CN115563872B - TBM hob abrasion loss prediction method based on DE-SVR algorithm - Google Patents
TBM hob abrasion loss prediction method based on DE-SVR algorithm Download PDFInfo
- Publication number
- CN115563872B CN115563872B CN202211237102.6A CN202211237102A CN115563872B CN 115563872 B CN115563872 B CN 115563872B CN 202211237102 A CN202211237102 A CN 202211237102A CN 115563872 B CN115563872 B CN 115563872B
- Authority
- CN
- China
- Prior art keywords
- abrasion loss
- tbm
- tbm hob
- svr
- hob abrasion
- 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
- 238000005299 abrasion Methods 0.000 title claims abstract description 172
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000012360 testing method Methods 0.000 claims abstract description 31
- 230000005641 tunneling Effects 0.000 claims abstract description 25
- 238000005457 optimization Methods 0.000 claims abstract description 18
- 238000012549 training Methods 0.000 claims description 12
- 230000035515 penetration Effects 0.000 claims description 11
- 230000009977 dual effect Effects 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 238000013210 evaluation model Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 230000035772 mutation Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000010276 construction Methods 0.000 abstract description 5
- 238000004458 analytical method Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 26
- 239000011435 rock Substances 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 239000010438 granite Substances 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000005422 blasting Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/10—Numerical modelling
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Biology (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Gear Processing (AREA)
Abstract
The invention discloses a TBM hob abrasion loss prediction method based on a DE-SVR algorithm, which comprises the steps of constructing a TBM hob abrasion loss sample data set; constructing an SVR model for predicting the abrasion loss of the TBM hob; constructing a DE-SVR prediction model: testing the accuracy of the DE-SVR by using the test set data; the unknown TBM hob abrasion loss is predicted, and the value of the unknown TBM hob abrasion loss is obtained. According to the method, the parameters are optimized through a DE optimization algorithm, and the analysis of the established DE-SVR prediction model shows that the positive correlation between the hob abrasion loss prediction value and the measured value is very good, so that the accuracy of the established hob abrasion loss prediction model is good, the hob abrasion loss prediction precision is effectively improved, and the overall tunneling construction efficiency is improved.
Description
Technical Field
The invention is applied to the technical field of predicting the abrasion loss of a TBM cutter head hob, and relates to a support vector machine (SVR) and a method for predicting the abrasion loss of the TBM hob after optimization.
Background
Tunnel Boring Machines (TBMs) are large, high-speed, high-automation, tunneling devices that are 3-10 times more efficient and faster than conventional methods such as drilling and blasting. At present, the TBM is increasingly widely applied to tunnel construction, especially for hard rock and long tunnels, but the rock breaking efficiency of the TBM applied under different geological conditions is different, and further the hob abrasion loss in the tunneling process is also different, so that the hob abrasion loss of the TBM needs to be predicted by an effective method. Currently, a support vector machine (SVR) algorithm is widely applied in model prediction, and a proper hob is selected by a prediction method for predicting the wear amount of a TBM hob, so that the cost of TBM engineering can be effectively reduced.
The existing method for analyzing the TBM hob is simple regression analysis, has great deviation in prediction accuracy and precision, and cannot intelligently predict the hob abrasion loss in the follow-up TBM tunneling process, so an intelligent prediction algorithm is provided to improve the prediction precision level.
Disclosure of Invention
The invention relates to an intelligent prediction method based on SVR algorithm improvement, which adopts differential evolution algorithm (DE) and support vector machine regression algorithm (SVR) to predict the hob tunneling wear amount of TBM (full-face tunnel boring machine), and effectively improves the accuracy of TBM hob wear amount prediction.
The technical scheme adopted by the invention is a TBM hob abrasion loss prediction method based on a DE-SVR algorithm, which comprises the following steps:
s100, constructing a TBM hob abrasion loss sample data set; analyzing and finishing according to TBM hob abrasion loss under different tunneling parameters acquired on site to form 15 groups of TBM hob abrasion sample data sets corresponding to different tunneling parameters;
s110: dividing a TBM hob abrasion sample data set into two groups of data, wherein the numbers 1-10 are the first group, and forming training set data; numbers 11-15 are the second group, forming test set data;
s200: constructing an SVR model for predicting the abrasion loss of the TBM hob;
s210: according to the kernel function of the SVR model for predicting the TBM hob abrasion loss, obtaining the kernel function of the SVR model for predicting the TBM hob abrasion loss through improvement, and then performing dual processing on the kernel function of the SVR model for predicting the TBM hob abrasion loss to obtain the SVR model for predicting the TBM hob abrasion loss;
s220: the kernel function K of the TBM hob wear SVR model is expressed as follows:
K(x i ，x j )＝φ(x i )·φ(x j )
the estimation function f (x) of the TBM hob wear amount is expressed as:
f(x)＝w·φ(x)+b
wherein,the mapping from a low-dimensional vector to a high-dimensional vector in the TBM hob abrasion loss is represented, χ is the vector of the TBM hob abrasion loss, ω is the vector independent variable of the TBM hob abrasion loss, and b is the displacement item parameter in the linear equation of the TBM hob abrasion loss;
s300: the method comprises the following specific steps of constructing a DE-SVR prediction model:
s310: selecting a kernel function of the DE-SVR model, and adopting a radial basis kernel function based on the established TBM hob abrasion loss SVR model:
in the radial basis function, the balance parameter C is the compromise between the proportion of TBM hob abrasion loss control misclassification samples and the algorithm complexity, and the kernel parameter sigma is the width of the radial basis of the TBM hob abrasion loss;
s320: optimizing SVR parameters by utilizing a Lagrangian multiplier method and a dual method, and converting the DE-SVR model into an SVR optimization problem, namely solving the SVR model based on an estimation function f (x) of the abrasion loss of the TBM hob in S220, wherein the regression estimation function of the obtained DE-SVR model of the abrasion loss of the TBM hob is as follows:
s330: optimizing by using a DE algorithm to obtain a TBM hob abrasion loss DE-SVR model, wherein the specific implementation steps are as follows;
s331, performing parameter optimization on a TBM hob abrasion loss SVR parameter by using a DE (differential evolution algorithm) optimization algorithm, wherein parameters to be optimized on the basis of a radial basis function DE-SVR model are a TBM hob abrasion loss balance parameter C and a TBM hob abrasion loss kernel parameter sigma;
s332: setting initial parameters, initializing TBM hob abrasion loss population scale N and evolution algebra k m The threshold values of the cross probability CR, the scaling factor a, the termination threshold value, the TBM hob abrasion loss balance parameter C and the TBM hob abrasion loss kernel parameter sigma;
s333: generating an initial TBM hob abrasion loss DE population randomly, setting the iteration number to be 0, and then carrying out DE algorithm variation operation and crossover operation;
s334: randomly selecting 3 individuals different from the current population to perform mutation operation, generating new variant individuals, further generating a new TBM hob abrasion loss population, and performing continuous iteration;
s335: training and predicting TBM hob abrasion loss sample data by using the currently obtained TBM hob abrasion loss (C, delta) as a parameter of SVR, comparing an adaptability function value with an expected value, and if a stopping condition is not met, entering the next generation of evolution;
s336: after a plurality of iterations, stopping iteration when the fitness function value is unchanged and the maximum iteration times are reached, and storing the optimal parameters (C, delta) of the SVR, thereby completing the selection of the optimal parameters of the TBM hob abrasion loss DE-SVR model;
s340: and training and predicting the TBM hob abrasion loss sample data according to the selected optimal parameters, and establishing a TBM hob abrasion loss prediction model of the DE-SVR.
S400: the accuracy of the DE-SVR is checked by using the test set data, the test set data is input into the DE-SVR prediction model obtained in the step S340, a predicted value is obtained, a performance index R2 is calculated by the predicted value and an actual measured value of the TBM hob abrasion loss, if the set TBM hob abrasion loss prediction accuracy requirement is met, the next step S500 is carried out, and the TBM hob abrasion loss prediction model of the DE-SVR is set as a final model; if the preset TBM tunneling prediction precision requirement is not met, repeating the steps of S320-S340, and continuing model training and parameter optimization until the preset TBM tunneling prediction precision is met;
s500: the unknown abrasion loss of the TBM hob is predicted: substituting the numerical values of the TBM cutter rotating speed n, the TBM cutter torque T, TBM cutter thrust F and the penetration P into a finally obtained TBM cutter abrasion loss prediction model of the DE-SVR, thereby obtaining the abrasion loss value of the unknown TBM cutter.
Step S400 checks the accuracy of the model. Mainly uses test set, R 2 Index evaluation model predicts model performance, R 2 The calculation formula of (2) is as follows:
wherein n is the number of samples in the TBM hob abrasion loss test set; y is i Representing the measured value, y in the TBM hob abrasion loss test set i ' represents a predicted value in the TBM hob abrasion loss test set;mean value of measurement values in TBM hob abrasion loss test set,/->Representing the average value of predicted values in the TBM hob abrasion loss test set; i=1, 2, …, n; performance index R of hob abrasion amount of TBM requiring prediction 2 Reaching more than 0.7, and meeting the precision requirement of predicting the abrasion loss of the TBM hob.
The set TBM hob abrasion loss prediction precision requirement is the goodness of fit R 2 And the relation between the cutter thrust, the cutter torque and the penetration and the cutter abrasion loss is better when the cutter thrust, the cutter torque and the penetration are larger than 0.7.
Compared with the prior art, the invention has the following advantages:
1. according to the method, the parameters are optimized through the DE optimization algorithm, the analysis of the established DE-SVR prediction model shows that the positive correlation between the hob abrasion loss prediction value and the measured value is very good, the algorithm shows that the accuracy of the established hob abrasion loss prediction model is good, and the hob abrasion loss prediction precision is effectively improved.
2. The SVR optimization algorithm is improved by a reasonable method, the algorithm is simple and convenient, parameter redundancy is reduced, the application is flexible, and the prediction model provided by the invention has remarkable advantages in predicting the abrasion loss of the TBM hob.
3. In actual tunneling construction, the model is used for carrying out hob abrasion loss prediction and hob life prediction, and based on cutterhead thrust, cutterhead torque and penetration tunneling parameters, the model is used for approximately predicting hob abrasion loss of every 300m tunneling mileage under geological conditions of high-strength and high-abrasion metamorphic rocks (mainly granite) encountered by engineering, and therefore the model has a great guiding effect on predicting hob life, reasonably planning tool inspection and replacement work and improving overall tunneling construction efficiency.
Drawings
FIG. 1 is a flow chart of a machine learning algorithm, hob wear prediction.
Fig. 2 is an overall frame diagram.
FIG. 3 is a graph of data statistics of the amount of hob wear in a certain segment.
Fig. 4 is a sample dataset diagram.
Fig. 5: data graph of TBM wear amount at test stage.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
As shown in fig. 1-4, the invention relates to an improved intelligent TBM hob abrasion loss prediction method based on SVR algorithm, which adopts a hybrid algorithm technology combining a differential optimization algorithm (DE) and an improved SVR algorithm to predict the TBM hob abrasion loss, and the algorithm can effectively improve the accuracy of TBM hob abrasion loss prediction.
Examples: a method for predicting hobbing cutter tunneling wear amount of TBM (full face tunnel boring machine) based on SVR algorithm comprises the following steps:
s100, constructing a TBM hob abrasion loss sample data set; analyzing and finishing according to TBM hob abrasion loss under different tunneling parameters acquired on site to form 15 groups of TBM hob abrasion sample data sets corresponding to different tunneling parameters;
s110: dividing a TBM hob abrasion sample data set into two groups of data, wherein the numbers 1-10 are the first group, and forming training set data; numbers 11-15 are the second group, forming test set data;
s200: constructing an SVR model for predicting the abrasion loss of the TBM hob;
s210: according to the kernel function of the SVR model for predicting the TBM hob abrasion loss, obtaining the kernel function of the SVR model for predicting the TBM hob abrasion loss through improvement, and then performing dual processing on the kernel function of the SVR model for predicting the TBM hob abrasion loss to obtain the SVR model for predicting the TBM hob abrasion loss;
s220: the kernel function K of the TBM hob wear SVR model is expressed as follows:
K(x i ，x j )＝φ(x i )·φ(x j )
the estimation function f (x) can be expressed as:
f(x)＝w·φ(x)+b
wherein,the mapping from a low-dimensional vector to a high-dimensional vector in the TBM hob abrasion loss is represented, χ is the vector of the TBM hob abrasion loss, ω is the vector independent variable of the TBM hob abrasion loss, and b is the displacement item parameter in the linear equation of the TBM hob abrasion loss;
s300: the method comprises the following specific steps of constructing a DE-SVR prediction model:
s310: selecting a kernel function of the DE-SVR model, and adopting a radial basis kernel function based on the built TBM hob abrasion loss prediction model:
in the radial basis function, the balance parameter C is a compromise between the proportion of the error division samples of the abrasion loss of the TBM hob and the algorithm complexity, and the kernel parameter sigma is the width of the radial basis of the abrasion loss of the TBM hob;
s320: solving the constraint optimization by utilizing a Lagrangian multiplier method, utilizing SVR dual problem representation, finally converting the DE-SVR model into an SVR optimization problem, and solving the SVR model to obtain a regression estimation function of the TBM hob abrasion loss DE-SVR model, wherein the regression estimation function is as follows:
s330: the DE algorithm is optimized to obtain a TBM hob abrasion loss DE-SVR algorithm model, and the specific implementation steps are as follows;
s331, performing parameter optimization on a TBM hob abrasion loss SVR parameter by using a DE (differential evolution algorithm) optimization algorithm, wherein parameters to be optimized on the basis of a radial basis function DE-SVR model are a TBM hob abrasion loss balance parameter C and a TBM hob abrasion loss kernel parameter sigma;
s332: setting initial parameters, initializing TBM hob abrasion loss population scale N and evolution algebra k m The threshold values of the cross probability CR, the scaling factor a, the termination threshold value, the TBM hob abrasion loss balance parameter C and the TBM hob abrasion loss kernel parameter sigma;
s333: generating an initial TBM hob abrasion loss DE population randomly, setting the iteration number to be 0, and then carrying out DE algorithm variation operation and crossover operation;
s334: randomly selecting 3 individuals different from the current population to perform mutation operation, generating new variant individuals, further generating a new TBM hob abrasion loss population, and performing continuous iteration;
s335: training and predicting TBM hob abrasion loss sample data by using the currently obtained TBM hob abrasion loss (C, delta) as a parameter of SVR, comparing an adaptability function value with an expected value, and if a stopping condition is not met, entering the next generation of evolution;
s336: after a plurality of iterations, stopping iteration when the fitness function value is unchanged and the maximum iteration times are reached, and storing the optimal parameters (C, delta) of the SVR, thereby completing the selection of the optimal parameters of the TBM hob abrasion loss DE-SVR model;
s340: and training and predicting the TBM hob abrasion loss sample data according to the selected optimal parameters, and establishing a TBM hob abrasion loss prediction model of the DE-SVR.
S400: the accuracy of the DE-SVR is checked by using the test set data, the test set data is input into the DE-SVR prediction model obtained in the step S340, a predicted value is obtained, a performance index R2 is calculated by the predicted value and an actual measured value of the abrasion loss of the TBM hob, if the set precision requirement of the prediction of the abrasion loss of the TBM hob is met, the next step S500 is carried out, and the DE-SVR prediction model is set as a final model; if the preset TBM tunneling prediction precision requirement is not met, repeating the steps of S320-S340, and continuing model training and parameter optimization until the preset TBM tunneling prediction precision is met;
s500: the unknown abrasion loss of the TBM hob is predicted: substituting the numerical values of the TBM cutter rotating speed n, the TBM cutter torque T, TBM cutter thrust F and the penetration P into a finally obtained TBM cutter abrasion loss prediction model of the DE-SVR, thereby obtaining the abrasion loss value of the unknown TBM cutter.
Step S400 checks the accuracy of the model. Mainly uses test set, R 2 Index evaluation model predicts model performance, R 2 The calculation formula of (2) is as follows:
wherein n is the number of samples in the TBM hob abrasion loss test set; y is i Representing the measured value, y in the TBM hob abrasion loss test set i ' represents a predicted value in the TBM hob abrasion loss test set;mean value of measurement values in TBM hob abrasion loss test set,/->Representing the average value of predicted values in the TBM hob abrasion loss test set; i=1, 2, …, n; performance index R of hob abrasion amount of TBM requiring prediction 2 Reaching more than 0.7, and meeting the precision requirement of predicting the abrasion loss of the TBM hob.
The following table shows the sample data collected:
the set TBM hob abrasion loss prediction precision requirement is the goodness of fit R 2 And the relation between the cutter thrust, the cutter torque and the penetration and the cutter abrasion loss is better when the cutter thrust, the cutter torque and the penetration are larger than 0.7.
As shown in FIG. 5, the data graph of the TBM abrasion loss in the test stage shows that the accuracy of the constructed hob abrasion prediction model is better. The model is used for carrying out hob abrasion loss prediction and hob life prediction, is used for approximately predicting hob abrasion loss of every 300m tunneling mileage under geological conditions of high-strength and strong-abrasion metamorphic rocks (mainly granite) encountered by engineering based on cutterhead thrust, cutterhead torque and penetration tunneling parameters, further predicts hob life, reasonably plans tool inspection and replacement work, and improves overall tunneling construction efficiency.
Claims (2)
1. A TBM hob abrasion loss prediction method based on a DE-SVR algorithm is characterized in that: the method comprises the following steps:
s100: constructing a TBM hob abrasion loss sample data set; analyzing and finishing according to TBM hob abrasion loss under different tunneling parameters acquired on site to form 15 groups of TBM hob abrasion sample data sets corresponding to different tunneling parameters;
s110: dividing a TBM hob abrasion sample data set into two groups of data, wherein the numbers 1-10 are the first group, and forming training set data; numbers 11-15 are the second group, forming test set data;
s200: constructing an SVR model for predicting the abrasion loss of the TBM hob;
s210: according to the kernel function of the SVR model for predicting the TBM hob abrasion loss, obtaining the kernel function of the SVR model for predicting the TBM hob abrasion loss through improvement, and then performing dual processing on the kernel function of the SVR model for predicting the TBM hob abrasion loss to obtain the SVR model for predicting the TBM hob abrasion loss;
s220: the kernel function K of the TBM hob wear SVR model is expressed as follows:
K(x i ，x j )＝φ(x i )·φ(x j )
the estimation function f (x) of the TBM hob wear amount is expressed as:
f(x)＝ω·φ(x)+b
wherein,the mapping from a low-dimensional vector to a high-dimensional vector in the TBM hob abrasion loss is represented, χ is the vector of the TBM hob abrasion loss, ω is the vector independent variable of the TBM hob abrasion loss, and b is the displacement item parameter in the linear equation of the TBM hob abrasion loss;
s300: the method comprises the following specific steps of constructing a DE-SVR prediction model:
s310: selecting a kernel function of the DE-SVR model, and adopting a radial basis kernel function based on the established TBM hob abrasion loss SVR model:
in the radial basis function, the balance parameter C is the compromise between the proportion of TBM hob abrasion loss control misclassification samples and the algorithm complexity, and the kernel parameter sigma is the width of the radial basis of the TBM hob abrasion loss;
s320: solving SVR by utilizing a Lagrangian multiplier method and a dual method, and converting a DE-SVR model into an SVR optimization problem, namely solving the SVR model based on an estimation function f (x) of the abrasion loss of the TBM hob in S220, wherein the regression estimation function of the obtained DE-SVR model of the abrasion loss of the TBM hob is as follows:
s330: optimizing by using a DE algorithm to obtain a TBM hob abrasion loss DE-SVR model, wherein the specific implementation steps are as follows;
s331, carrying out parameter optimization on a TBM hob abrasion loss SVR parameter by utilizing a DE differential evolution optimization algorithm, wherein parameters to be optimized on the basis of a radial basis function DE-SVR model are a TBM hob abrasion loss balance parameter C and a TBM hob abrasion loss kernel parameter sigma;
s332: setting initial parameters, initializing TBM hob abrasion loss population scale N and evolution algebra k m The threshold values of the cross probability CR, the scaling factor a, the termination threshold value, the TBM hob abrasion loss balance parameter C and the TBM hob abrasion loss kernel parameter sigma;
s333: generating an initial TBM hob abrasion loss DE population randomly, setting the iteration number to be 0, and then carrying out DE algorithm variation operation and crossover operation;
s334: randomly selecting 3 individuals different from the current population to perform mutation operation, generating new variant individuals, further generating a new TBM hob abrasion loss population, and performing continuous iteration;
s335: training and predicting TBM hob abrasion loss sample data by using the currently obtained TBM hob abrasion loss (C, delta) as a parameter of SVR, comparing an adaptability function value with an expected value, and if a stopping condition is not met, entering the next generation of evolution;
s336: after a plurality of iterations, stopping iteration when the fitness function value is unchanged and the maximum iteration times are reached, and storing the optimal parameters (C, delta) of the SVR, thereby completing the selection of the optimal parameters of the TBM hob abrasion loss DE-SVR model;
s340: training and predicting TBM hob abrasion loss sample data according to the selected optimal parameters, and establishing a TBM hob abrasion loss prediction model of the DE-SVR;
s400: the accuracy of the DE-SVR is checked by using the test set data, the test set data is input into the DE-SVR prediction model obtained in the step S340 to obtain a predicted value, and the performance index R is calculated by the predicted value and the actual measurement value of the abrasion loss of the TBM hob 2 If the set TBM hob abrasion loss prediction precision requirement is met, performing the next step S500, and setting a TBM hob abrasion loss prediction model of the DE-SVR as a final model; if the preset TBM tunneling prediction precision requirement is not met, repeating the steps of S320-S340, and continuing model training and parameter optimization until the preset TBM tunneling prediction precision is met;
s500: the unknown abrasion loss of the TBM hob is predicted: substituting the numerical values of the TBM cutter rotating speed n, the TBM cutter torque T, TBM cutter thrust F and the penetration P into a TBM cutter abrasion loss prediction model of the final DE-SVR, thereby obtaining the abrasion loss value of the unknown TBM cutter;
step S400, checking the accuracy of the model; using test set, R 2 Index evaluation model predicts model performance, R 2 The calculation formula of (2) is as follows:
wherein n is the number of samples in the TBM hob abrasion loss test set; y is i Representing the measured value, y in the TBM hob abrasion loss test set i ' represents a predicted value in the TBM hob abrasion loss test set;mean value of measurement values in TBM hob abrasion loss test set,/->Representing the average value of predicted values in the TBM hob abrasion loss test set; i=1, 2, …, n; performance index R of hob abrasion amount of TBM requiring prediction 2 Reaching more than 0.7, and meeting the precision requirement of predicting the abrasion loss of the TBM hob.
2. The TBM hob abrasion loss prediction method based on the DE-SVR algorithm according to claim 1, wherein the method comprises the following steps: the set TBM hob abrasion loss prediction precision requirement is the goodness of fit R 2 And the relation between the cutter thrust, the cutter torque and the penetration and the cutter abrasion loss is better when the cutter thrust, the cutter torque and the penetration are larger than 0.7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211237102.6A CN115563872B (en) | 2022-10-10 | 2022-10-10 | TBM hob abrasion loss prediction method based on DE-SVR algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211237102.6A CN115563872B (en) | 2022-10-10 | 2022-10-10 | TBM hob abrasion loss prediction method based on DE-SVR algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115563872A CN115563872A (en) | 2023-01-03 |
| CN115563872B true CN115563872B (en) | 2024-04-12 |
Family
ID=84744757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211237102.6A Active CN115563872B (en) | 2022-10-10 | 2022-10-10 | TBM hob abrasion loss prediction method based on DE-SVR algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115563872B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000204884A (en) * | 1999-01-12 | 2000-07-25 | Mitsui Eng & Shipbuild Co Ltd | Method and device for detecting wear of cutter bit of shield machine |
| JP2001040983A (en) * | 1999-08-02 | 2001-02-13 | Hitachi Constr Mach Co Ltd | Tunnel boring machine and method for controlling the same |
| CN106570275A (en) * | 2016-11-07 | 2017-04-19 | 沈阳工业大学 | Method based on CAI value for predicting abrasion of TBM hobbing cutter |
| CN108710722A (en) * | 2018-04-04 | 2018-10-26 | 郑州大学 | A kind of prediction technique of TBM hob abrasion |
| CN111300146A (en) * | 2019-11-29 | 2020-06-19 | 上海交通大学 | Numerical control machine tool cutter abrasion loss online prediction method based on spindle current and vibration signal |
| CN113051683A (en) * | 2021-03-26 | 2021-06-29 | 长安大学 | Method, system, equipment and storage medium for predicting service life of numerical control machine tool cutter |
| CN113222273A (en) * | 2021-05-26 | 2021-08-06 | 中铁十八局集团有限公司 | TBM tunneling rate prediction method based on GWO-FW-MKL-SVR algorithm |
| CN114417697A (en) * | 2021-12-07 | 2022-04-29 | 山东大学 | Neural network-based TBM hob abrasion real-time prediction method and system |
-
2022
- 2022-10-10 CN CN202211237102.6A patent/CN115563872B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000204884A (en) * | 1999-01-12 | 2000-07-25 | Mitsui Eng & Shipbuild Co Ltd | Method and device for detecting wear of cutter bit of shield machine |
| JP2001040983A (en) * | 1999-08-02 | 2001-02-13 | Hitachi Constr Mach Co Ltd | Tunnel boring machine and method for controlling the same |
| CN106570275A (en) * | 2016-11-07 | 2017-04-19 | 沈阳工业大学 | Method based on CAI value for predicting abrasion of TBM hobbing cutter |
| CN108710722A (en) * | 2018-04-04 | 2018-10-26 | 郑州大学 | A kind of prediction technique of TBM hob abrasion |
| CN111300146A (en) * | 2019-11-29 | 2020-06-19 | 上海交通大学 | Numerical control machine tool cutter abrasion loss online prediction method based on spindle current and vibration signal |
| CN113051683A (en) * | 2021-03-26 | 2021-06-29 | 长安大学 | Method, system, equipment and storage medium for predicting service life of numerical control machine tool cutter |
| CN113222273A (en) * | 2021-05-26 | 2021-08-06 | 中铁十八局集团有限公司 | TBM tunneling rate prediction method based on GWO-FW-MKL-SVR algorithm |
| CN114417697A (en) * | 2021-12-07 | 2022-04-29 | 山东大学 | Neural network-based TBM hob abrasion real-time prediction method and system |
Non-Patent Citations (3)
| Title |
|---|
| New Tool Wear Estimation Method of the Milling Process Based on Multisensor Blind Source Separation;Chen Gao;《Mathematical Problems in Engineering 2021》;全文 * |
| 基于数据驱动的铣削刀具寿命预测研究;黄志平;黄新宇;李亮;郭月龙;;制造技术与机床(第01期);全文 * |
| 数据驱动的钛合金铣削刀具寿命预测与管理技术研究;黄志平;《中国优秀硕士学位论文全文数据库 工程科技I辑》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115563872A (en) | 2023-01-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Prediction model of rock mass class using classification and regression tree integrated AdaBoost algorithm based on TBM driving data | |
| CN106778010B (en) | TBM cutter life prediction method based on data-driven support vector regression machine | |
| CN109358582B (en) | Big-data-based large-mining-height fully-mechanized coal mining face multi-equipment cooperative operation control method | |
| Song et al. | A new fuzzy c-means clustering-based time series segmentation approach and its application on tunnel boring machine analysis | |
| CN108470095B (en) | TBM (tunnel boring machine) propulsion prediction method based on data-driven radial basis function model | |
| CN110378069A (en) | A method of prediction boring machine cutter accumulated quality loss late | |
| TWI584134B (en) | Method for analyzing variation causes of manufacturing process and system for analyzing variation causes of manufacturing process | |
| CN113404502B (en) | Shield hob abrasion monitoring device and method based on ballast piece morphology | |
| CN113414638B (en) | Variable working condition milling cutter wear state prediction method based on milling force time sequence diagram deep learning | |
| CN114662699A (en) | Shield attitude prediction method based on machine learning | |
| CN115563863A (en) | Method and terminal for determining abrasion loss of slurry shield hobbing cutter in real time based on deep learning | |
| CN110704963A (en) | Rapid method for optimizing tunneling parameters of earth pressure balanced type shield tunneling machine | |
| CN110633504A (en) | Prediction method for coal bed gas permeability | |
| CN107368463B (en) | Roadway nonlinear deformation prediction method based on fiber bragg grating sensor network data | |
| CN112966355B (en) | Method for predicting residual service life of shield machine cutter based on deep learning | |
| CN115563872B (en) | TBM hob abrasion loss prediction method based on DE-SVR algorithm | |
| CN111340275B (en) | Tunnel support mode selection real-time prediction method based on detection while drilling technology | |
| CN109543268A (en) | The recognition methods of TBM propulsive force major influence factors based on kriging model | |
| CN117252086A (en) | Method, system and equipment for health evaluation and degradation prediction of shield tunneling machine cutterhead | |
| CN117172360A (en) | Drilling mechanical drilling speed optimization method, system, equipment and medium based on MLP and high-efficiency PSO | |
| CN116108339A (en) | Actual measurement data outlier detection and correction method for tunnel boring machine | |
| CN113222273B (en) | TBM tunneling rate prediction method based on GWO-FW-MKL-SVR algorithm | |
| CN114818493A (en) | Method for quantitatively evaluating integrity degree of tunnel rock mass | |
| Vidner et al. | Multidisciplinary design optimization of a Mobile Miner using the OpenMDAO platform | |
| CN115774912A (en) | Mechanism and data combined driving TBM hob abrasion prediction method |
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 |