CN-121996947-A - Tunnel deformation prediction method based on DTW-SSA-DBN and application

CN121996947ACN 121996947 ACN121996947 ACN 121996947ACN-121996947-A

Abstract

Aiming at the problems of low accuracy of a prediction result caused by few influence factors and too few sample data in the conventional tunnel deformation prediction method, the invention provides a tunnel deformation prediction method based on a DTW-SSA-DBN and application thereof, and belongs to the technical field of tunnel deformation monitoring. The method comprises the steps of constructing a tunnel deformation prediction index system based on surrounding rock geological parameters, tunnel design parameters and construction factors, constructing a tunnel deformation prediction model based on a DTW-SSA-DBN based on the tunnel deformation prediction index system, performing model training and verification to obtain a target tunnel deformation prediction model, inputting tunnel deformation prediction index data of a target section to be predicted into the target tunnel deformation prediction model, and outputting tunnel deformation under specified monitoring time. According to the invention, by adding a geological phase change driving force item in the energy function of the DBN network, the abrupt change of the activation mode of the hidden layer neurons can be forced, and the surrounding rock is correspondingly plastically deformed or destroyed, so that the accuracy of a prediction model can be improved.

Inventors

HOU WANGBIN
HU SHIYU
Cai Zeyun
SONG ZHANPING
ZHANG YUWEI
XING PENGTAO
MA YINGJIAN
TIAN WEIZHI
Yang Runpei
Gao Shaochao
LI JIALUN
GAO JUN
ZHAO ZHUOHUI

Assignees

中铁一局集团有限公司
中铁一局集团有限公司第三工程分公司
广西乐望高速公路有限公司
西安建筑科技大学
中铁一局集团电务工程有限公司
中铁一局集团市政环保工程有限公司
中铁一局集团第七工程有限公司

Dates

Publication Date: 20260508
Application Date: 20251226

Claims (10)

1. A method for predicting tunnel deformation based on DTW-SSA-DBN, the method comprising: Constructing a tunnel deformation prediction index system based on surrounding rock geological parameters, tunnel design parameters and construction factors; Constructing a tunnel deformation prediction model based on a DTW-SSA-DBN based on a tunnel deformation prediction index system, and performing model training and verification to obtain a target tunnel deformation prediction model; And inputting the tunnel deformation prediction index data of the target section to be predicted into a target tunnel deformation prediction model, and outputting a tunnel deformation prediction value under the specified monitoring time.
2. The method of claim 1, wherein the tunnel deformation predictor system comprises 9 primary indices of rock saturation uniaxial compressive strength, rock mass integrity index, rock weight, elastic resistance coefficient, elastic modulus, poisson's ratio and internal friction angle, burial depth, and monitoring time.
3. The method of claim 2, wherein the constructing a tunnel deformation prediction model based on the DTW-SSA-DBN based on the tunnel deformation prediction index system, and performing model training and verification to obtain the target tunnel deformation prediction model, comprises: Based on a tunnel deformation prediction index system, acquiring actual monitoring data of each level index in tunnel construction, constructing a tunnel deformation prediction data set, and dividing the tunnel deformation prediction data set into a training set and a testing set; based on the improved DBN model, establishing a tunnel deformation prediction model based on the DTW-SSA-DBN; Inputting training set data into a tunnel deformation prediction model based on a DTW-SSA-DBN, adjusting the super parameters of the DBN network, and outputting a DBN network model after the last parameter adjustment; Performing model evaluation on the DBN network model subjected to the last parameter adjustment on the test set, and outputting a predicted value of the DBN network model; And comparing and verifying the predicted values of different sections with the actual monitored values, and taking the verified model as a target tunnel deformation prediction model.
4. The method of claim 3, wherein the establishing a DTW-SSA-DBN based tunnel deformation prediction model comprises: The dynamic time normalization module is used for classifying the input tunnel deformation prediction data set based on a dynamic time normalization DTW algorithm to obtain a standardized data set with consistent time sequence; The deep confidence network module is used for predicting tunnel deformation by utilizing the improved DBN network based on the standardized data set with consistent time sequence; And the super-parameter optimization module is used for optimizing the super-parameters of the improved DBN model based on the SSA algorithm.
5. The method of claim 4, wherein the improved DBN network is based on an original DBN network model, and is obtained by introducing a geological phase change driving force term into an energy function of a limited boltzmann machine RBM thereof, and modeling surrounding rock deformation mutation to reconstruct the energy function.
6. The method of claim 5, wherein the reconstructed energy function is: ; Wherein, the For the surrounding rock stress gradient corresponding to the jth feature, Is a critical threshold value for the yield of the rock mass, Is the phase-change coupling coefficient, In order for the softening factor to be a good factor, The bias term representing the visible layer is presented, The bias term of the hidden layer is represented, Indicating the state of the visible layer, Indicating the state of the hidden layer(s), Representing the connection weights between the visible and hidden layers, F representing the set of rock mass units.
7. A tunnel deformation prediction apparatus based on DTW-SSA-DBN, characterized in that it is realized based on any one of the methods of claims 1-6, comprising: the index system construction module is used for constructing a tunnel deformation prediction index system based on surrounding rock geological parameters, tunnel design parameters and construction factors; the model building module is used for building a tunnel deformation prediction model based on the DTW-SSA-DBN based on the tunnel deformation prediction index system, and performing model training and verification to obtain a target tunnel deformation prediction model; The prediction module is used for inputting the tunnel deformation prediction index data of the target section to be predicted into the target tunnel deformation prediction model and outputting the tunnel deformation prediction value under the appointed monitoring time.
8. The apparatus of claim 7, wherein the model building module comprises: The data set construction unit is used for acquiring actual monitoring data of each level of index in tunnel construction, constructing a tunnel deformation prediction data set and dividing the tunnel deformation prediction data set into a training set and a testing set; the construction unit is used for establishing a tunnel deformation prediction model based on the DTW-SSA-DBN based on the improved DBN model; The training unit is used for inputting training set data into the tunnel deformation prediction model based on the DTW-SSA-DBN, adjusting the super parameters of the DBN network and outputting the DBN network model after the last parameter adjustment; The evaluation unit is used for carrying out model evaluation on the DBN network model after the last parameter adjustment on the test set and outputting the predicted value of the DBN network model; and the verification unit is used for comparing and verifying the predicted values of the different sections with the actual monitored values, and taking the verified model as a target tunnel deformation prediction model.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory, the processor executing the computer program to implement the method of any of claims 1-6.
10. A readable storage medium, wherein a program or instructions is stored on the readable storage medium, which when executed by a processor, implements the method according to any of claims 1-6.

Description

Tunnel deformation prediction method based on DTW-SSA-DBN and application Technical Field The application relates to the technical field of tunnel deformation monitoring, in particular to a tunnel deformation prediction method based on a DTW-SSA-DBN and application thereof. Background Tunnel deformation is an important index of engineering safety, and accurate prediction of deformation trend is important for guaranteeing construction and operation safety. The tunnel deformation directly reflects the stress state of the structure, and the overlarge deformation can cause the instability of the structure and cause serious safety accidents such as collapse, cracking and the like. Tunnel deformation may cause subsidence of the earth's surface and threaten superstructure safety. In recent years, as the progress of urbanization accelerates and the depth of development of the underground space increases, the importance of tunnel deformation prediction becomes more and more remarkable. The conventional tunnel deformation prediction methods have respective defects, such as an empirical/semi-empirical formula method can only be used under specific geological conditions, and the accuracy is obviously reduced when facing complex and heterogeneous rock masses. The numerical simulation method needs a large amount of accurate rock and soil parameters, and is difficult to obtain reliable input data under complex geological conditions. The prediction method of machine learning can automatically capture the nonlinear relations of different rock and soil conditions, supporting modes, construction procedures and other factors by learning a large amount of historical monitoring data, so that a unified prediction model can be provided under different geological environments, and the limit of an empirical formula which can only aim at specific conditions is broken through. In numerical modeling, parameter uncertainty is a major source of error. The machine learning model can directly take monitoring quantity (displacement, sedimentation, stress and the like) and construction working condition parameters as input, so that strict requirements on accurate numerical values of elastic modulus, friction coefficient and the like are omitted, and better prediction accuracy can be maintained in complex geology with parameters difficult to acquire. However, most of the current methods for predicting tunnel deformation based on the neural network method consider less influencing factors, generally only the monitoring time is used as an input index of a prediction model, but influences of stratum geological factors (such as uniaxial compressive strength, rock mass integrity index and the like), tunnel design (burial depth) and construction factors (monitoring time) on tunnel deformation are ignored, and some researches also have the problem of too few sample data, so that the prediction model obtained by training has limited applicability and poor generalization capability, such as the problem of invalid traffic volume generated by repeated parking seeking in the automobile parking process in the RTCN in-road parking berth prediction model considering regional occupancy, and the RTCN short-time idle berth prediction model considering regional occupancy is proposed. It is pointed out that in the prediction study of parking berth in roads, the problem of too few parking samples in roads exists. Current Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) have drawbacks in the study of short-term predictions of free berths, and the existence of this problem further limits the effectiveness of the prediction model. Because of insufficient sample number, the model is difficult to comprehensively learn the change rules of the idle berth under different time and space. In summary, how to consider multiple factors affecting tunnel deformation, to build a more perfect and comprehensive tunnel deformation prediction model to improve prediction accuracy, and to effectively ensure engineering safety is a current urgent problem to be solved. Disclosure of Invention In view of the above problems, the present application proposes a tunnel deformation prediction method and apparatus based on DTW-SSA-DBN in consideration of stratum geological factors, tunnel design and construction factors, so as to overcome or at least partially solve the above problems. In a first aspect, an embodiment of the present application provides a method for predicting tunnel deformation based on DTW-SSA-DBN, including: Constructing a tunnel deformation prediction index system based on surrounding rock geological parameters, tunnel design parameters and construction factors; Constructing a tunnel deformation prediction model based on a DTW-SSA-DBN based on a tunnel deformation prediction index system, and performing model training and verification to obtain a target tunnel deformation prediction model; And inputting the tunnel deformation prediction index data