CN108985340B - Shield construction ground settlement prediction method based on dual-model fusion - Google Patents
Shield construction ground settlement prediction method based on dual-model fusion Download PDFInfo
- Publication number
- CN108985340B CN108985340B CN201810650131.2A CN201810650131A CN108985340B CN 108985340 B CN108985340 B CN 108985340B CN 201810650131 A CN201810650131 A CN 201810650131A CN 108985340 B CN108985340 B CN 108985340B
- Authority
- CN
- China
- Prior art keywords
- data
- population
- ground settlement
- historical
- parameter
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000010276 construction Methods 0.000 title claims abstract description 29
- 230000004927 fusion Effects 0.000 title claims abstract description 17
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000005641 tunneling Effects 0.000 claims description 76
- 238000009412 basement excavation Methods 0.000 claims description 63
- 238000012544 monitoring process Methods 0.000 claims description 30
- 238000013528 artificial neural network Methods 0.000 claims description 29
- 238000005457 optimization Methods 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 27
- 230000002159 abnormal effect Effects 0.000 claims description 24
- 238000004422 calculation algorithm Methods 0.000 claims description 24
- 230000002068 genetic effect Effects 0.000 claims description 20
- 238000007477 logistic regression Methods 0.000 claims description 20
- 238000012706 support-vector machine Methods 0.000 claims description 15
- 238000005516 engineering process Methods 0.000 claims description 14
- 210000000349 chromosome Anatomy 0.000 claims description 4
- 230000035772 mutation Effects 0.000 claims description 3
- 230000009977 dual effect Effects 0.000 claims 1
- 238000007405 data analysis Methods 0.000 abstract 1
- 230000008569 process Effects 0.000 description 14
- 238000004458 analytical method Methods 0.000 description 13
- 230000006870 function Effects 0.000 description 13
- 238000004088 simulation Methods 0.000 description 13
- 239000002689 soil Substances 0.000 description 13
- 238000011160 research Methods 0.000 description 7
- 230000004913 activation Effects 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 241000512668 Eunectes Species 0.000 description 1
- 229910001374 Invar Inorganic materials 0.000 description 1
- 208000025174 PANDAS Diseases 0.000 description 1
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 description 1
- 240000000220 Panda oleosa Species 0.000 description 1
- 235000016496 Panda oleosa Nutrition 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000009933 burial Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 239000004927 clay Substances 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000246 remedial effect Effects 0.000 description 1
- 238000004062 sedimentation Methods 0.000 description 1
- 239000011800 void material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
- G06F18/2414—Smoothing the distance, e.g. radial basis function networks [RBFN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a shield construction ground settlement prediction method based on big data analysis, which mainly solves the problem of low prediction precision of ground settlement in the shield construction big data scene in the prior art, and the prediction method comprises the following steps: preprocessing data; acquiring parameter data influencing ground settlement; establishing a settlement prediction model; optimizing parameters of the prediction model; retraining the model with the optimal parameters; model fusion; and acquiring real-time ground settlement data. The whole scheme of the invention has rigorous and complete design and high efficiency and accuracy of ground settlement prediction, is used for ground settlement prediction of subway construction, and guarantees the engineering quality and safety.
Description
Technical Field
The invention belongs to the technical field of industrial big data, relates to a ground settlement prediction method, and particularly relates to a ground settlement prediction method based on the fusion of an optimized BP neural network and a support vector regression machine, which can be used for predicting the ground settlement of a plurality of monitoring points in the shield construction process.
Background
Along with the high-speed development of Chinese economy, a great number of people are rushed into each large city, and due to the limited urban planning area, the existing transportation conditions can not meet the demand of people for fast traveling gradually. With the continuous promotion of the urbanization process of China, the urban traffic faces unprecedented challenges due to the large increase of the population quantity of numerous cities, the continuous deterioration of the urban traffic condition causes high attention of governments at all levels, and in the face of increasingly prominent civil problems such as traffic jam, time consumption for trip and the like, the traffic problems faced by the various super-large cities, big cities and the like of all countries in the world are solved by improving the transport capacity of the existing urban rail traffic, the occupational environment of the cities is improved, so that the urban rail traffic needs to be vigorously developed, and therefore the subway construction in China has huge growth potential.
The important point of subway construction lies in the excavation of subway tunnels. In tunnel construction, the shield method is one of the most common ways for subway tunnel construction due to its obvious economic and technical advantages and the characteristic of small influence on the surrounding environment. A plurality of problems are involved in the shield construction and tunneling process, the most important is the construction safety problem, wherein the most important factor influencing the construction safety is the ground settlement in the tunneling process. The ground subsidence is the phenomenon of ground subsidence caused by the combined action of soil loss caused by tunnel excavation and remolding soil reconsolidation of the shield body to the disturbance of the surrounding soil layer in the tunneling process of the shield machine, and if the control is improper or remedial measures cannot be taken in advance, great threats can be caused to roads, buildings, structures, underground oil and gas pipelines and operators around the construction line. The influence of construction on surrounding soil bodies is reduced to the maximum extent, the influence of construction on adjacent buildings and pipelines is reduced, the surface subsidence caused by construction is reasonably controlled, and the timely, accurate and intelligent prediction of the ground subsidence amount has great engineering significance. The prediction method mainly comprises the following steps of: a Peck formula method, a numerical simulation method, a finite element analysis method, an artificial intelligence model prediction method and the like.
In terms of Peck formula, in 1969, Peck researches a large number of subway tunneling engineering examples, and proposes a famous Peck formula through analysis of surface subsidence monitoring data. Under the assumption of no drainage, Peck considers that the volume of the settling tank is approximately equal to the formation loss amount in the tunneling process, and the curve shape of the settling tank in the cross section of the tunnel is like a normal distribution curve, so that the ground settlement is predicted according to the formula. At present, the Peck formula in a plurality of shield construction projects still has important practical significance in the aspect of ground settlement prediction. Based on the Peck formula, Chenchun Lai, Zhao City Li and other people research a ground settlement prediction method in a double-line shield tunneling mode, establish and correct a three-dimensional Peck formula, and propose a method for calculating the total ground settlement by superposing earlier settlement and later settlement. Clough and Schmidt et al performed a study summary of the settler width coefficients. A great deal of research is also carried out on the calculation of the width coefficient of the settling tank by O' Reilly and New, and the results show that the tunnel axis is buried deep and has a linear relation, so that the conclusion lays a foundation for the subsequent numerous correction methods. Celestino et al recognized the deficiencies of the Peck formula in ground subsidence prediction by collating measured data and analyzing results of finite element calculations. The main difference in recognition is the problem of the settler shape, which they believe no longer presents a normal distribution shape under certain construction conditions, but a shape similar to a "plug". Meanwhile, a settlement curve formula is proposed based on the conclusion.
In the aspect of numerical simulation, Wulonghai researches and summarizes ground settlement and deformation rules, and popularizes the rules to collapsible loess areas based on numerical simulation and on-site monitoring data. And then, the stratum soil loss rate and the width coefficient of the settling tank are corrected, the core idea is the application of linear regression and the least square method in solving the relevant parameters of the Peck formula, and the article practice proves that the method has better performance under specific geological conditions. Zhang Dongmei, Huang hongwei and the like fully analyze the influence factors influencing the long-term sedimentation of the ground, construct a soft clay timeliness constitutive model on the basis, and simulate the local seepage phenomenon of the lining. And finally, simulating the ground long-term settlement under the soft stratum condition by adopting a numerical simulation method. Liu hong zhou, grand jun et al first introduced the nature of soft soil stratum and the influence to ground subsidence, then selected appropriate soil mechanics and material parameters, and based on this determined the finite element model and optimized the net.
In the aspect of finite element analysis, Rowe et al simulate the displacement caused by excavation in the tunneling process by using a three-dimensional elasto-plastic finite element analysis model, and calculate the soil loss. Karakus, Fowell and the like simulate the states and influences of tunnel excavation faces under different excavation modes by means of a finite element analysis modeling method. Izumi and Norrish introduce the application of the finite element method in shield construction in soft soil regions, and propose that the parameter estimation is difficult due to the soil characteristics and the like, and the simulation of tunnel engineering by using the finite element method is difficult. Shi, Ortigao and the like calculate the soil loss in the tunneling process, and simulate the ground settlement caused by shield construction by using a three-dimensional elasto-plastic finite element analysis method. Zhang Hai ripples, invar and the like develop three-dimensional nonlinear finite element analysis research on ground settlement caused by an earth pressure balance shield machine, deeply research on the influence of tunneling parameters on the ground settlement in the tunneling process, and finally, the influence of various influencing factors on the ground settlement is integrated by utilizing a mathematical expression form. It is worth noting that in terms of finite element Analysis, FLAC (fast Lagrange Analysis of Continua), a continuous medium body rapid Lagrangian Analysis software, is most widely used. The software is developed by ITASCA company in America, and is often used in mechanical analysis of geotechnical and geological engineering as a finite element difference method calculation program.
In the aspect of artificial intelligence models, Kim, Bae and the like indicate that the ground settlement is predicted by using an empirical formula or a semi-empirical formula, and all related factors cannot be considered at the same time, so that the prediction accuracy needs to be improved. In order to solve the problems, the method for capturing the implicit association relationship in the shield construction data set by using the capability of artificial neural network pattern recognition and memory is provided. Besides, abnormal values in the tunneling data are processed, and a prediction model is optimized. Pourtaghi, Lotfollahi-Yaghin and the like propose a wavelet network (wavenet) aiming at the ground settlement prediction problem based on a wavelet theory and an artificial neural network theory, wherein the wavenet is a single-hidden-layer feed-forward network and is characterized in that a network activation function is a wavelet function. The experimental result shows that the fitting capability and the generalization capability of the neural network model with the wavelet function as the activation function are stronger than those of the traditional neural network. The artificial neural network is firstly researched for predicting ground settlement in China by grand jun, Yuanjinrong and the like, and the prediction result is compared with the field actual measurement result, so that the finding effect is better. Shi, Ortigao, etc. analyze the association between the ground subsidence and the shield tunneling parameters, and obtain the input of a BP neural network training model based on the analysis result: the diameter of the shield machine, the buried depth of the shield, the tunneling speed, the mechanical parameters of the soil body and the like. The method is high in forever strength, the artificial neural network algorithm is adopted for building the ground settlement prediction model in the tunneling process by people who are in the sea and the like, and the effectiveness of the algorithm is verified through examples. In the process of constructing the ground settlement prediction model, georgia, fangyangli and the like not only remove abnormal values of input data and ensure the data quality, but also optimize the step length in the process of training the BP neural network model and construct the settlement prediction model by adopting a variable step length method. Practice proves that the method can effectively solve the problem that the BP neural network algorithm is easy to converge to a local optimal solution. Aiming at the problems that a BP neural network model is long in training time, slow in training speed, easy to converge to a local minimum point and the like, Li Hongxian, Zhao Xinhua and the like adopt a Genetic Algorithm (GA) to optimize parameters of the BP neural network, a ground subsidence prediction model of the GA + BP neural network is constructed, and the fact proves that the optimization strategy of the model has obvious positive effects. A ground settlement prediction model is constructed by adopting a Support Vector Regression (SVR) in the Wangyuan, a relative Error and a Mean Square Error (MSE) are adopted to evaluate the prediction model, and meanwhile, the model performances under different parameters are compared.
Summarizing the above current research situation, it is found that the artificial intelligence method, especially the BP neural network, is widely applied to ground subsidence prediction and obtains a better prediction result because the artificial intelligence method has good data understanding and mining capabilities. For example, xiebao 29710, et al, filed 2016 for a patent entitled "analysis method of ground subsidence caused by subway excavation based on neural network", with application publication No. CN 106021717 a and publication date 2016.10.12, and disclosed a method for analyzing ground subsidence caused by subway excavation based on neural network, which applies the neural network to take geotechnical parameter data of each existing subway as input values, trains by taking ground subsidence data at detection points as output values at different positions of the horizontal distance between a tunneling surface and a monitoring point, and analyzes the subsidence condition of the ground above other subways to be constructed by using the trained network. When parameter data influencing ground settlement is obtained, the ground settlement is mainly related to elastic modulus, bulk density, Poisson's ratio, cohesive force, internal friction angle and 5 stratum parameter data according to a geotechnical theory, the 5 stratum parameter data and the relative distance parameter data are used as the parameter data influencing the ground settlement, but only the influence of geological parameters on the ground settlement is considered, and the tunneling parameters influencing the ground settlement are not considered, so that the reliability of the established ground settlement prediction model is not high, when the ground prediction model is established, the difference of parameter values of the ground settlement prediction model is not considered, so that errors can occur in the prediction result, and only the condition of a single prediction model is considered, so that the condition that the parameter data are not suitable for the prediction model can occur.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, provides a shield construction ground settlement prediction method based on dual-model fusion, and aims to solve the technical problem of low prediction precision in the prior art.
In order to achieve the purpose, the technical scheme adopted by the invention comprises the following steps:
1. a shield construction ground settlement prediction method based on double model fusion is characterized by comprising the following steps:
(1) preprocessing historical shield tunneling parameter data, geological parameter data and ground settlement data of a plurality of monitoring points:
filling null values in historically collected geological parameter data and tunneling parameter data, removing abnormal values in the obtained complete historical geological parameter data and tunneling parameter data, and then normalizing to obtain normalized historical geological parameter data and normalized historical tunneling parameter data; meanwhile, after null value filling is carried out on the ground subsidence data of a plurality of historically collected monitoring points, abnormal values are removed, wherein the abnormal values comprise historical subsidence data at a position which is-20 meters away from an excavation surface, historical subsidence data at a position which is-15 meters away from the excavation surface, historical subsidence data at a position which is-10 meters away from the excavation surface, historical subsidence data at a position which is-5 meters away from the excavation surface, historical subsidence data at a position which is 0 meters away from the excavation surface, historical subsidence data at a position which is 5 meters away from the excavation surface and historical subsidence data at a position which is 10 meters away from the excavation surface, and the historical ground subsidence data of the plurality of preprocessed monitoring points are obtained;
(2) acquiring parameter data influencing ground settlement;
taking normalized historical geological parameter data and normalized historical tunneling parameter data as input of Logistic regression, taking historical ground settlement data of a plurality of pretreated monitoring points as output of Logistic regression, constructing a ground settlement prediction model based on Logistic regression, outputting coefficient weights of all parameters from the built ground settlement prediction model of Logistic regression, wherein the coefficient weights are weights of the normalized historical geological parameter data and the normalized historical tunneling parameter data, setting a threshold value for extracting the weight of the parameter data influencing ground settlement as delta, and taking the normalized historical geological parameter data and the normalized historical tunneling parameter data with the weights larger than delta as the parameter data influencing ground settlement;
(3) establishing a ground subsidence prediction model based on a BP neural network and a support vector machine and training the ground subsidence prediction model:
taking the normalized historical geological parameter data and the normalized historical tunneling parameter data as the input of a BP (Back propagation) neural network, taking the preprocessed historical ground settlement data of a plurality of monitoring points as the output of the BP neural network, constructing a ground settlement prediction model based on the BP neural network, and training the ground settlement prediction model by utilizing the parameter data influencing ground settlement and the preprocessed historical ground settlement data to obtain a ground settlement prediction model M1(ii) a Meanwhile, normalized historical geological parameter data and normalized historical tunneling parameter data affecting ground settlement are used as input of a support vector machine, preprocessed ground settlement data of a plurality of monitoring points are used as output of the support vector machine, a ground settlement prediction model based on the support vector machine is constructed, the ground settlement prediction model is trained by using the parameter data affecting ground settlement and the preprocessed historical ground settlement data, and a ground settlement prediction model M is obtained2；
(4) To M1Initial weight and threshold ofOptimizing the value:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution1Optimizing the initial weight and the threshold value to obtain the optimal initial weight wbAnd an optimum threshold value thetab；
(5) To M1Training is carried out:
the optimal initial weight value w is obtainedbOptimal threshold value thetabSubstituting parameter data influencing ground settlement and preprocessed historical ground settlement data into model M1In, to M1Training to obtain a ground settlement prediction model M1 *；
(6) To M2The Gama parameter and the Sigma parameter in (1) are optimized:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution2The Gama parameter and the Sigma parameter in the system are optimized to obtain an optimal Gama parameter value and an optimal Sigma parameter value;
(7) to M2Training is carried out:
the optimal Gama parameter value, the optimal Sigma parameter value, the parameter data influencing the ground settlement and the preprocessed historical ground settlement data are brought into the model M2In, to M2Training to obtain a ground settlement prediction model M2 *；
(8) Model M for predicting ground settlement1 *Model M for predicting ground settlement2 *Carrying out fusion:
setting the weight of model fusion as alphaiTraversing all weight combinations to obtain the optimal weight combination alpha1_bestAnd alpha2_bestAnd use of alpha1_bestAnd alpha2_bestModel M for predicting ground settlement1 *Model M for predicting ground settlement2 *Carrying out weighted linear combination to obtain a fused model O*，
(9) Acquiring real-time ground settlement data:
input model O for acquiring real-time geological parameter data and tunneling parameter data*And outputting the ground settlement data of the plurality of monitoring points, wherein the ground settlement data respectively comprise settlement data at a position which is-20 meters away from the excavation surface, settlement data at a position which is-15 meters away from the excavation surface, settlement data at a position which is-10 meters away from the excavation surface, settlement data at a position which is-5 meters away from the excavation surface, settlement data at a position which is 0 meters away from the excavation surface, settlement data at a position which is 5 meters away from the excavation surface and settlement data at a position which is 10 meters away from the excavation surface.
Compared with the prior art, the invention has the following advantages:
1. according to the method, the ground settlement prediction model based on the BP neural network is optimized, the ground prediction model based on the support vector machine is optimized, and then fusion is carried out to obtain the fused ground settlement prediction model, so that the situation that parameter data is not suitable for the prediction model due to the fact that parameter values of the models in the prior art are different and the prediction result error is large is avoided, and the situation that the parameter data is not suitable for the prediction model is only considered in the prior art, and the prediction accuracy is improved.
2. According to the method, the normalized historical geological parameter data and the normalized historical tunneling parameter data are used as the input of the Logistic regression through the Logistic regression algorithm, the preprocessed historical ground settlement data of the monitoring points are used as the output of the Logistic regression, a ground settlement prediction model based on the Logistic regression is constructed, the parameter data influencing the ground settlement is obtained according to the ground settlement prediction model based on the Logistic regression, the problem that the reliability of the trained model is low due to the fact that the prior art cannot obtain all the parameter data influencing the ground settlement is solved, and the prediction accuracy is further improved.
3. According to the method, the settlement of a plurality of important monitoring points is predicted simultaneously by utilizing the tunneling parameter data and the geological parameter data which are collected in real time, and the predicted settlement data of the plurality of monitoring points are compared and analyzed simultaneously, so that the problem that errors can occur when the predicted settlement data exceeds the danger limit value in the construction process and the tunneling parameters are adjusted due to the fact that the settlement data of a single monitoring point is only considered to be predicted in the prior art is avoided.
Drawings
FIG. 1 is a flow chart of an implementation of the present invention.
Detailed description of the invention
The invention is described in further detail below with reference to the figures and the specific embodiments.
Referring to fig. 1, the implementation steps of the invention are as follows:
step 1) preprocessing historical shield tunneling parameter data, geological parameter data and ground settlement data of a plurality of monitoring points:
filling null values in historically collected geological parameter data and tunneling parameter data, removing abnormal values in the obtained complete historical geological parameter data and tunneling parameter data, and then normalizing to obtain normalized historical geological parameter data and normalized historical tunneling parameter data; meanwhile, after null value filling is carried out on the ground subsidence data of a plurality of historically collected monitoring points, abnormal values are removed, wherein the abnormal values comprise historical subsidence data at a position which is-20 meters away from an excavation surface, historical subsidence data at a position which is-15 meters away from the excavation surface, historical subsidence data at a position which is-10 meters away from the excavation surface, historical subsidence data at a position which is-5 meters away from the excavation surface, historical subsidence data at a position which is 0 meters away from the excavation surface, historical subsidence data at a position which is 5 meters away from the excavation surface and historical subsidence data at a position which is 10 meters away from the excavation surface, and the historical ground subsidence data of the plurality of preprocessed monitoring points are obtained;
the method comprises the following specific steps of preprocessing historical shield tunneling parameter data, geological parameter data and ground settlement data of a plurality of monitoring points:
step 1a) judging whether missing values exist in tunneling parameter data, geological parameter data and ground settlement data or not, if so, calling a fillna function in a pandas module by utilizing a python programming language to perform average filling on the tunneling parameter data, the geological parameter data and the ground settlement data with the missing values to obtain complete tunneling parameter data, geological parameter data and ground settlement data;
step 1b) judging whether p (x-mu is more than 3 sigma) is less than or equal to 0.003, if so, taking an observed value x in the complete tunneling parameter data, geological parameter data and ground settlement data as an abnormal value of the tunneling parameter data, and executing the step (1c), otherwise, executing the step (1d) if no abnormal value exists in the complete tunneling parameter data, geological parameter data and ground settlement data, wherein x is an observed value in the complete tunneling parameter data, geological parameter data and ground settlement data, mu is an average value of each parameter data in the complete tunneling parameter data, geological parameter data and ground settlement data, sigma is a standard deviation of each parameter data in the complete tunneling parameter data, geological parameter data and ground settlement data, and p is a probability that a difference value between the observed value x and the average value mu exceeds 3 times of the standard deviation sigma;
step 1c), replacing abnormal values of corresponding parameter data by using average values of all parameter data in complete tunneling parameter data, geological parameter data and ground settlement data to obtain complete tunneling parameter data, geological parameter data and preprocessed ground settlement data without abnormal values;
step 1d) Using the formula x*＝(x-xmin)/(xmax-xmin) Normalizing the complete tunneling parameter data and geological parameter data without abnormal values to obtain normalized historical geological parameter data and normalized historical tunneling parameter data, wherein x is*Is a normalized data set, x is complete tunneling parameter data and geological parameter data without abnormal values, and x isminMinimum value, x, of complete tunneling parameter data and geological parameter data without abnormal valuemaxThe maximum value of the complete tunneling parameter data and geological parameter data without abnormal values is obtained.
Step 2) acquiring parameter data influencing ground settlement;
taking normalized historical geological parameter data and normalized historical tunneling parameter data as input of Logistic regression, taking historical ground settlement data of a plurality of pretreated monitoring points as output of Logistic regression, constructing a ground settlement prediction model based on Logistic regression, outputting coefficient weights of all parameters from the built ground settlement prediction model of Logistic regression, wherein the coefficient weights are weights of the normalized historical geological parameter data and the normalized historical tunneling parameter data, setting a threshold value for extracting the weight of the parameter data influencing ground settlement as delta, and taking the normalized historical geological parameter data and the normalized historical tunneling parameter data with the weights larger than delta as the parameter data influencing ground settlement;
step 3) establishing a ground subsidence prediction model based on a BP neural network and a support vector machine and training the ground subsidence prediction model:
taking the normalized historical geological parameter data and the normalized historical tunneling parameter data as the input of a BP (Back propagation) neural network, taking the preprocessed historical ground settlement data of a plurality of monitoring points as the output of the BP neural network, constructing a ground settlement prediction model based on the BP neural network, and training the ground settlement prediction model by utilizing the parameter data influencing ground settlement and the preprocessed historical ground settlement data to obtain a ground settlement prediction model M1(ii) a Meanwhile, normalized historical geological parameter data and normalized historical tunneling parameter data affecting ground settlement are used as input of a support vector machine, preprocessed ground settlement data of a plurality of monitoring points are used as output of the support vector machine, a ground settlement prediction model based on the support vector machine is constructed, the ground settlement prediction model is trained by using the parameter data affecting ground settlement and the preprocessed historical ground settlement data, and a ground settlement prediction model M is obtained2；
Step 4) for M1Optimizing the initial weight and the threshold value:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution1Optimizing the initial weight and the threshold value to obtain the optimal initial weight wbAnd an optimum threshold value thetab；
Genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution1The specific implementation steps for optimizing the initial weight and the threshold value are as follows:
step 4a) self-defining wbPopulation sum thetabPopulation size, crossover probability of genetic algorithm of heuristic search and optimization technique of natural evolutionVariation probability and optimization criteria;
step 4b) according to the user-defined wbPopulation sum thetabSize of the population, randomly generating an initial w comprising a plurality of individualsbPopulation and initial thetabA population, each individual representing a genotype of a chromosome;
step 4c) calculating the initial wbPopulation and initial thetabThe fitness value of each individual in the population is judged, whether the fitness value meets the optimization criterion is judged, and if yes, the initial w is obtainedbPopulation and initial thetabBest individual in population and optimal initial weight w represented by the best individualbAnd an optimum threshold value thetabAnd outputs the optimal initial weight wbAnd an optimum threshold value thetab(ii) a Otherwise, will wbIndividual composition w of population not satisfying fitnessb1Group of thetabThe composition theta of individuals in the population that do not meet the fitnessb1Population and performing the step (4 d);
step 4d) according to the cross probability and the mutation probability, in wb1Crossing and varying individuals in the population to obtain offspring wb2Group of, at the same time as thetab1Crossing and varying individuals in the population to obtain offspring thetab2Population;
step 4e) of converting wb2Population as initial wbGroup of thetab2Population as initial thetabPopulation, execute (4 c).
Step 5) for M1Training is carried out:
the optimal initial weight value w is obtainedbOptimal threshold value thetabSubstituting parameter data influencing ground settlement and preprocessed historical ground settlement data into model M1In, to M1Training to obtain a ground settlement prediction model M1 *；
Step 6) for M2The Gama parameter and the Sigma parameter in (1) are optimized:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution2The Gama parameter and the Sigma parameter in the system are optimized to obtain the optimal Gama parameter value and the optimal Sigma parameterA value;
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution2The specific steps for optimizing the Gama parameter and the Sigma parameter in the method are as follows:
step 6a) self-defining the sizes of the Gama population and the Sigma population, and cross probability, variation probability and optimization criteria of a genetic algorithm of a heuristic search and optimization technology of natural evolution;
step 6b) randomly generating an initial Gama population and an initial Sigma population containing a plurality of individuals according to the sizes of the customized Gama population and Sigma population, wherein each individual represents the genotype of the chromosome;
step 6c) calculating the fitness value of each individual in the initial Gama population and the initial Sigma population, judging whether the fitness value meets the optimization criterion, if so, obtaining the best individual in the initial Gama population and the initial Sigma population, the optimal Gama parameter value and the optimal Sigma parameter value which are represented by the best individual, and outputting the optimal Gama parameter value and the optimal Sigma parameter value; otherwise, the individuals which do not meet the fitness in the Gama population are combined into Gama1The individuals which do not meet the fitness in the Sigma population are combined into the Sigma1Population and performing step (6 d);
step 6d) according to the cross probability and the variation probability, in Gama1Crossing and varying individuals in the population to obtain offspring Gama2Population while in Sigma1Crossing and varying individuals in the population to obtain offspring Sigma2Population;
step 6e) Gama2The population was used as the initial Gama population, Sigma2Population as initial Sigma population, perform (6 c);
step 7) for M2Training is carried out:
the optimal Gama parameter value, the optimal Sigma parameter value, the parameter data influencing the ground settlement and the preprocessed historical ground settlement data are brought into the model M2In, to M2Training to obtain a ground settlement prediction model M2 *；
Step 8) predicting model M for ground settlement1 *To the groundSurface subsidence prediction model M2 *Carrying out fusion:
setting the weight of model fusion as alphaiTraversing all weight combinations to obtain the optimal weight combination alpha1_bestAnd alpha2_bestAnd use of alpha1_bestAnd alpha2_bestModel M for predicting ground settlement1 *Model M for predicting ground settlement2 *Carrying out weighted linear combination to obtain a fused model O*，
Step 9), acquiring real-time ground settlement data:
input model O for acquiring real-time geological parameter data and tunneling parameter data*And outputting the ground settlement data of the plurality of monitoring points, wherein the ground settlement data respectively comprise settlement data at a position which is-20 meters away from the excavation surface, settlement data at a position which is-15 meters away from the excavation surface, settlement data at a position which is-10 meters away from the excavation surface, settlement data at a position which is-5 meters away from the excavation surface, settlement data at a position which is 0 meters away from the excavation surface, settlement data at a position which is 5 meters away from the excavation surface and settlement data at a position which is 10 meters away from the excavation surface.
The effect of the present invention is further illustrated by the following simulation experiments:
1. simulation conditions are as follows:
the data simulation experiment is carried out under the Intel (R) core (TM)2Duo of the main frequency 2.4GHZ, the hardware environment of the memory 4GB and the Anaconda spyder software environment;
the data adopted in the experiment are the real data of Ningbo subway gym-Ming building segment shield tunneling machine tunneling, 35 characteristics are provided in total, and 600 data records are provided in total;
the tunneling parameter data comprises a ring number, a target mileage, an inner ring temperature, an outer ring temperature, an average grouting amount, a screw machine rotating speed, an average grouting pressure, a shield tunneling machine burial depth, a hinge horizontal deviation, a hinge vertical deviation, a notch horizontal deviation, a notch vertical deviation, a shield tail horizontal deviation, a shield tail vertical deviation, a shield tail gap upper part, a shield tail gap lower part, a shield tail gap left part, a shield tail gap right part, an earth bunker average pressure, a jack average thrust force, a cutter head rotating speed, a cutter head torque and a tunneling speed;
the geological parameters comprise the specific gravity, cohesive force, internal friction angle, void ratio and compression modulus of soil;
the ground settlement data comprises settlement at a position which is 20 meters away from the excavation surface, settlement at a position which is 15 meters away from the excavation surface, settlement at a position which is 10 meters away from the excavation surface, settlement at a position which is 5 meters away from the excavation surface, settlement at a position which is 0 meters away from the excavation surface, settlement at a position which is 5 meters away from the excavation surface and settlement at a position which is 10 meters away from the excavation surface;
2. simulation content:
the prediction method of the ground settlement amount in the shield construction process based on the dual-model fusion is adopted to carry out prediction simulation on the real data of Ningbo subway stadium-Ming building shield tunneling machine, and the simulation result is shown in table 2.1.
The specific simulation steps are as follows:
1): data preprocessing:
the method comprises the steps of firstly filling missing values by using a mean value filling method, detecting abnormal values by using a 3 sigma method, replacing the abnormal values by using the average number of the columns of the abnormal values, and finally carrying out normalization processing to normalize the abnormal values to a [0,1] interval to obtain preprocessed experimental data, wherein the original data may have the missing values and abnormal value data caused by shield startup and shutdown, shield faults, shield driver blind operation and the like. The experimental data are shown in table 1 below:
TABLE 1 Experimental data
2): acquiring parameter data characteristics influencing ground settlement:
taking normalized historical geological parameter data and normalized historical tunneling parameter data as input of Logistic regression, taking preprocessed historical ground subsidence data as output of Logistic regression, constructing a ground subsidence prediction model based on Logistic regression, extracting a regression coefficient of each feature vector as weights of the normalized historical geological parameter data and the normalized historical tunneling parameter data, setting a threshold value for extracting the feature weight of the parameter data influencing ground subsidence as 1, and taking the normalized historical geological parameter data and the normalized historical tunneling parameter data with the feature weight larger than 1 as the parameter data features influencing ground subsidence to obtain parameter mathematical features as follows:
TABLE 2 Parametric data characterization
Step 3): establishing a ground subsidence prediction model based on a BP neural network and a support vector machine and training the ground subsidence prediction model:
establishing a ground subsidence prediction model based on a BP (back propagation) neural network, wherein the network structure of the BP neural network model is [28,57,1], namely the number of input layer nodes is 28, the number of single hidden layer nodes is 57, the number of output layer nodes is 1, the learning rate is 0.01, a hidden layer activation function is a tansig function, and an output layer activation function is a sigmoid function; establishing a ground settlement prediction model based on a support vector machine, wherein a Radial Basis Function (RBF) Function is selected as a kernel Function, and a kernel Function parameter gamma and a penalty factor C are respectively 0.1 and 1;
step 4): to M1Optimizing the initial weight and the threshold value:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution1The initial weight and the threshold value are optimized, the population size in GA parameters of a genetic algorithm is set to be 50, the maximum iteration number is 100, the crossing rate is 0.8, the variation rate is 0.0005, the optimization criterion is that the individual fitness is 0.001, and the optimal initial weight is 0.5 and the optimal threshold value is 0.2;
step 5): to M1Training is carried out:
the optimal initial weight value 0.5, the optimal threshold value 0.2, the parameter data influencing the ground settlement and the preprocessed historical ground settlement data are brought into the model M1In, to M1Training to obtain a ground settlement prediction model M1 *；
Step 6): to M2The Gama parameter and the Sigma parameter in (1) are optimized:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution2The Gama parameter and the Sigma parameter in the genetic algorithm are optimized, the population size in the GA parameter of the genetic algorithm is set to be 50, the maximum iteration number is 100, the cross rate is 0.8, the variation rate is 0.0005, and the optimization criterion is that the individual fitness is 0.001, so that the optimal Gama parameter value 10 and the optimal Sigma parameter value 1 are obtained;
step 7): to M1Training is carried out:
the optimal Gama parameter value 10, the optimal Sigma parameter value 1, the parameter data influencing the ground settlement and the preprocessed historical ground settlement data are brought into the model M2In, to M2Training to obtain a ground settlement prediction model M1 *；
Step 8): adopting a grid search method to violently search the optimal weight combination of the combined model, setting the weight step length in the grid search method to be 0.05, and carrying out prediction on the ground settlement model M1 *Model M for predicting ground settlement2 *Carrying out weighted linear combination to obtain a fused model O*，
Step 9): and inputting the geological parameter data and the tunneling parameter data acquired in real time into the fused prediction model, and predicting the ground settlement by using the extracted key parameter data characteristics.
3. And (3) simulation result analysis:
TABLE 2.1
As can be seen from the table 2.1, the prediction relative error of the subsidence data at the position 20 meters away from the excavation surface is 1.53%, the prediction relative error of the subsidence data at the position 15 meters away from the excavation surface is 1.84%, the prediction relative error of the subsidence data at the position 10 meters away from the excavation surface is 1.89%, the prediction relative error of the subsidence data at the position 5 meters away from the excavation surface is 1.95%, the prediction relative error of the subsidence data at the position 0 meters away from the excavation surface is 2.12%, the prediction relative error of the subsidence data at the position 5 meters away from the excavation surface is 1.57%, the prediction relative error of the subsidence data at the position 10 meters away from the excavation surface is 1.71%, and the prediction accuracy of the ground subsidence data is more than 95%, so the prediction accuracy of the invention is higher.
Claims (3)
1. A shield construction ground settlement prediction method based on double model fusion is characterized by comprising the following steps:
(1) preprocessing historical shield tunneling parameter data, geological parameter data and ground settlement data of a plurality of monitoring points:
filling null values in historically collected geological parameter data and tunneling parameter data, removing abnormal values in the obtained complete historical geological parameter data and tunneling parameter data, and then normalizing to obtain normalized historical geological parameter data and normalized historical tunneling parameter data; meanwhile, after null value filling is carried out on the ground subsidence data of a plurality of historically collected monitoring points, abnormal values are removed, wherein the abnormal values comprise historical subsidence data at a position which is-20 meters away from an excavation surface, historical subsidence data at a position which is-15 meters away from the excavation surface, historical subsidence data at a position which is-10 meters away from the excavation surface, historical subsidence data at a position which is-5 meters away from the excavation surface, historical subsidence data at a position which is 0 meters away from the excavation surface, historical subsidence data at a position which is 5 meters away from the excavation surface and historical subsidence data at a position which is 10 meters away from the excavation surface, and the historical ground subsidence data of the plurality of preprocessed monitoring points are obtained;
(2) acquiring parameter data influencing ground settlement;
taking normalized historical geological parameter data and normalized historical tunneling parameter data as input of Logistic regression, taking historical ground settlement data of a plurality of pretreated monitoring points as output of Logistic regression, constructing a ground settlement prediction model based on Logistic regression, outputting coefficient weights of all parameters from the built ground settlement prediction model of Logistic regression, wherein the coefficient weights are weights of the normalized historical geological parameter data and the normalized historical tunneling parameter data, setting a threshold value for extracting the weight of the parameter data influencing ground settlement as delta, and taking the normalized historical geological parameter data and the normalized historical tunneling parameter data with the weights larger than delta as the parameter data influencing ground settlement;
(3) establishing a ground subsidence prediction model based on a BP neural network and a support vector machine and training the ground subsidence prediction model:
taking the normalized historical geological parameter data and the normalized historical tunneling parameter data as the input of a BP (Back propagation) neural network, taking the preprocessed historical ground settlement data of a plurality of monitoring points as the output of the BP neural network, constructing a ground settlement prediction model based on the BP neural network, and training the ground settlement prediction model by utilizing the parameter data influencing ground settlement and the preprocessed historical ground settlement data to obtain a ground settlement prediction model M1(ii) a Meanwhile, normalized historical geological parameter data and normalized historical tunneling parameter data affecting ground settlement are used as input of a support vector machine, preprocessed ground settlement data of a plurality of monitoring points are used as output of the support vector machine, a ground settlement prediction model based on the support vector machine is constructed, the ground settlement prediction model is trained by using the parameter data affecting ground settlement and the preprocessed historical ground settlement data, and a ground settlement prediction model M is obtained2；
(4) To M1Optimizing the initial weight and the threshold value:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution1Optimizing the initial weight and the threshold value to obtain the optimal initial weight wbAnd an optimum threshold value thetab；
(5) To M1Training is carried out:
the optimal initial weight value w is obtainedbOptimal threshold value thetabSubstituting parameter data influencing ground settlement and preprocessed historical ground settlement data into model M1In, to M1Training to obtain a ground settlement prediction model M1 *；
(6) To M2The Gama parameter and the Sigma parameter in (1) are optimized:
genetic algorithm pair M adopting heuristic search and optimization technology based on natural evolution2The Gama parameter and the Sigma parameter in the system are optimized to obtain an optimal Gama parameter value and an optimal Sigma parameter value;
(7) to M2Training is carried out:
the optimal Gama parameter value, the optimal Sigma parameter value, the parameter data influencing the ground settlement and the preprocessed historical ground settlement data are brought into the model M2In, to M2Training to obtain a ground settlement prediction model M2 *；
(8) Model M for predicting ground settlement1 *Model M for predicting ground settlement2 *Carrying out fusion:
setting the weight of model fusion as alphaiTraversing all weight combinations to obtain the optimal weight combination alpha1_bestAnd alpha2_bestAnd use of alpha1_bestAnd alpha2_bestModel M for predicting ground settlement1 *Model M for predicting ground settlement2 *Carrying out weighted linear combination to obtain a fused model O*，
(9) Acquiring real-time ground settlement data:
input model O for acquiring real-time geological parameter data and tunneling parameter data*And outputting the ground settlement data of the plurality of monitoring points, wherein the ground settlement data respectively comprise settlement data at a position which is-20 meters away from the excavation surface, settlement data at a position which is-15 meters away from the excavation surface, settlement data at a position which is-10 meters away from the excavation surface, settlement data at a position which is-5 meters away from the excavation surface, settlement data at a position which is 0 meters away from the excavation surface, settlement data at a position which is 5 meters away from the excavation surface and settlement data at a position which is 10 meters away from the excavation surface.
2. The method for predicting ground settlement in shield construction based on dual-model fusion as claimed in claim 1, wherein the genetic algorithm pair M in step (4) using heuristic search and optimization technique based on natural evolution1The initial weight and the threshold value are optimized, and the practical steps are as follows: obtaining the optimal initial weight wbAnd an optimum threshold value thetab
(4a) Custom wbPopulation sum thetabThe size of the population, and the cross probability, the variation probability and the optimization criterion of the genetic algorithm of the heuristic search and optimization technology of natural evolution;
(4b) according to user-defined wbPopulation sum thetabSize of the population, randomly generating an initial w comprising a plurality of individualsbPopulation and initial thetabA population, each individual representing a genotype of a chromosome;
(4c) calculating an initial wbPopulation and initial thetabThe fitness value of each individual in the population is judged, whether the fitness value meets the optimization criterion is judged, and if yes, the initial w is obtainedbPopulation and initial thetabBest individual in population and optimal initial weight w represented by the best individualbAnd an optimum threshold value thetabAnd outputs the optimal initial weight wbAnd an optimum threshold value thetab(ii) a Otherwise, will wbIndividual composition w of population not satisfying fitnessb1Group of thetabThe composition theta of individuals in the population that do not meet the fitnessb1Population and performing the step (4 d);
(4d) according to the cross probability and the mutation probability, at wb1Crossing and varying individuals in the population to obtain offspring wb2Group of, at the same time as thetab1Crossing and varying individuals in the population to obtain offspring thetab2Population;
(4e) will wb2Population as initial wbGroup of thetab2Population as initial thetabPopulation, execute (4 c).
3. The dual model fusion based shield construction ground settlement of claim 1The quantity prediction method is characterized in that the genetic algorithm pair M adopting the heuristic search and optimization technology based on natural evolution in the step (6)2The Gama parameter and the Sigma parameter in the method are optimized, and the practical steps are as follows: obtaining the optimal Gama parameter value and the optimal Sigma parameter value
(6a) Self-defining the sizes of the Gama population and the Sigma population, and cross probability, variation probability and optimization criteria of a genetic algorithm of a naturally evolved heuristic search and optimization technology;
(6b) randomly generating an initial Gama population and an initial Sigma population which comprise a plurality of individuals according to the sizes of the customized Gama population and the Sigma population, wherein each individual represents the genotype of a chromosome;
(6c) calculating the fitness value of each individual in the initial Gama population and the initial Sigma population, judging whether the fitness value meets the optimization criterion, if so, obtaining the best individual in the initial Gama population and the initial Sigma population, the optimal Gama parameter value and the optimal Sigma parameter value which are represented by the best individual, and outputting the optimal Gama parameter value and the optimal Sigma parameter value; otherwise, the individuals which do not meet the fitness in the Gama population are combined into Gama1The individuals which do not meet the fitness in the Sigma population are combined into the Sigma1Population and performing step (6 d);
(6d) according to the cross probability and the mutation probability in Gama1Crossing and varying individuals in the population to obtain offspring Gama2Population while in Sigma1Crossing and varying individuals in the population to obtain offspring Sigma2Population;
(6e) will Gama2The population was used as the initial Gama population, Sigma2Population as initial Sigma population (6c) was performed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810650131.2A CN108985340B (en) | 2018-06-22 | 2018-06-22 | Shield construction ground settlement prediction method based on dual-model fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810650131.2A CN108985340B (en) | 2018-06-22 | 2018-06-22 | Shield construction ground settlement prediction method based on dual-model fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108985340A CN108985340A (en) | 2018-12-11 |
| CN108985340B true CN108985340B (en) | 2021-06-01 |
Family
ID=64538086
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810650131.2A Active CN108985340B (en) | 2018-06-22 | 2018-06-22 | Shield construction ground settlement prediction method based on dual-model fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108985340B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188947B (en) * | 2019-05-30 | 2021-04-27 | 上海隧道工程有限公司 | Method and system for predicting current ring target in shield deviation correction |
| CN110362905B (en) * | 2019-07-03 | 2021-03-05 | 中国科学院武汉岩土力学研究所 | Method for determining instability probability of shield tunnel excavation surface |
| CN114894154B (en) * | 2019-09-06 | 2024-01-02 | 北京住总集团有限责任公司 | Method and equipment for judging stability of shield construction section |
| CN110617074B (en) * | 2019-09-20 | 2020-10-09 | 西安电子科技大学 | Incidence relation method for ground settlement and tunneling parameters in shield construction |
| CN111259594A (en) * | 2020-02-10 | 2020-06-09 | 中国机械设备工程股份有限公司 | Long-term post-construction settlement prediction method based on settlement monitoring result of filling engineering |
| CN113204824B (en) * | 2021-05-21 | 2023-04-07 | 上海大学 | Multi-model fusion shield construction settlement prediction method and system |
| CN113239439B (en) * | 2021-05-21 | 2022-04-05 | 上海大学 | Shield construction ground surface settlement prediction system and method |
| CN114462185B (en) * | 2021-11-30 | 2022-12-06 | 中国建筑第五工程局有限公司 | Method for estimating five-dimensional evolution of space form of surface subsider caused by tunnel construction |
| CN114413832B (en) * | 2021-12-02 | 2023-12-15 | 中国通信建设第三工程局有限公司 | Road monitoring system and method based on optical fiber sensing |
| CN115130734A (en) * | 2022-06-06 | 2022-09-30 | 北京城建设计发展集团股份有限公司 | Method and system for predicting construction influence of penetration project based on LightGBM and deep learning algorithm |
| CN115018164B (en) * | 2022-06-09 | 2023-04-07 | 广州市第三市政工程有限公司 | Method, system, equipment and storage medium for predicting sedimentation of slurry balance pipe jacking construction |
| CN117474170B (en) * | 2023-11-17 | 2024-05-28 | 中电建铁路建设投资集团有限公司 | Shield settlement deformation model construction method based on neural network |
| CN118376203B (en) * | 2024-06-21 | 2024-09-20 | 四川旷想科技有限公司 | Pipeline settlement displacement monitoring method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101699477A (en) * | 2009-10-21 | 2010-04-28 | 东南大学 | Neural network method for accurately predicting dam deformation |
| CN103942433A (en) * | 2014-04-21 | 2014-07-23 | 南京市测绘勘察研究院有限公司 | Building settlement prediction method based on historical data analysis |
| CN104766129A (en) * | 2014-12-31 | 2015-07-08 | 华中科技大学 | Subway shield construction surface deformation warning method based on temporal and spatial information fusion |
| WO2017031615A1 (en) * | 2015-08-24 | 2017-03-02 | 深圳朝伟达科技有限公司 | Method of constructing ground subsidence prediction model of two-cavity salt cavern reservoir |
| CN107153843A (en) * | 2017-05-03 | 2017-09-12 | 西安电子科技大学 | Surface subsidence forecasting system and method based on SVMs |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170221152A1 (en) * | 2016-02-02 | 2017-08-03 | Loss Technology Services, Inc. | Water damage mitigation estimating system and method |
-
2018
- 2018-06-22 CN CN201810650131.2A patent/CN108985340B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101699477A (en) * | 2009-10-21 | 2010-04-28 | 东南大学 | Neural network method for accurately predicting dam deformation |
| CN103942433A (en) * | 2014-04-21 | 2014-07-23 | 南京市测绘勘察研究院有限公司 | Building settlement prediction method based on historical data analysis |
| CN104766129A (en) * | 2014-12-31 | 2015-07-08 | 华中科技大学 | Subway shield construction surface deformation warning method based on temporal and spatial information fusion |
| WO2017031615A1 (en) * | 2015-08-24 | 2017-03-02 | 深圳朝伟达科技有限公司 | Method of constructing ground subsidence prediction model of two-cavity salt cavern reservoir |
| CN107153843A (en) * | 2017-05-03 | 2017-09-12 | 西安电子科技大学 | Surface subsidence forecasting system and method based on SVMs |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108985340A (en) | 2018-12-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108985340B (en) | Shield construction ground settlement prediction method based on dual-model fusion | |
| CN113239439B (en) | Shield construction ground surface settlement prediction system and method | |
| CN101344389A (en) | Method for estimating tunnel surrounding rock displacement by neural network | |
| CN113187556A (en) | Tunnel earthquake risk analysis method under fault dislocation based on total probability | |
| Wu et al. | Multi-objective optimization of shield construction parameters based on random forests and NSGA-II | |
| Xue et al. | Optimum design calculation method for the reasonable buried depth: A case study from Hong Kong-Zhuhai-Macao immersed tunnel | |
| Lv et al. | Study on settlement prediction model of deep foundation pit in sand and pebble strata based on grey theory and BP neural network | |
| CN105302934A (en) | Multi-objective intelligent optimization design method for anchoring and protecting network structure of coal mine underground roadway | |
| Tang et al. | Predicting settlement along railway due to excavation using empirical method and neural networks | |
| CN115130772A (en) | Ground surface settlement prediction method based on PSO (particle swarm optimization) optimization BP (back propagation) neural network | |
| Ding et al. | A long-term tunnel settlement prediction model based on BO-GPBE with SHM data | |
| Xiao et al. | Safety monitoring of expressway construction based on multisource data fusion | |
| CN116127788B (en) | Intelligent prediction method for surface damage of directional cutting, roof crushing, expanding and filling mining working face | |
| CN116629125A (en) | Whole process data model analysis method for reinforced peat soil | |
| CN112257154A (en) | Urban tunnel transparent construction monitoring method and system | |
| An et al. | [Retracted] Comparison and Economic Envelope Structure Schemes for Deep Foundation Pit of Subway Stations Based on Fuzzy Logic | |
| CN113836812B (en) | Shield construction pose adjusting method for identifying hard rock thickness by using intelligent algorithm | |
| CN115169243A (en) | GA-PSO-GLSSVM algorithm-based soil-rock composite stratum deep foundation pit deformation time sequence prediction method | |
| Qiao et al. | Artificial neural network to predict the surface maximum settlement by shield tunneling | |
| CN108446413A (en) | Method for optimally measuring pile diameter of grouting-formed club-footed pile | |
| Chen et al. | Prediction of tunneling-induced surface settlement using a Bi-LSTM model with high generalization performance | |
| Yang et al. | Monitoring and prediction of highway foundation settlement based on particle swarm optimization and support vector machine | |
| CN117057632B (en) | Method for evaluating precipitation recharging feasibility of pebble layer deep foundation pit | |
| Cui et al. | Analysis and Prediction of the Factors Influencing Postconstruction Surface Deformation of Pipe‐Jacking Tunnel in Soft Clay Strata in China | |
| Shi et al. | A Study of Several Supervised Learning Methods for the Prediction of Pit Settlements |
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 |