CN110990938B - Soft measurement method for tunnel face in rock digging state - Google Patents
Soft measurement method for tunnel face in rock digging state Download PDFInfo
- Publication number
- CN110990938B CN110990938B CN201911384893.3A CN201911384893A CN110990938B CN 110990938 B CN110990938 B CN 110990938B CN 201911384893 A CN201911384893 A CN 201911384893A CN 110990938 B CN110990938 B CN 110990938B
- Authority
- CN
- China
- Prior art keywords
- rock mass
- tunnel face
- parameters
- state
- soft measurement
- 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
- 239000011435 rock Substances 0.000 title claims abstract description 213
- 238000000691 measurement method Methods 0.000 title claims abstract description 20
- 230000005641 tunneling Effects 0.000 claims abstract description 97
- 238000005259 measurement Methods 0.000 claims abstract description 65
- 239000011159 matrix material Substances 0.000 claims abstract description 52
- 238000013528 artificial neural network Methods 0.000 claims abstract description 28
- 230000004927 fusion Effects 0.000 claims abstract description 20
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000010219 correlation analysis Methods 0.000 claims abstract description 8
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000012417 linear regression Methods 0.000 claims description 14
- 238000009412 basement excavation Methods 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000013507 mapping Methods 0.000 claims description 11
- 230000035515 penetration Effects 0.000 claims description 6
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 238000003786 synthesis reaction Methods 0.000 claims description 5
- 230000002194 synthesizing effect Effects 0.000 claims description 4
- 230000001174 ascending effect Effects 0.000 claims description 3
- 150000001875 compounds Chemical class 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 12
- 238000000034 method Methods 0.000 description 9
- 108010074864 Factor XI Proteins 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 4
- 230000005284 excitation Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012627 multivariate algorithm Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Excavating Of Shafts Or Tunnels (AREA)
Abstract
The invention discloses a soft measurement method for a tunnel face in a rock digging state, which comprises the following steps: extracting historical data of rock machine parameters; preprocessing the historical data of the rock machine parameters to obtain a tunneling parameter matrix and a tunnel face rock mass state parameter matrix; performing correlation analysis on the state parameters and the tunneling parameters of the tunnel face rock mass, and screening the tunneling parameters with strong correlation; respectively establishing corresponding tunnel face rock mass state soft measurement models by utilizing step regression, BP neural network and support vector regression, and performing decision-making level fusion on rock mass state parameter initial estimated values predicted by the three tunnel face rock mass state soft measurement models based on an improved D-S evidence theory to obtain an updated rock mass state soft measurement model; and acquiring new tunneling parameters, acquiring a new tunneling parameter matrix through pretreatment, substituting the new tunneling parameter matrix into the updated rock mass state soft measurement model, and predicting the proposed values of the rock mass joint and the uniaxial compressive strength under the current geological condition. The invention has high measurement precision and strong real-time and robustness.
Description
Technical Field
The invention belongs to the technical field of tunnel engineering TBM construction, and particularly relates to a soft measurement method for a tunnel face in a rock digging state.
Background
In the TBM construction process, rock mass joints and uniaxial compressive strength are key variables for representing the integrity and strength characteristics of a tunnel face in rock mass excavation, but the rock mass state parameters can only be obtained through manual on-site sketch, sampling and indoor tests during the stop and maintenance of the TBM, the acquisition means is laggard, the advance and real-time performance of rock mass state information cannot be realized, so that the timely adjustment of an excavation scheme and control parameters is difficult to make when the TBM encounters stratum changes or complex geological conditions, the TBM excavation efficiency is low, the economy is seriously wasted, and the serious accidents of blocking, damage, scrapping and even casualties of the TBM are caused. On the other hand, the TBM tunneling parameters can be measured on line by arranging related sensors, and the TBM tunneling parameters such as the rotating speed of a cutterhead, the propelling speed, the total thrust, the torque of the cutterhead, the penetration degree of the cutterhead and the like and the state parameters of the rock mass of the tunnel face have inseparable relation, so that the TBM tunneling parameters can be used as important auxiliary variables for representing the conditions of the rock mass in front.
In order to fully integrate the advantages of online measurement of TBM tunneling parameters, a rock mass state soft measurement method is provided and applied to modeling of a mapping relation between a measurable auxiliary variable which is easy to measure and a key variable which is difficult to directly measure, so that the in-tunnel rock mass state of a tunnel face is sensed in real time. The existing rock mass state soft measurement method mainly comprises a linear regression model and an artificial intelligence model, wherein the linear regression model has the advantages of strong interpretability and simplicity and convenience in application, but the model establishment needs researchers to have strong specialty due to the fact that the number of influencing factors is large, the modeling process is complex and the like. Second, the artificial intelligence model has strong non-linear representation capability, but it is susceptible to sample size and data quality. In summary, the rock mass state soft measurement model based on the existing linear regression or artificial intelligence has certain defects, so that the development of the rock mass state soft measurement method with strong instantaneity and robustness has strong practical significance for solving the problem of inaccurate measurement and slow measurement of the rock mass state parameters.
Disclosure of Invention
The invention provides a soft measurement method of a tunnel face in rock digging state, aiming at solving the problems of the existing soft measurement model of the rock mass state that the modeling process is complex and the model is easily influenced by the sample capacity and the like, and solving the problems of inaccurate and not fast measurement of the rock mass state parameters.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a soft measurement method for a rock digging state of a tunnel face comprises the following steps:
s1, extracting rock machine parameter historical data reflecting the rock mass state of a tunnel face through a TBM big data monitoring platform;
s2, preprocessing the acquired rock machine parameter historical data to acquire a TBM tunneling parameter matrix A and a tunnel face rock mass state parameter matrix Y;
s3, performing correlation analysis on the tunnel face rock mass state parameters and the TBM tunneling parameters according to the TBM tunneling parameter matrix A and the tunnel face rock mass state parameter matrix Y, and screening TBM tunneling parameters with strong correlation;
s4, respectively establishing three corresponding tunnel face rock mass state soft measurement models for the screened TBM tunneling parameters and tunnel face rock mass state parameters based on stepwise regression, BP neural network and support vector regression, and performing decision-level fusion on rock mass state parameter initial estimated values predicted by the three tunnel face rock mass state soft measurement models based on an improved D-S evidence theory to further obtain an updated tunnel face rock mass state soft measurement model;
and S5, acquiring new TBM tunneling parameters on line, acquiring a new TBM tunneling parameter matrix according to the step S2, substituting the new TBM tunneling parameter matrix into the updated tunnel face rock mass state soft measurement model, and predicting the rock mass joint and uniaxial compressive strength recommended value under the current geological condition through the updated tunnel face rock mass state soft measurement model.
In the step S1, the rock machine parameters comprise TBM tunneling parameters and tunnel face rock mass state parameters, wherein the TBM tunneling parameters comprise cutter head torque T, total thrust F, cutter head rotating speed n, propelling speed v, cutter head penetration P, cutter head power P and host machine belt machine pump motor current I; the state parameters of the tunnel face rock mass comprise a rock mass joint Jv and uniaxial compressive strength UCS.
In step S2, the rock machine parameter preprocessing includes the steps of:
s2.1, dividing the obtained TBM tunneling parameters into an ascending section, a stable section and a stopping section, and rejecting stopping section data;
s2.2, standardizing TBM tunneling parameters after the shutdown section is removed;
s2.3, extracting a TBM tunneling parameter matrix and a tunnel face rock mass state parameter matrix.
In step S3, the tunnel face rock mass state parameter and TBM tunneling parameter correlation analysis includes the following steps:
s3.1, according to the preprocessed rock machine parameters, solving a correlation coefficient Cor between the tunnel face rock mass state parameters and the TBM tunneling parameters, wherein a calculation formula of the correlation coefficient Cor is as follows:
in the formula, z and u respectively represent two different TBM tunneling parameters in a TBM tunneling parameter matrix A; i represents an ith TBM tunneling parameter sample;
s3.2 screening the correlation coefficient Cor according to the criterion that the larger the correlation coefficient is and the stronger the correlation is>0.7 of TBM tunneling parameters, so that a TBM tunneling parameter matrix A '= [ n', T ', p', F ] with strong correlation is screened out n ′]。
In step S4, the soft measurement model of the rock mass state of the face based on stepwise regression is as follows:
Y pre1 =A′*B;
in the formula, B represents a linear regression coefficient vector, Y pre1 Representing the initial estimated value of the rock mass state parameter based on step regression;
the soft measurement model of the rock mass state of the tunnel face based on the BP neural network comprises the following steps:
in the formula, delta t Indicating the bias, a, from the hidden layer to the output layer st Representing the weight of the hidden layer to the output layer, H s Representing the hidden layer output matrix,/ 2 Indicating the number of nodes of the hidden layer, s indicating the node of the s-th hidden layer, Y pre2 Representing an initial rock mass state parameter estimated value based on a BP neural network;
the soft measurement model of the rock mass state of the tunnel face based on the support vector regression is as follows:
in the formula, b represents the bias from the hidden layer to the output layer, w represents the weight from the hidden layer to the output layer, k (-) represents a kernel mapping function, N represents the total number of TBM tunneling parameter samples, and Y represents pre3 And representing initial estimated values of rock mass state parameters based on support vector regression.
The decision-making level fusion of rock state parameter initial estimated values predicted by three tunnel face rock state soft measurement models based on the improved D-S evidence theory comprises the following steps:
(1) defining weight coefficients of three tunnel face rock mass state soft measurement models obtained through D-S fusion function mapping as basic probability values, namely BPA values;
(2) according to the initial estimated value and error of the rock state parameters predicted by stepwise regression, the initial estimated value and error of the rock state parameters predicted by BP neural network, and the initial estimated value and error of the rock state parameters predicted by support vector regression, the BPA is estimated;
(3) correcting the BPA value by adopting a credibility factor, synthesizing the corrected BPA value according to Dempster synthesis rules, and updating the BPA value;
(4) and respectively fusing rock mass state parameter initial estimation values predicted based on linear regression, BP neural network and support vector regression according to the updated BPA value to obtain optimal rock mass state parameter estimation.
The BPA value is m (C), the m (C) represents the weight coefficient of each soft measurement model of the rock mass state of the face through a D-S fusion function m (-) mapping, and the m (C) meets the following rule:
in the formula, C represents a soft measurement model of the rock mass state of the tunnel face and can be taken as C 1 ,C 2 ,C 3 ,C 1 ,C 2 ,C 3 Respectively representing a soft measurement model of the rock mass state of the palmar face based on linear regression, BP neural network and support vector regression, wherein theta represents a model corpus space and theta = { C 1 ,C 2 ,C 3 And m (-) represents a fusion function of D-S evidence theory.
The initial estimation of the BPA value comprises the following steps:
randomly extracting N groups of rock mass state parameters and TBM tunneling parameter data from a tunnel face rock mass state parameter matrix Y and a TBM tunneling parameter matrix A' with strong correlation as an evidence set;
secondly, on the ith evidence set, rock mass state parameter initial estimation values predicted according to stepwise regressionAnd error E i1 Based on the initial estimation value of the rock mass state parameter predicted by the BP neural network->And error E i2 Initial estimate of a rock mass state parameter predicted by support vector regression>And error E i3 To initially estimate BPA.
The calculation formula for the initial estimate of the BPA value is:
in the formula, w i1 、w i2 、w i3 Respectively representing the ith evidence set initial weight, m, of the rock mass state soft measurement model based on step regression, BP neural network and support vector regression i (C 1 )=w i1 ,m i (C 2 )=w i2 ,m i (C 3 )=w i3 。
in the formula (I), the compound is shown in the specification,reflecting the conflict situation of the evidence set, the coefficient 1/(1-q) is the regularization factor.
the invention has the beneficial effects that:
the method fully integrates the advantages of on-line measurement of TBM tunneling parameters, and can sense the in-tunneling-rock-mass state of the tunnel face in real time by modeling the mapping relation between measurable and easily-measured auxiliary variables (TBM tunneling parameters such as the rotation speed of a cutter head, the propelling speed, the total thrust, the torque of the cutter head, the penetration degree of the cutter head and the like) and key variables (tunnel-face rock-mass state parameters such as rock mass joints, uniaxial compressive strength and the like) which are difficult to directly measure; on the other hand, the soft measurement method for the rock mass state provided by the invention can improve the defects of low precision, poor robustness and the like of the existing linear regression and artificial intelligence model in the rock mass state parameter prediction, thereby providing a proper tunneling scheme and a control parameter adjustment suggestion for a TBM driver.
The method is simple, high in measurement accuracy, strong in instantaneity and robustness, does not interfere with normal tunneling of the TBM, improves the measurement means of the tunnel face rock mass state, improves the tunneling efficiency of the TBM, and is high in safety and high in economy.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive effort based on the embodiments of the present invention, are within the scope of the present invention.
A soft measurement method for a tunnel face in a rock digging state comprises the following steps as shown in figure 1:
s1, obtaining historical data of rock machine parameters: extracting a rock machine parameter historical database reflecting the rock mass state of the tunnel face through a TBM big data monitoring platform;
the rock machine parameters comprise TBM tunneling parameters and tunnel face rock mass state parameters, and the TBM tunneling parameters comprise cutterhead torque T, total thrust F, cutterhead rotating speed n, propelling speed v, cutterhead penetration P, cutterhead power P and main machine belt machine pump motor current I; the state parameters of the tunnel face rock mass comprise a rock mass joint Jv and uniaxial compressive strength UCS.
S2, pre-processing rock machine parameters: preprocessing the acquired rock machine parameter historical data to acquire a TBM tunneling parameter matrix and a tunnel face rock mass state parameter matrix;
the rock machine parameter preprocessing comprises the following steps:
s2.1, dividing the obtained TBM tunneling parameters into ascending section data, stabilizing section data and stopping section data, and rejecting the stopping section data and abnormal data;
s2.2, standardizing TBM tunneling data according to the following formula, and zooming each tunneling parameter;
wherein n ', T', F ', … …, p', F n Respectively representing the standardized rotating speed of a cutter head, the torque of the cutter head, the total thrust, … …, the penetration degree of the cutter head and the positive stress of a single cutter, wherein M represents the number of hobs on the cutter head, and f represents the frictional resistance borne by the TBM during tunneling, and can be obtained by estimating according to experience; the standardized TBM tunneling data can also be obtained by dividing the TBM tunneling data by a TBM tunneling parameter delivery value corresponding to the equipment;
s2.3, extracting a TBM tunneling parameter matrix A and a tunnel face rock mass state parameter matrix Y, A = [ n ', T', F ', … …, p', F) according to standardized rock machine parameter historical data n ′],Y=[Jv,UCS]。
S3, rock machine parameter correlation analysis: performing correlation analysis on the tunnel face rock mass state parameters and the TBM tunneling parameters according to the preprocessed rock mass state parameters, and screening a TBM tunneling parameter matrix with strong correlation;
the rock machine parameter correlation analysis comprises the following steps:
s3.1, according to the preprocessed rock machine parameters, solving a correlation coefficient Cor between the tunnel face rock mass state parameters and the TBM tunneling parameters according to the following formula;
in the formula, z and u respectively represent two different TBM tunneling parameters in a TBM tunneling parameter matrix A; i represents an ith TBM tunneling parameter sample;
s3.2 screening the correlation coefficient Cor according to the criterion that the larger the correlation coefficient is and the stronger the correlation is>0.7 of TBM tunneling parameters, so that a TBM tunneling parameter matrix A '= [ n', T ', p', F ] with strong correlation is screened out n ′];
S4, constructing a rock mass state soft measurement model: respectively constructing corresponding rock mass state soft measurement models for the TBM tunneling parameter matrix A' after screening and the tunnel face rock mass state parameter matrix Y by utilizing step regression, BP neural network and support vector regression, and then performing decision-level fusion on initial values of rock mass state parameters estimated by the step regression, BP neural network and support vector regression by utilizing multivariate algorithm fusion based on an improved D-S evidence theory to obtain a constructed rock mass state soft measurement model and realize optimal estimation on a rock mass joint Jv and uniaxial compressive strength UCS;
s4.1 step regression-based initial value Y of rock mass state parameter pre1 Predicting;
TBM tunneling parameter matrix with strong correlation constructed by using least square methodAnd palm face rock mass state parameter matrix>Step regression model in between, obtain the tunnel face rock mass state parameter Y estimated by the step regression model pre1 And a regression equation between the TBM tunneling parameter matrix A':
Y pre1 =A′*B;
in the formula, N represents the total number of TBM tunneling parameter samples, i represents the ith TBM tunneling parameter sample, and B represents a linear regression coefficient vector;
s4.2 rock mass state parameter initial value Y based on BP neural network pre2 Predicting;
s4.2.1 network initialization: setting the number of nodes of the BP neural network input layer as l 1 The number of nodes in the hidden layer is l 2 The number of nodes of the output layer is l 3 (ii) a Weights alpha of input layers to hidden layers rs Weight α from hidden layer to output layer st The bias from the input layer to the hidden layer is gamma s The bias from hidden layer to output layer is delta t (ii) a Learning rate is eta, and excitation function is g (x); the excitation function g (x) selects a Sigmoid function, and a TBM tunneling parameter matrix A' with strong correlation is used as an input layer parameter and substituted into the excitation function g (x) to obtain a formula:
s4.2.2 implicit layer output calculation: according to the TBM tunneling parameter matrix A' with strong correlation, inputting the weight alpha from the layer to the hidden layer rs And an offset of gamma s Solving the hidden layer output matrix H s The formula of (1) is as follows:
in the formula (II), A' r Representing a strong correlation TBM tunneling parameter matrix, gamma, corresponding to the r-th input layer node s Representing the bias from the input layer corresponding to the s-th hidden layer node to the hidden layer;
s4.2.3 output layer output calculation: outputting matrix H according to hidden layer s Weight α from hidden layer to output layer st And an offset delta t And outputting rock mass state parameter initial estimation Y output by the layer based on the BP neural network pre2 Comprises the following steps:
s4.3 rock mass state parameter initial value Y based on support vector regression pre3 Predicting;
s4.3.1 support vector parameter initialization: setting a bias b from a support vector model hidden layer to an output layer, a weight w from the hidden layer to the output layer, a model hyperparameter-penalty factor gamma and a kernel mapping function k (-) and mapping a TBM tunneling parameter matrix A' with strong correlation to a high-dimensional space, wherein gamma meets the condition that gamma is greater than 0;
s4.3.2 objective function J (w, ξ) calculation;
in the formula, xi represents the error of the rock mass state parameter predicted by the support vector regression model;
s4.3.3 support vector regression-based rock mass state parameter initial value Y pre3 Estimating;
s4.4 based on the improved D-S evidence theory, the initial value Y of the rock mass state parameter based on stepwise regression pre1 Rock mass state parameter initial value Y based on BP neural network pre2 Initial value Y of rock mass state parameter based on support vector regression pre3 Performing multivariate algorithm fusion to predict the predicted value Y of the optimal rock mass state parameter pre ;
(1) Basic Probability value definition (BPA);
defining soft measurement models of tunnel face rock mass states based on linear regression, BP neural network and support vector regression as C 1 ,C 2 ,C 3 Θ represents the model corpus space, and Θ = { C 1 ,C 2 ,C 3 And f, m (-) represents a fusion function of the D-S evidence theory, m (C) represents a weight coefficient of each palmar rock mass state soft measurement model obtained by mapping the fusion function m (-) and is also called as a basic probability value, and the following rules are satisfied:
in the formula, C represents a soft measurement model of the rock mass state of the tunnel face and can be taken as C 1 ,C 2 ,C 3 。
(2) Initial estimation of a BPA value;
randomly extracting N groups of rock mass state parameters and TBM tunneling parameter data from a tunnel face rock mass state parameter matrix Y and a TBM tunneling parameter matrix A' with strong correlation as an evidence set;
secondly, on the ith evidence set, rock mass state parameter initial estimation values predicted according to stepwise regressionAnd error E i1 And the initial estimation value of the rock mass state parameter predicted by the BP neural network>And error E i2 Initial estimation value of rock mass state parameter predicted by support vector regression>And error E i3 And initially estimating BPA, giving a larger weight to the rock mass state soft measurement model with small prediction error, and giving a smaller weight to the rock mass state soft measurement model with large prediction error.
The implementation of the initial estimation of BPA is specifically set forth below:
respectively endowing the rock mass state soft measurement model based on step regression, BP neural network and support vector regression with initial weight w aiming at the ith evidence set i1 、w i2 、w i3 According to the rock mass state parameter prediction result after the improved D-S evidence theory is fused, the result isThe calculation formula is as follows:
mean error value E of fusion results i And variance D (E) i ) The calculation formulas of (a) and (b) are respectively as follows:
E i =w i1 E i1 +w i2 E i2 +w i3 E i3 ;
for the prediction errors of the soft measurement models of the rock mass states of different working faces, the prediction errors are assumed to be independent of each other, namely:
to make the error variance D (E) of the prediction result i ) Minimum, pair D (E) i ) Calculating a deviation guide, order
When satisfying w i1 +w i2 +w i3 When =1, the following results can be obtained:
in the formula, w i1 、w i2 、w i3 BPA as the ith evidence set, i.e., m i (C 1 )=w i1 ,m i (C 2 )=w i2 ,m i (C 3 )=w i3 。
(3) Updating the BPA value;
the synthesis rule is the key of the D-S evidence theory, and the traditional D-S evidence theory can generate Zaldeh paradox when synthesizing high-conflict evidence sources. In order to solve the above problem, the present embodiment introduces a reliability factor ξ to measure the reliability of the evidence source, and modifies the evidence source according to the reliability factor ξ to reduce conflicts between different evidence sources, and the specific method is as follows:
firstly, defining the distance u between the BPA value of the ith evidence set and the BPA value of the residual evidence sets of different tunnel face rock mass state soft measurement models i :
In the formula, N represents the number of evidence sets, namely the total number of samples of TBM tunneling parameters; c a Representing the a rock mass state soft measurement model;
distance u i The sizes of the first evidence set and the second evidence set reflect the consistency degree of the BPA values of the ith evidence set and the BPA values of the remaining evidence sets of the soft measurement model of the rock mass states of different tunnel faces; if u i The smaller the confidence factor xi, the higher the consistency i The higher; otherwise, the confidence factor xi i The lower.
synthesizing the corrected BPA on different evidence sets according to Dempster synthesis rules to obtain the optimal weight of soft measurement models of rock mass states of different working faces, and taking the optimal weight as a weight coefficient for decision fusion of linear regression, BP neural network and support vector regression by improving D-S evidence theory;
the Dempster synthesis rule is:
in the formula (I), the compound is shown in the specification,reflecting the conflict situation of evidence, the coefficient 1/(1-q) is the regularization factor.
(4) Estimating optimal rock mass state parameters;
according to the weight coefficient after the fusion of the improved D-S evidence theory, the initial values of the rock state parameters predicted by linear regression, BP neural network and support vector regression are respectively fused to obtain the predicted value Y of the optimal rock state parameter pre Namely, the formula of the soft measurement model of the rock mass state of the tunnel face is as follows:
s5, online prediction of rock mass state parameters: and (3) acquiring new TBM tunneling parameters on line, acquiring a new TBM tunneling parameter matrix according to the step S2, and substituting the new TBM tunneling parameter matrix into the established tunnel face rock mass state soft measurement model, so as to predict the rock mass joint Jv and the uniaxial compressive strength UCS recommended value under the current geological condition.
S6, evaluating and verifying the soft measurement model of the rock mass state of the tunnel face: obtaining initial estimated value Y of rock mass state parameter predicted by stepwise regression pre1 Initial estimation value Y of rock mass state parameter predicted by BP neural network pre2 Initial estimation value Y of rock mass state parameter predicted by support vector regression pre3 And improving the optimal value Y of the rock mass state parameter predicted by the D-S evidence theory pre Then, in this embodiment, three evaluation indexes, namely, pearson Correlation Coefficient (PCC), root Mean Square Error (RMSE), and Accuracy (Accuracy), are introduced to quantify the prediction errors of the rock mass state soft measurement models corresponding to different methods, specifically as follows:
in the formula, y represents the measured value of the rock mass state parameter,representing the estimated value of the rock mass state parameter predicted by the rock mass state soft measurement model. Generally speaking, the larger the PCC and Accuracy values are, the better the soft measurement model effect of the tunnel face rock mass state is, while for the RMSE index, the smaller the value thereof is, the higher the prediction Accuracy of the soft measurement model of the tunnel face rock mass state is.
From the aspects of prediction precision and fitting effect, the soft measurement model of the rock mass state of the tunnel face based on the improved D-S evidence theory can realize decision-level fusion of prediction results of different models, and has higher prediction precision and stronger robustness.
According to the method and the device, the condition of the tunnel face rock mass can be sensed in real time, and the current geological information is fed back to the TBM driver according to the recommended value predicted by the tunnel face rock mass state soft measurement model, so that the TBM driver can conveniently make a tunneling scheme and control parameter adjustment in time when the TBM encounters stratum changes or complex geological conditions, the tunneling efficiency is improved, the economic loss is reduced, and major accidents are avoided.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (9)
1. A soft measurement method for a rock digging state of a tunnel face is characterized by comprising the following steps:
s1, extracting rock machine parameter historical data reflecting the rock mass state of a tunnel face through a TBM big data monitoring platform;
s2, preprocessing the acquired rock machine parameter historical data to acquire a TBM tunneling parameter matrix A and a tunnel face rock mass state parameter matrix Y;
s3, performing correlation analysis on the tunnel face rock mass state parameters and the TBM tunneling parameters according to the TBM tunneling parameter matrix A and the tunnel face rock mass state parameter matrix Y, and screening TBM tunneling parameters with strong correlation;
the tunnel face rock mass state parameter and TBM tunneling parameter correlation analysis comprises the following steps:
s3.1, according to the preprocessed rock machine parameters, solving a correlation coefficient Cor between the tunnel face rock mass state parameters and the TBM tunneling parameters, wherein a calculation formula of the correlation coefficient Cor is as follows:
in the formula, z and u respectively represent two different TBM tunneling parameters in a TBM tunneling parameter matrix A; i represents an ith TBM tunneling parameter sample;
s3.2 screening the correlation coefficient Cor according to the criterion that the larger the correlation coefficient is and the stronger the correlation is>0.7 of TBM tunneling parameters, thereby screening out a TBM tunneling parameter matrix A '= [ n', T ', p', F ] with strong correlation n ′];
S4, respectively establishing three corresponding tunnel face rock mass state soft measurement models for the screened TBM tunneling parameters and tunnel face rock mass state parameters based on step regression, BP neural network and support vector regression, and performing decision-level fusion on rock mass state parameter initial estimation values predicted by the three tunnel face rock mass state soft measurement models based on an improved D-S evidence theory to obtain an updated tunnel face rock mass state soft measurement model;
and S5, acquiring new TBM tunneling parameters on line, acquiring a new TBM tunneling parameter matrix according to the step S2, substituting the new TBM tunneling parameter matrix into the updated tunnel face rock mass state soft measurement model, and predicting the rock mass joint and uniaxial compressive strength recommended value under the current geological condition through the updated tunnel face rock mass state soft measurement model.
2. The soft measurement method for the rock mass excavation state of the tunnel face according to claim 1, wherein in step S1, the rock machine parameters comprise TBM tunneling parameters and tunnel face rock mass state parameters, and the TBM tunneling parameters comprise cutterhead torque T, total thrust F, cutterhead rotating speed n, propelling speed v, cutterhead penetration P, cutterhead power P and main machine belt machine pump motor current I; the state parameters of the tunnel face rock mass comprise a rock mass joint Jv and uniaxial compressive strength UCS.
3. The soft measurement method for rock mass excavation of a tunnel face according to claim 2, characterized in that in step S2, the rock machine parameter preprocessing comprises the following steps:
s2.1, dividing the obtained TBM tunneling parameters into an ascending section, a stable section and a stopping section, and rejecting stopping section data;
s2.2, standardizing TBM tunneling parameters after the shutdown section is removed;
s2.3, extracting a TBM tunneling parameter matrix and a tunnel face rock mass state parameter matrix.
4. The soft measurement method for the rock mass excavation state of the tunnel face according to the claim 1 or 3, characterized in that in the step S4, the soft measurement model for the rock mass state of the tunnel face based on stepwise regression is:
Y pre1 =A′*B;
wherein B represents a linear regression coefficient vector, Y pre1 Representing the initial estimated value of the rock mass state parameter based on step regression;
the soft measurement model of the rock mass state of the tunnel face based on the BP neural network comprises the following steps:
in the formula, delta t Indicating the bias from the hidden layer to the output layer, α st Representing the weight of the hidden layer to the output layer, H s Representing the hidden layer output matrix,/ 2 Indicating the number of nodes of the hidden layer, s indicating the s-th hidden layer node, Y pre2 Representing an initial rock mass state parameter estimated value based on a BP neural network;
the soft measurement model of the rock mass state of the tunnel face based on the support vector regression is as follows:
in the formula, b represents the bias from the hidden layer to the output layer, w represents the weight from the hidden layer to the output layer, k (-) represents a kernel mapping function, N represents the total number of TBM tunneling parameter samples, and Y represents pre3 And representing initial estimated values of rock mass state parameters based on support vector regression.
5. The soft measurement method for the rock mass excavation state of the tunnel face according to claim 4, wherein the decision-level fusion of the initial estimated values of the rock mass state parameters predicted by the soft measurement models for the rock mass states of the three tunnel faces based on the improved D-S evidence theory comprises the following steps:
(1) defining weight coefficients of three tunnel face rock mass state soft measurement models obtained through D-S fusion function mapping as basic probability values, namely BPA values;
(2) according to the initial estimated value and error of the rock state parameters predicted by stepwise regression, the initial estimated value and error of the rock state parameters predicted by BP neural network, and the initial estimated value and error of the rock state parameters predicted by support vector regression;
(3) correcting the BPA value by adopting a credibility factor, synthesizing the corrected BPA value according to a Dempster synthesis rule and updating the BPA value;
(4) and respectively fusing rock mass state parameter initial estimated values predicted based on linear regression, BP neural network and support vector regression according to the updated BPA value to obtain the optimal rock mass state parameter estimation.
6. The soft measurement method for the rock mass excavation state of the working face according to claim 5, wherein the BPA value is m (C), the m (C) represents a weight coefficient of each soft measurement model for the rock mass excavation state of the working face obtained by mapping a D-S fusion function m (-) and the m (C) satisfies the following rule:
in the formula, C represents a soft measurement model of the rock mass state of the tunnel face and can be taken as C 1 ,C 2 ,C 3 ,C 1 ,C 2 ,C 3 Respectively representing a soft measurement model of the rock mass state of the palmar face based on linear regression, BP neural network and support vector regression, wherein theta represents a model corpus space and theta = { C 1 ,C 2 ,C 3 And m (-) represents a fusion function of D-S evidence theory.
7. The soft measurement method for rock mass excavation state of a tunnel face according to claim 6, wherein the initial estimation of the BPA value comprises the following steps:
randomly extracting N groups of rock mass state parameters and TBM tunneling parameter data from a tunnel face rock mass state parameter matrix Y and a TBM tunneling parameter matrix A' with strong correlation as an evidence set;
secondly, on the ith evidence set, rock mass state parameter initial estimation values predicted according to stepwise regressionAnd error E i1 And the initial estimation value of the rock mass state parameter predicted by the BP neural network>And error E i2 Initial estimate of a rock mass state parameter predicted by support vector regression>And error E i3 To initially estimate the BPA value;
the calculation formula for the initial estimate of the BPA value is:
in the formula, w i1 、w i2 、w i3 Respectively representing the initial weight m of the ith evidence set of the rock mass state soft measurement model based on step regression, BP neural network and support vector regression i (C 1 )=w i1 ,m i (C 2 )=w i2 ,m i (C 3 )=w i3 。
8. The soft measurement method for rock mass excavation state of tunnel face according to claim 7, characterized in that the updated BPA value is defined asThe formula of (1) is:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911384893.3A CN110990938B (en) | 2019-12-28 | 2019-12-28 | Soft measurement method for tunnel face in rock digging state |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911384893.3A CN110990938B (en) | 2019-12-28 | 2019-12-28 | Soft measurement method for tunnel face in rock digging state |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110990938A CN110990938A (en) | 2020-04-10 |
CN110990938B true CN110990938B (en) | 2023-04-18 |
Family
ID=70076678
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911384893.3A Active CN110990938B (en) | 2019-12-28 | 2019-12-28 | Soft measurement method for tunnel face in rock digging state |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110990938B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113513331A (en) * | 2021-04-15 | 2021-10-19 | 上海交通大学 | Tunneling face rock-soil type identification method, system and medium based on shield machine operation parameters |
CN113268799B (en) * | 2021-05-27 | 2024-04-30 | 深圳市岩土综合勘察设计有限公司 | Method and system for predicting karst cave burial depth and size based on while-drilling data |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105259331B (en) * | 2015-11-06 | 2017-06-30 | 三峡大学 | A kind of jointed rock mass uniaxial strengeth Forecasting Methodology |
CN107577862B (en) * | 2017-08-30 | 2019-12-03 | 中铁工程装备集团有限公司 | A kind of TBM is in pick rock mass state real-time perception system and method |
CN109685378B (en) * | 2018-12-27 | 2020-04-24 | 中铁工程装备集团有限公司 | TBM construction surrounding rock digchability grading method based on data mining |
CN110020694A (en) * | 2019-04-19 | 2019-07-16 | 中铁工程装备集团有限公司 | A kind of TBM unfavorable geology discrimination method based on intelligent drives model |
CN110110419B (en) * | 2019-04-28 | 2022-11-18 | 中铁工程装备集团有限公司 | TBM tunneling parameter prediction method based on multi-target learning |
-
2019
- 2019-12-28 CN CN201911384893.3A patent/CN110990938B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN110990938A (en) | 2020-04-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109885907B (en) | Cloud model-based satellite attitude control system health state assessment and prediction method | |
WO2021203796A1 (en) | Disease prognosis prediction system based on deep semi-supervised multi-task learning survival analysis | |
CN108960303B (en) | Unmanned aerial vehicle flight data anomaly detection method based on LSTM | |
CN112052992B (en) | Deep learning-based construction project progress prediction system and method | |
Li et al. | Data-model interactive prognosis for multi-sensor monitored stochastic degrading devices | |
Leng et al. | A hybrid data mining method for tunnel engineering based on real-time monitoring data from tunnel boring machines | |
CN110990938B (en) | Soft measurement method for tunnel face in rock digging state | |
CN110348752B (en) | Large industrial system structure safety assessment method considering environmental interference | |
CN111598352A (en) | Concrete beam type bridge comprehensive evaluation method based on Bayesian network | |
CN109490072B (en) | Detection system for civil engineering building and detection method thereof | |
CN109131452A (en) | A kind of train status on-line prediction method based on long memory network in short-term | |
CN111813084A (en) | Mechanical equipment fault diagnosis method based on deep learning | |
CN101436057A (en) | Numerical control machining tool heat error Bayes network compensation method | |
CN112948932A (en) | Surrounding rock grade prediction method based on TSP forecast data and XGboost algorithm | |
CN116150897A (en) | Machine tool spindle performance evaluation method and system based on digital twin | |
CN114970688A (en) | Landslide monitoring data preprocessing method based on LSTMAD algorithm and Hermite interpolation method | |
CN111754034A (en) | Time sequence prediction method based on chaos optimization neural network model | |
Li et al. | Three‐way decisions based on some Hamacher aggregation operators under double hierarchy linguistic environment | |
CN111797364A (en) | Landslide multilevel safety evaluation method based on composite cloud model | |
CN111861238A (en) | Expressway bridge engineering risk assessment method and device and computer equipment | |
CN111666684B (en) | Circumferential weld risk prediction method and device for conveying pipeline and readable storage medium | |
CN111859814A (en) | Rock aging deformation prediction method and system based on LSTM deep learning | |
CN112288594A (en) | Data quality transaction processing method and system based on real-time event triggering | |
CN114065639B (en) | Closed-loop real-time inversion method for construction parameters of dredger | |
CN115759409A (en) | Water gate deformation prediction method for optimizing LSTM (least Square TM) model by multi-time mode attention mechanism |
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 |