared with other machine learning techniques, corresponding MAE, RMSE and R 2 As shown in fig. 5. Compared with a Random Forest (RF), xgboost (extreme Gradient boosting) method, the model trained by GCN has higher precision and lower overfitting tendency.
Further, in the present invention, SHAP (Shapley Additive ExPlations, explain machine learning model) is used as an interpretable artificial intelligence technology, and the black box model forming the GCN regression model is simplified by an Additive feature factorization method to obtain the importance of each feature:
Figure 98050DEST_PATH_IMAGE054
wherein the content of the first and second substances,
Figure 394033DEST_PATH_IMAGE002
the number of features that are to be input is indicated,
Figure 69865DEST_PATH_IMAGE055
representation of features
Figure 980052DEST_PATH_IMAGE004
The simplified characteristic value of (a) is,
Figure 515070DEST_PATH_IMAGE056
to simplify the reference values of the model, it is often desirable to predict the outcome of the sample,
Figure 208220DEST_PATH_IMAGE057
is characterized in thatiThe value of the SHAP of (1),
Figure 332033DEST_PATH_IMAGE058
for the value of the importance of a feature,
Figure 960592DEST_PATH_IMAGE059
can be calculated by the following formula:
Figure 841960DEST_PATH_IMAGE060
wherein the content of the first and second substances,
Figure 666697DEST_PATH_IMAGE061
the output value of the original model can be obtained
Figure 458066DEST_PATH_IMAGE062
And (4) calculating.
Figure 710056DEST_PATH_IMAGE002
Representing the number of features of the input, S is a non-0 subset of the features used in the model, and N is the set of all input features.
Further, in the present invention, soil parameter uncertainty is simulated by monte carlo.
Given the probability distribution of the parameters, the probability distribution is generated according to the distribution
Figure 813141DEST_PATH_IMAGE063
And (4) grouping the samples. Due to the fact that
Figure 989039DEST_PATH_IMAGE063
In actual calculation, the value is usually very large, and the sample mean value obeys normal distribution by the central limit theorem. After taking the Monte Carlo sample mean value as an output predicted value after considering the uncertainty of the soil property parameters, the mean value can be calculated according to the following formulaA signal interval:
Figure 759549DEST_PATH_IMAGE064
wherein the content of the first and second substances,
Figure 182440DEST_PATH_IMAGE065
is the predicted value mean of the monte carlo samples,
Figure 913767DEST_PATH_IMAGE012
the standard deviation of the predicted values for the monte carlo samples,
Figure 80306DEST_PATH_IMAGE066
is at a significant level of
Figure 705322DEST_PATH_IMAGE067
The value of the standard normal distribution of time,
Figure 112164DEST_PATH_IMAGE015
is the confidence interval.
In particular, it is assumed in the examples that all soil parameters follow a normal distribution with standard deviation
Figure 517737DEST_PATH_IMAGE068
And all are
Figure 894492DEST_PATH_IMAGE069
The ratio of (d) is 10%, and the mean value is set to be the same as the input value. 10000 samples were taken during the monte carlo simulation. Significance of confidence interval was set to 0.01.
The uncertainty of the GCN prediction model is quantified by the Prediction Interval (PI) of the regression process. The method comprises the following steps: taking data of an original data set as
Figure 514960DEST_PATH_IMAGE070
Predicted value of GCN prediction model as
Figure 279654DEST_PATH_IMAGE071
By using the commonThe least squares (OLS) method can train a linear regression model with the following format:
Figure 844628DEST_PATH_IMAGE072
at a given sample capacity
Figure 369281DEST_PATH_IMAGE073
Then, the corresponding prediction interval PI can be calculated by the following formula:
Figure 827944DEST_PATH_IMAGE074
wherein the content of the first and second substances,
Figure 701222DEST_PATH_IMAGE075
is based on
Figure 628858DEST_PATH_IMAGE076
The predicted value of the regression model is determined,
Figure 940891DEST_PATH_IMAGE077
is composed of
Figure 129427DEST_PATH_IMAGE078
Degree of freedom
Figure 48972DEST_PATH_IMAGE079
The value of t at the significant level is,
Figure 916434DEST_PATH_IMAGE026
in order to be the standard error of the residual error,
Figure 704262DEST_PATH_IMAGE080
and
Figure 888249DEST_PATH_IMAGE081
are respectively as
Figure 165647DEST_PATH_IMAGE082
The mean and the variance of (a) is,PIis a prediction interval.
Specifically, in order to further test the performance of the prediction model meta-model, a linear regression model between the sample values and the predicted values is established:
Figure 192509DEST_PATH_IMAGE083
wherein the content of the first and second substances,
Figure 393814DEST_PATH_IMAGE070
is a function of the actual value of the measured value,
Figure 884838DEST_PATH_IMAGE084
is a predicted value.
The prediction interval of the meta-model may be determined from a linear regression model. As baseline results, a significance level of 0.01 was also used to obtain the interval. The application of the prediction interval can ensure the reliability of the machine learning model.
Furthermore, the influence of the uncertainty of the soil texture and the prediction model on the result is considered at the same time, and the upper and lower bounds of the prediction interval are substituted into the calculation of the confidence interval to obtain the new upper and lower bounds of the confidence interval as follows:
Figure 270820DEST_PATH_IMAGE085
wherein the content of the first and second substances,
Figure 660344DEST_PATH_IMAGE086
and
Figure 852291DEST_PATH_IMAGE087
respectively are the average values of the upper and lower bounds of the prediction interval of the Monte Carlo sample,
Figure 401084DEST_PATH_IMAGE088
and
Figure 567755DEST_PATH_IMAGE089
respectively after simultaneous consideration of soil and model uncertaintyUpper and lower bounds of confidence interval.
Further, in the invention, for safety risk analysis, the confidence interval and the target value retrievable range are compared, and the percentage of the intersection in the confidence interval is the system reliability coefficient. I.e. if the confidence interval is lower bound
Figure 631526DEST_PATH_IMAGE090
Not more thanUAnd upper bound of confidence interval
Figure 564847DEST_PATH_IMAGE088
Not less thanLThen r =1, otherwise one proceeds as follows: if it is not
Figure 577933DEST_PATH_IMAGE091
And is
Figure 368034DEST_PATH_IMAGE092
Then, then
Figure 591205DEST_PATH_IMAGE093
Otherwise, if
Figure 141267DEST_PATH_IMAGE094
And is
Figure 258127DEST_PATH_IMAGE095
Then, then
Figure 422392DEST_PATH_IMAGE096
Otherwise, if
Figure 211488DEST_PATH_IMAGE097
And is provided with
Figure 80087DEST_PATH_IMAGE098
Then, then
Figure 457978DEST_PATH_IMAGE099
And if not, the step (B),
Figure 137353DEST_PATH_IMAGE100
wherein, in the step (A),
Figure 663012DEST_PATH_IMAGE101
、
Figure 476247DEST_PATH_IMAGE102
the upper and lower bounds of the target value range are,
Figure 521695DEST_PATH_IMAGE103
is a reliability coefficient. More specifically, the operation method is as follows:
Figure 152396DEST_PATH_IMAGE104
wherein the content of the first and second substances,U、Lthe upper and lower limits of the target value range are set,rfor the reliability coefficient, the closer the coefficient is to 1, the smaller the system security risk.
Specifically, under the condition that a nearby structure exists, the maximum allowable existing tunnel settlement can be 15 mm during excavation, the maximum allowable ground surface settlement can be 30mm, and the lower limit is 0mm on the assumption that the lifting effect is not needed. The allowable shield surface pressure depends on the actual situation, and the maximum pressure in the example can be 100 kpa.
In the invention, for safety risk analysis, the final reliability of the system is taken as an average value under the condition of considering all indexes. However, in practical situations, when the risk of a certain index is too large, the safety problem of the whole system cannot be guaranteed, so if the reliability of a certain index is lower than a threshold value
Figure 571876DEST_PATH_IMAGE105
The overall system reliability is noted as 0. Namely, if the reliability coefficients of the supporting force of the excavation surface, the deformation of the existing tunnel and the accumulated settlement of the earth surface are not less than the threshold value
Figure 1852DEST_PATH_IMAGE106
Then, the calculation model of the overall reliability coefficient (safety risk coefficient) of the existing tunnel passing through the shield tunnel is as follows:
Figure 947811DEST_PATH_IMAGE107
otherwise, the overall reliability coefficient (safety risk coefficient) of the shield tunnel passing through the existing tunnel is 0.
Specifically, the reliability threshold is set to 0.3. The depth change range of the newly-built tunnel is set as
Figure 624780DEST_PATH_IMAGE108
1m, angle variation range of
Figure 610184DEST_PATH_IMAGE108
And 8 degrees are combined to calculate the safety risk under 4 scenes. Reliability coefficient under different scenesrShown in fig. 6.
In the invention, in order to make the machine learning operation process interactive and the operation result visualized, the Grasshopper plug-in which all code parts are embedded in the Rhino comprises the following steps: the method comprises the steps of GCN prediction model training, model interpretability analysis and system security risk analysis. The analysis result is displayed in the 3D model through parametric modeling, and real-time analysis warning is given. Specifically, all parameters in the operation process can be adjusted in an interface, and the result can be visually embodied through a 3D model. The lower the system security risk, the green tunnel color, and the red, otherwise.
After the uncertainty of the soil condition and the error of the machine learning model are considered, it can be seen from fig. 6 that the influence of the change of the included angle on the system safety risk does not have an obvious trend, but when the newly-built tunnel is shallow in burial depth, the system reliability coefficient is obviously improved. When the buried depth is +1m, the reliability is only 0.51, and when the buried depth is-1 m, the reliability becomes 0.88. Meanwhile, as can be seen from fig. 6, even if all the index predicted values are within the available range, a certain safety risk still exists; on the contrary, even if the index prediction value exceeds the acceptable range, the safety risk is not 100%. Therefore, after considering the uncertainty of the soil texture and the proxy model, the designer will not make decisions according to extreme conditions any more, and the design according to the method will be more robust.
The implementation basis of the various embodiments of the present invention is realized by programmed processing performed by a device having a processor function. Therefore, in engineering practice, the technical solutions and functions thereof of the embodiments of the present invention can be packaged into various modules. Based on the actual situation, on the basis of the foregoing embodiments, embodiments of the present invention provide a visualized safety assessment system for a shield tunnel to pass through an existing tunnel, where the system is configured to execute the visualized safety assessment method for a shield tunnel to pass through an existing tunnel in the foregoing method embodiments. Referring to fig. 2, the apparatus includes: the first main module is used for constructing a tunnel construction parameter index system and comprises the following steps: stratum parameters, soil parameters and shield tunneling parameters, and acquiring the existing data; the second main module is used for constructing a GCN regression model to predict the accumulated surface settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation face supporting force (EFP) according to the existing data; the third main module is used for explaining and predicting a proxy model of the ground surface accumulated settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation surface supporting force (EFP) through the SHAP technology to obtain the sensitivity of the ground surface accumulated settlement (GSS), the Existing Tunnel Deformation (ETD) and the excavation surface supporting force (EFP) to different characteristics so as to simplify and explain the black box model; the fourth main module is used for simulating uncertainty in soil parameters by using Monte Carlo, and simultaneously performing linear regression on predicted values and sample actual values of surface accumulated settlement (GSS), Existing Tunnel Deformation (ETD) and excavation face supporting force (EFP) to obtain a confidence interval of a prediction result when soil quality and model uncertainty are considered simultaneously, so that the safety risk of the large-diameter shield in the process of constructing the existing tunnel penetrating downwards is calculated; and the fifth main module is used for training a GCN model for predicting surface accumulated settlement (GSS), Existing Tunnel Deformation (ETD) and excavation face supporting force (EFP), SHAP interpretability analysis of the model and a parameterized modeling process for embedding three parts of machine learning codes for safety risk analysis in the construction process of passing a large-diameter shield through an existing tunnel into Grasshopper through a Grasshopper plug-in a Rhino software, rendering a 3D model and presenting a calculation result, and realizing interaction of an artificial intelligence operation process and visualization of the result in the 3D model.
Based on the content of the above device embodiment, as an optional embodiment, the system of the embodiment of the present invention further includes: a third sub-module, configured to implement the method that shap (simple Additive explantations) is used as an interpretable artificial intelligence technology, and a black box model is simplified by an Additive feature factorization method, and the importance of each feature is obtained:
Figure 889856DEST_PATH_IMAGE054
wherein the content of the first and second substances,
Figure 909896DEST_PATH_IMAGE002
the number of features that are to be input is indicated,
Figure 492187DEST_PATH_IMAGE055
representation feature
Figure 276472DEST_PATH_IMAGE004
The simplified characteristic value of (a) is,
Figure 48250DEST_PATH_IMAGE109
in order to simplify the reference values of the model,
Figure 578589DEST_PATH_IMAGE110
is characterized in that
Figure 721994DEST_PATH_IMAGE004
The value of the SHAP of (1),
Figure 603362DEST_PATH_IMAGE111
is an importance value of a feature;
the characteristics
Figure 420973DEST_PATH_IMAGE004
The calculation model of the SHAP value of (1) is as follows:
Figure 196031DEST_PATH_IMAGE060
wherein the content of the first and second substances,
Figure 870857DEST_PATH_IMAGE112
is the output value of the original model, S is a non-0 subset of the features used in the model, N is the set of all input features,
Figure 505101DEST_PATH_IMAGE112
can pass through
Figure 415419DEST_PATH_IMAGE113
And (4) calculating.
Based on the content of the above device embodiment, as an optional embodiment, the system of the embodiment of the present invention further includes: and the fourth sub-module is used for realizing that the uncertainty of the soil parameters is simulated through Monte Carlo. Given the probability distribution of the parameters, the probability distribution is generated according to the distribution
Figure 513825DEST_PATH_IMAGE063
And (4) grouping the samples. Due to the fact that
Figure 343241DEST_PATH_IMAGE063
In actual calculation, the value is usually very large, and the sample mean value obeys normal distribution by the central limit theorem. After the Monte Carlo sample mean value is taken as an output predicted value after the uncertainty of the soil property parameters is considered, the confidence interval of the mean value can be calculated according to the following formula:
Figure 340147DEST_PATH_IMAGE114
wherein the content of the first and second substances,
Figure 506686DEST_PATH_IMAGE115
is the predicted value mean of the monte carlo samples,
Figure 866123DEST_PATH_IMAGE012
the standard deviation of the predicted values for the monte carlo samples,
Figure 272965DEST_PATH_IMAGE066
is at a significant level of
Figure 678538DEST_PATH_IMAGE116
The value of the standard normal distribution of time,
Figure 320872DEST_PATH_IMAGE015
is the confidence interval.
Based on the content of the above device embodiment, as an optional embodiment, the fourth sub-module of the system according to the embodiment of the present invention is further configured to implement that the uncertainty of the model is quantified by using a Prediction Interval (PI) of a regression process. The method comprises the following steps: taking data of an original data set as
Figure 941341DEST_PATH_IMAGE070
Predicted value of GCN prediction model as
Figure 706034DEST_PATH_IMAGE117
Using the least squares method, a linear regression model of the following format can be trained:
Figure 271008DEST_PATH_IMAGE118
at a given sample capacity
Figure 592399DEST_PATH_IMAGE073
Then, the corresponding prediction interval can be calculated by the following formula:
Figure 254324DEST_PATH_IMAGE119
wherein the content of the first and second substances,
Figure 862023DEST_PATH_IMAGE075
is based on
Figure 461763DEST_PATH_IMAGE076
The predicted value of the regression model is determined,
Figure 258949DEST_PATH_IMAGE120
is composed of
Figure 837698DEST_PATH_IMAGE121
Degree of freedom
Figure 960506DEST_PATH_IMAGE122
The value of t at the significant level is,
Figure 500071DEST_PATH_IMAGE026
in order to be the standard error of the residual error,
Figure 412533DEST_PATH_IMAGE123
and
Figure 721154DEST_PATH_IMAGE081
are respectively as
Figure 749284DEST_PATH_IMAGE124
The mean and the variance of (a) is,PIis a prediction interval.
Based on the content of the above device embodiment, as an optional embodiment, the fourth sub-module of the system in the embodiment of the present invention is further configured to re-determine a new confidence interval on the basis of considering the influence of the soil property and the uncertainty of the proxy model on the prediction result: meanwhile, the influence of soil texture and proxy model uncertainty on the result is considered, and the upper and lower bounds of the prediction interval are substituted into the calculation of the confidence interval, so that the new upper and lower bounds of the confidence interval are obtained as follows:
Figure 900780DEST_PATH_IMAGE125
wherein, the first and the second end of the pipe are connected with each other,
Figure 226719DEST_PATH_IMAGE086
and
Figure 468476DEST_PATH_IMAGE126
respectively are the average values of the upper and lower bounds of the prediction interval of the Monte Carlo sample,
Figure 979091DEST_PATH_IMAGE088
and
Figure 493249DEST_PATH_IMAGE089
respectively an upper bound and a lower bound of a confidence interval after simultaneously considering the uncertainty of the soil and the model.
Based on the content of the foregoing device embodiment, as an optional embodiment, the fourth sub-module of the system according to the embodiment of the present invention is further configured to compare the foregoing confidence interval with a desirable range of a target value, where a percentage of the intersection in the confidence interval is a system reliability coefficient. The operation mode is as follows:
Figure 435929DEST_PATH_IMAGE127
wherein the content of the first and second substances,U、Lthe upper and lower limits of the target value range are set,rfor the reliability coefficient, the closer the coefficient is to 1, the smaller the system security risk.
Based on the content of the above device embodiment, as an optional embodiment, the fourth sub-module of the system according to the embodiment of the present invention is further configured to implement security risk analysis, and the final reliability of the system is taken as an average value in consideration of all indexes. However, in practical situations, when the risk of a certain index is too large, the safety problem of the whole system cannot be guaranteed, so if the reliability of a certain index is lower than a threshold value
Figure 843776DEST_PATH_IMAGE105
The overall system reliability is noted as 0. The method specifically comprises the following steps:
Figure 400659DEST_PATH_IMAGE128
based on the content of the above device embodiment, as an optional embodiment, the system of the embodiment of the present invention further includes: a fifth sub-module, configured to implement the Grasshopper plug-in that embeds all code portions in Rhino to make the machine learning operation process interactive and the operation result visualized, including: GCN model training, model interpretability analysis and system security risk analysis. The analysis result is displayed in the 3D model through parametric modeling, and real-time analysis warning is given.
The method of the embodiment of the invention is realized by depending on the electronic equipment, so that the related electronic equipment is necessarily introduced. To this end, an embodiment of the present invention provides an electronic apparatus, as shown in fig. 3, including: the system comprises at least one processor (processor), a communication Interface (communication Interface), at least one memory (memory) and a communication bus, wherein the at least one processor, the communication Interface and the at least one memory are communicated with each other through the communication bus. The at least one processor may invoke logic instructions in the at least one memory to perform all or a portion of the steps of the methods provided by the various method embodiments described above.
In addition, the logic instructions in the at least one memory may be implemented in software functional units and stored in a computer readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the method embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on the understanding, the above technical solutions substantially or otherwise contributing to the prior art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the various embodiments or some parts of the embodiments.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. Based on this recognition, each block in the flowchart or block diagrams may represent a module, a program segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (10)

1. A visual safety assessment method for a shield tunnel to pass through an existing tunnel is characterized by comprising the following steps:
step one, constructing a tunnel construction parameter index system, wherein the index system comprises stratum parameters, soil parameters and shield tunneling parameters, and acquiring the existing data;
step two, constructing a GCN prediction model of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface based on the GCN regression model and the existing data;
thirdly, explaining a GCN prediction model by using a SHAP model to obtain the sensitivity of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface to different characteristics;
simulating uncertainty in soil parameters by using Monte Carlo, and linearly regressing the predicted values and actual values of accumulated surface settlement, existing tunnel deformation and excavation face supporting force to obtain a confidence interval of a prediction result when the uncertainty of the soil and the model is considered at the same time, so that the safety risk of the large-diameter shield in the construction process of passing through the existing tunnel is calculated;
and fifthly, visualizing the interaction and the safety risk of the operation process.
2. The visual safety assessment method for the existing tunnel penetrated by the shield tunnel according to claim 1, wherein the second step specifically comprises the following steps:
(1) dividing the existing data into a training set and a testing set according to a specified proportion;
(2) constructing a GCN regression model, and training the GCN regression model by taking the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface as output based on a training set to obtain a GCN prediction model of the accumulated surface settlement, the deformation of the existing tunnel and the supporting force of the excavation surface;
(3) testing the GCN prediction model by adopting a test set;
(4) and (4) carrying out prediction precision analysis on the GCN prediction model, and evaluating the GCN prediction model by adopting a mean square error and a root mean square error.
3. The visual safety assessment method for the existing tunnel penetrated by the shield tunnel according to claim 1, wherein the third step specifically comprises the following steps:
through the SHAP model, the black box model of the GCN prediction model is simplified, and the sensitivity of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface to each characteristic is obtained:
Figure 501323DEST_PATH_IMAGE001
wherein the content of the first and second substances,
Figure 548914DEST_PATH_IMAGE002
the number of features that are to be input is indicated,
Figure 202880DEST_PATH_IMAGE003
representation feature
Figure 70342DEST_PATH_IMAGE004
The simplified characteristic value of (a) is,
Figure 592590DEST_PATH_IMAGE005
in order to simplify the reference values of the model,
Figure 42157DEST_PATH_IMAGE006
is characterized in thatiThe value of the SHAP of (1),
Figure 319555DEST_PATH_IMAGE007
is an importance value of a feature;
the characteristicsiThe calculation model of the SHAP value of (1) is as follows:
Figure 80837DEST_PATH_IMAGE008
wherein the content of the first and second substances,
Figure 547722DEST_PATH_IMAGE009
for the output values of the original model, S is a non-0 subset of the features used in the model and N is the set of all input features.
4. The visual safety assessment method for the existing tunnel penetrated by the shield tunnel according to claim 1, wherein in step four, the simulation of uncertainty in soil parameters by using monte carlo comprises: given the probability distribution of soil parameters, generating according to the distributionnAnd (3) grouping samples, wherein the sample mean value obeys normal distribution, and after the Monte Carlo sample mean value is taken as an output predicted value after the uncertainty of the soil parameters is considered, the confidence interval of the mean value can be calculated according to the following formula:
Figure 38746DEST_PATH_IMAGE010
wherein the content of the first and second substances,
Figure 159149DEST_PATH_IMAGE011
is the predicted value mean of the monte carlo samples,
Figure 814252DEST_PATH_IMAGE012
the standard deviation of the predicted values for the monte carlo samples,
Figure 678303DEST_PATH_IMAGE013
is at a significant level of
Figure 23834DEST_PATH_IMAGE014
The value of the standard normal distribution of time,
Figure 721662DEST_PATH_IMAGE015
is the confidence interval.
5. The visual safety assessment method for the shield tunnel to pass through the existing tunnel according to claim 4, wherein in step four, the model uncertainty is quantified by a prediction interval of a regression process, including: taking data of an original data set as
Figure 785433DEST_PATH_IMAGE016
The predicted value of the GCN prediction model is used as
Figure 187596DEST_PATH_IMAGE017
Using the least squares method, a linear regression model of the following format can be trained:
Figure 466262DEST_PATH_IMAGE018
at a given sample capacity
Figure 521942DEST_PATH_IMAGE019
Then, the corresponding prediction interval can be calculated by the following formula:
Figure 479534DEST_PATH_IMAGE020
wherein the content of the first and second substances,
Figure 91912DEST_PATH_IMAGE021
is based on
Figure 84139DEST_PATH_IMAGE022
The predicted value of the regression model is determined,
Figure 45142DEST_PATH_IMAGE023
is composed of
Figure 896554DEST_PATH_IMAGE024
Degree of freedom
Figure 906098DEST_PATH_IMAGE025
The value of t at the significant level is,
Figure 80728DEST_PATH_IMAGE026
in order to be the standard error of the residual error,
Figure 291260DEST_PATH_IMAGE027
and
Figure 223444DEST_PATH_IMAGE028
are respectively as
Figure 364576DEST_PATH_IMAGE029
The mean and the variance of (a) is,PIis a prediction interval.
6. The visual safety assessment method for the existing tunnel penetrated by the shield tunnel according to claim 5, characterized in that the influence of soil and model uncertainty on the result is considered simultaneously in the fourth step, and the upper and lower bounds of the prediction interval are substituted into the calculation of the confidence interval to obtain the new upper and lower bounds of the confidence interval as follows:
Figure 472340DEST_PATH_IMAGE030
wherein the content of the first and second substances,
Figure 447249DEST_PATH_IMAGE031
and
Figure 460205DEST_PATH_IMAGE032
respectively are the average values of the upper and lower bounds of the prediction interval of the Monte Carlo sample,
Figure 952497DEST_PATH_IMAGE033
and
Figure 773822DEST_PATH_IMAGE034
respectively an upper bound and a lower bound of a confidence interval after simultaneously considering the uncertainty of the soil and the model.
7. The visual safety assessment method for the existing tunnel passing through the shield tunnel according to claim 6, characterized in that: in the fourth step, for security risk analysis, the new confidence interval is compared with the retrievable range of the target value, and the percentage of the intersection in the confidence interval is the reliability coefficient, and the operation mode is as follows: if the confidence interval is lower bound
Figure 247529DEST_PATH_IMAGE035
Not more thanUAnd upper bound of confidence interval
Figure 295251DEST_PATH_IMAGE033
Not less thanLThen r =1, otherwise one proceeds as follows: if it is not
Figure 450289DEST_PATH_IMAGE036
And is
Figure 454017DEST_PATH_IMAGE037
Then, then
Figure 177253DEST_PATH_IMAGE038
Otherwise, if
Figure 899221DEST_PATH_IMAGE039
And is
Figure 592371DEST_PATH_IMAGE040
Then, then
Figure 529234DEST_PATH_IMAGE041
Otherwise, if
Figure 282426DEST_PATH_IMAGE042
And is
Figure 226112DEST_PATH_IMAGE043
Then, then
Figure 863898DEST_PATH_IMAGE044
And if not, the step (B),
Figure 514322DEST_PATH_IMAGE045
wherein, in the step (A),
Figure 766312DEST_PATH_IMAGE046
、
Figure 275921DEST_PATH_IMAGE047
the upper and lower bounds of the target value range are,
Figure 310874DEST_PATH_IMAGE048
is a reliability coefficient.
8. The utility model provides a visual safety assessment system of existing tunnel is worn under shield tunnel which characterized in that includes: the first main module is used for constructing a tunnel construction parameter index system, wherein the index system comprises stratum parameters, soil parameters and shield tunneling parameters and acquires the existing data; the second main module is used for constructing a GCN prediction model of accumulated earth surface settlement, existing tunnel deformation and excavation face supporting force based on the GCN regression model and the existing data; the third main module is used for explaining the GCN regression model by adopting an SHAP model to obtain the sensitivity of the accumulated settlement of the earth surface, the deformation of the existing tunnel and the supporting force of the excavation surface to different characteristics; the fourth main module is used for simulating uncertainty in soil parameters by using Monte Carlo and linearly regressing the predicted values and actual values of accumulated settlement of the earth surface, deformation of the existing tunnel and supporting force of an excavation surface to obtain a confidence interval of a prediction result when the uncertainty of the soil and the model is considered at the same time, so that the safety risk of the large-diameter shield in the process of constructing the existing tunnel penetrating through the large-diameter shield is calculated; and the fifth main module is used for visualizing the interaction and the safety risk of the operation process.
9. An electronic device, comprising:
at least one processor, at least one memory, and a communication interface; wherein the content of the first and second substances,
the processor, the memory and the communication interface are communicated with each other;
the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the visual safety assessment method of shield tunneling through existing tunnel according to any one of claims 1 to 7.
10. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the visual security assessment of shield tunneling through an existing tunnel according to any one of claims 1 to 7.
CN202210928018.2A 2022-08-03 2022-08-03 Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel Active CN114996830B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210928018.2A CN114996830B (en) 2022-08-03 2022-08-03 Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210928018.2A CN114996830B (en) 2022-08-03 2022-08-03 Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel

Publications (2)

Publication Number Publication Date
CN114996830A true CN114996830A (en) 2022-09-02
CN114996830B CN114996830B (en) 2022-11-18

Family

ID=83022357

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210928018.2A Active CN114996830B (en) 2022-08-03 2022-08-03 Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel

Country Status (1)

Country Link
CN (1) CN114996830B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116562070A (en) * 2023-03-15 2023-08-08 青岛理工大学 Method, device, equipment and medium for determining priority of shield pile cutting parameters

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120245906A1 (en) * 2011-03-25 2012-09-27 Kabushiki Kaisha Toshiba Monte carlo analysis execution controlling method and monte carlo analysis execution controlling apparatus
CN106296475A (en) * 2016-07-29 2017-01-04 山东大学 Tunnels and underground engineering is dashed forward discharge disaster polymorphic type combining evidences appraisal procedure
CN111985804A (en) * 2020-08-18 2020-11-24 华中科技大学 Shield approaching existing tunnel safety evaluation method based on data mining and data fusion
CN112100574A (en) * 2020-08-21 2020-12-18 西安交通大学 Resampling-based AAKR model uncertainty calculation method and system
CN112149962A (en) * 2020-08-28 2020-12-29 中国地质大学（武汉） Risk quantitative evaluation method and system for cause behavior of construction accident
CN112418683A (en) * 2020-11-26 2021-02-26 华中科技大学 Construction risk evaluation method for shield underpass existing structure
CN112541666A (en) * 2020-12-08 2021-03-23 同济大学 Shield tunnel risk assessment method considering uncertainty of earthquake vulnerability model
AU2021101934A4 (en) * 2021-04-15 2021-06-03 Huazhong University Of Science And Technology Intelligent prediction of shield existing tunnels and multi objective optimization control method of construction parameters based on data drive
US20210382198A1 (en) * 2020-06-03 2021-12-09 Chevron U.S.A. Inc. Uncertainty-aware modeling and decision making for geomechanics workflow using machine learning approaches

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120245906A1 (en) * 2011-03-25 2012-09-27 Kabushiki Kaisha Toshiba Monte carlo analysis execution controlling method and monte carlo analysis execution controlling apparatus
CN106296475A (en) * 2016-07-29 2017-01-04 山东大学 Tunnels and underground engineering is dashed forward discharge disaster polymorphic type combining evidences appraisal procedure
US20210382198A1 (en) * 2020-06-03 2021-12-09 Chevron U.S.A. Inc. Uncertainty-aware modeling and decision making for geomechanics workflow using machine learning approaches
CN111985804A (en) * 2020-08-18 2020-11-24 华中科技大学 Shield approaching existing tunnel safety evaluation method based on data mining and data fusion
CN112100574A (en) * 2020-08-21 2020-12-18 西安交通大学 Resampling-based AAKR model uncertainty calculation method and system
CN112149962A (en) * 2020-08-28 2020-12-29 中国地质大学（武汉） Risk quantitative evaluation method and system for cause behavior of construction accident
CN112418683A (en) * 2020-11-26 2021-02-26 华中科技大学 Construction risk evaluation method for shield underpass existing structure
CN112541666A (en) * 2020-12-08 2021-03-23 同济大学 Shield tunnel risk assessment method considering uncertainty of earthquake vulnerability model
AU2021101934A4 (en) * 2021-04-15 2021-06-03 Huazhong University Of Science And Technology Intelligent prediction of shield existing tunnels and multi objective optimization control method of construction parameters based on data drive

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
LIMAO ZHANG等: "Multi-objective optimization for limiting tunnel-induced damages considering uncertainties", 《RELIABILITY ENGINEERING AND SYSTEM SAFETY》 *
YUE PAN等: "Data-driven multi-output prediction for TBM performance during tunnel excavation: An attention-based graph convolutional network approach", 《AUTOMATION IN CONSTRUCTION》 *
吴贤国等: "基于模糊贝叶斯证据理论的盾构下穿既有隧道安全风险评价", 《隧道建设》 *
李泽荃等: "基于事故树和贝叶斯网络的隧道塌陷风险概率估计方法研究", 《煤炭工程》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116562070A (en) * 2023-03-15 2023-08-08 青岛理工大学 Method, device, equipment and medium for determining priority of shield pile cutting parameters
CN116562070B (en) * 2023-03-15 2023-11-24 青岛理工大学 Method, device, equipment and medium for determining priority of shield pile cutting parameters

Also Published As

Publication number Publication date
CN114996830B (en) 2022-11-18

Similar Documents

Publication Publication Date Title
Yang et al. New pore space characterization method of shale matrix formation by considering organic and inorganic pores
US11295048B2 (en) Machine learning assisted reservoir simulation
Eshkalak et al. Geomechanical properties of unconventional shale reservoirs
CN104769215A (en) System and method for characterizing uncertainty in subterranean reservoir fracture networks
CN105804730B (en) Method and system for resource identification using historical well data
KR101625660B1 (en) Method for making secondary data using observed data in geostatistics
CN114996830B (en) Visual safety assessment method and equipment for shield tunnel to pass through existing tunnel
CN113296152A (en) Fault detection method and device
US20220244424A1 (en) Geological Grid Analysis
Tzu-hao et al. Reservoir uncertainty quantification using probabilistic history matching workflow
Barros et al. Quantitative assessment of monitoring strategies for conformance verification of CO2 storage projects
Lin et al. Probabilistic safety risk assessment in large-diameter tunnel construction using an interactive and explainable tree-based pipeline optimization method
Conde et al. Role of Fluid Injection in Induced Seismicity
CN112036424B (en) Submarine landslide risk analysis method based on unsupervised machine learning
Pan et al. Two-stage support vector machine-enabled deep excavation settlement prediction considering class imbalance and multi-source uncertainties
Armstrong et al. The application of data mining techniques to characterize agricultural soil profiles
CN117035454A (en) Soil pollution repair model training method, system, electronic equipment and medium
US20130132055A1 (en) Method for optimizing the development of an underground medium by means of a reservoir study comprising optimized upscaling
Wang et al. Data-driven analysis of soil consolidation with prefabricated vertical drains considering stratigraphic variation
WO2021240650A1 (en) Pipeline vulnerability estimation system, pipeline vulnerability estimation method, model creation device, and program
CN115238565A (en) Resistivity model reconstruction network training method, electromagnetic inversion method and device
Villegas et al. Simultaneous characterization of geological shapes and permeability distributions in reservoirs using the level set method
CN108875163B (en) Method and system for evaluating three-dimensional fracture network connectivity
WO2021206755A1 (en) Systems and methods for evaluating a simulation model of a hydrocarbon field
Saputra et al. Adapting Fracture Corridors and Diffuse Fractures in Single Porosity Model of Ujung Pangkah Carbonate Reservoir

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant