CN108868805B - Shield method for correcting error based on statistical analysis in conjunction with XGboost - Google Patents
Shield method for correcting error based on statistical analysis in conjunction with XGboost Download PDFInfo
- Publication number
- CN108868805B CN108868805B CN201810583493.4A CN201810583493A CN108868805B CN 108868805 B CN108868805 B CN 108868805B CN 201810583493 A CN201810583493 A CN 201810583493A CN 108868805 B CN108868805 B CN 108868805B
- Authority
- CN
- China
- Prior art keywords
- shield
- data
- attitude
- tunneling
- 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 49
- 238000007619 statistical method Methods 0.000 title claims abstract description 12
- 238000012937 correction Methods 0.000 claims abstract description 51
- 230000005641 tunneling Effects 0.000 claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 23
- 230000000153 supplemental effect Effects 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 7
- 238000010276 construction Methods 0.000 claims description 17
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 230000009467 reduction Effects 0.000 claims description 7
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000013461 design Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 4
- 241000208340 Araliaceae Species 0.000 claims description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 3
- 235000008434 ginseng Nutrition 0.000 claims description 3
- 230000006698 induction Effects 0.000 claims description 3
- 239000002689 soil Substances 0.000 claims description 3
- 230000008676 import Effects 0.000 claims 1
- 125000003367 polycyclic group Chemical group 0.000 abstract description 3
- 238000006073 displacement reaction Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000007637 random forest analysis Methods 0.000 description 3
- 238000009412 basement excavation Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 241000512668 Eunectes Species 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
- 206010034719 Personality change Diseases 0.000 description 1
- 238000012098 association analyses Methods 0.000 description 1
- 239000004567 concrete Substances 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 239000011178 precast concrete Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000011435 rock Substances 0.000 description 1
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21D—SHAFTS; TUNNELS; GALLERIES; LARGE UNDERGROUND CHAMBERS
- E21D9/00—Tunnels or galleries, with or without linings; Methods or apparatus for making thereof; Layout of tunnels or galleries
- E21D9/06—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining
- E21D9/093—Control of the driving shield, e.g. of the hydraulic advancing cylinders
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21D—SHAFTS; TUNNELS; GALLERIES; LARGE UNDERGROUND CHAMBERS
- E21D9/00—Tunnels or galleries, with or without linings; Methods or apparatus for making thereof; Layout of tunnels or galleries
- E21D9/06—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining
- E21D9/08—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining with additional boring or cutting means other than the conventional cutting edge of the shield
- E21D9/0875—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining with additional boring or cutting means other than the conventional cutting edge of the shield with a movable support arm carrying cutting tools for attacking the front face, e.g. a bucket
- E21D9/0879—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining with additional boring or cutting means other than the conventional cutting edge of the shield with a movable support arm carrying cutting tools for attacking the front face, e.g. a bucket the shield being provided with devices for lining the tunnel, e.g. shuttering
Landscapes
- Engineering & Computer Science (AREA)
- Mining & Mineral Resources (AREA)
- Environmental & Geological Engineering (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Geology (AREA)
- Excavating Of Shafts Or Tunnels (AREA)
Abstract
The invention proposes a kind of shield method for correcting error based on statistical analysis in conjunction with XGboost to realize step for solving the lower technical problem of correction precision existing in the prior art are as follows: obtains shield supplemental characteristic packet；Obtain training set data and test set data；Obtain the shield attitude Partial Linear Models based on XGboost；It obtains multiple groups and tunnels sample data；It rectifies a deviation to attitude of shield machine.Pass through statistical analysis, attitude parameter value is divided into multiple sections, and each section tunneling data is calculated by the regression model that XGboost is established, obtain every group of multi-group data comprising a plurality of shield data, so that each section provides one group of shield reference data, parameter setting provides foundation when giving polycyclic correction, passes through the regression model being fitted based on XGboost, the accuracy of given shield reference data is effectively increased, and then improves the accuracy of driving correction.
Description
Technical field
The invention belongs to tunnel machine equipment technology method fields, are related to for shield axis in shield tunneling process
Method for correcting error when posture deviates.
Background technique
Shield method is the Fully-mechanized construction method of one of shallow mining method.It is by shield machine in the earth formation according to setting
The route of meter is promoted, and is supported surrounding country rock by shield shell and section of jurisdiction and is prevented the collapsing in tunnel.Exist simultaneously
Carries out soil excavation with cutting apparatus in front of excavation face, by being unearthed, machinery is transported outside hole, and lean on jack in rear pressurization jacking,
And assembled precast concrete section of jurisdiction, form a kind of mechanized construction method of tunnel structure.
During shield-tunneling construction, since underground construction environment is very complicated, especially geological condition, shield appearance is influenced
The factor of state is very more.If shield attitude changes, driving axis will be deviated with original design axis, that is to say, that
Tunneling direction will deviate.When deviateing, then axis must be corrected into back former design axis direction in time, if not
Can timely correction, shield driving will substantial deviation design route, it will causes to delay work and serious economic loss, and entangles
Inclined process is and its complexity, factor in need of consideration are especially more again.
Shield attitude measuring system monitors shield attitude parameter in real time in tunneling process, and one ring of every driving will be to shield
Tail gap measures.It must then rectify a deviation when shield axis attitude parameter is greater than set correction threshold value.Correction is first
First it needs to be determined that various parameters combination correction scheme, and according at this time gap of the shield tail and safe shield tail spacing determine point
Several rings are rectified a deviation, determine boring parameter with grouting parameter, and then again under several rings carry out driving correction.If shield attitude parameter
It is not above threshold value, boring parameter, grouting parameter are normally arranged at this time, continue shield driving.The correction of shield machine at present
Mainly correction is carried out using manual mode setting according to the result of measurement by operator to operate.This method often more according to
Rely experience and correction accuracy it is not high enough, therefore to the research of shield-tunneling construction axis method for correcting error be extremely it is necessary to.
The main thought of shield method for correcting error is that axis correction is simulated by founding mathematical models at present, then passes through experience
Formula calculates related shield attitude parameter, then by carrying out correction setting compared with designing axis.This method more relies on
Artificial experience causes correction precision lower, in order to avoid correction precision excessively relies on the defect of manual deviation rectification experience, the prior art
It proposes a kind of by being positioned to shield machine and by compared with desired trajectory, then transmits a signal to driving mechanism and entangled
Inclined method, such as application publication number are CN105736007 A, entitled " the shield machine positioning and correction of fusion formation information
The patent application of system and method " discloses a kind of shield machine method for correcting error for merging formation information, by by inertial navigation system
The signal that system issues is converted into attitude angle, displacement and the axial position in tri- directions shield machine X, Y, Z through data operation, then leads to
Attitude angle and axial position are crossed, shield machine is positioned in real time, by real-time positioning compared with desired trajectory, is sent out by controller
Mechanism kinematic is rolled in signal driving out, tunnels shield machine according to desired trajectory, and then realize correction.This method can be with tradition
Method compare, eliminate the reliance on manual deviation rectification experience, improve correction precision to a certain extent.But its existing deficiency
Place is: 1, this method is to be converted to X, Y, Z attitude angle and displacement by calculate to signal, then pass through attitude angle with displacement
Amount positions shield machine, compares acquisition departure with desired trajectory again after positioning and rectifies a deviation, departure cannot be direct
It obtains, so that identified correction accuracy of measurement is not high enough；2, this method does not consider to be necessary to ensure that during correction a certain amount of
Safe shield tail clearance issues, being only applicable to small range deviation, (correction amount taken when tunneling correction, in each ring makes shield
It is 0 that structure machine, which twists to gap of the shield tail, and using bias at this time as critical value, a wide range of deviation is greater than the bias
Deviation, small range deviation are to be less than the deviation of the bias).
Summary of the invention
It is an object of the invention to overcome the problems of the above-mentioned prior art, provide it is a kind of based on statistical analysis with
Attitude parameter value is divided into multiple interval ranges, and right by statistical analysis by the shield method for correcting error that XGboost is combined
Each interval range tunneling data is calculated by the regression model that XGboost is established, and is obtained every group and is included a plurality of shield
The multi-group data of data, so that each interval range provides one group of shield reference data, parameter setting is provided when giving polycyclic correction
Foundation effectively increases the accuracy of given shield reference data by the regression model being fitted based on XGboost,
And then the accuracy of driving correction is improved, concrete methods of realizing includes the following steps:
(1) shield supplemental characteristic packet is obtained:
(1a) is filled the missing values in shield machine shield data collected, obtains complete shield data, rejects
Exceptional value in complete shield data, or the exceptional value in complete shield data is carried out by the median of complete shield data
Replacement, obtains new shield data；
(1b) calculates the mean value of every ring shield data in new shield data, obtains the shield data based on ring, and to base
Data Dimensionality Reduction is carried out in the shield data of ring, obtains the shield supplemental characteristic comprising shield driving data and shield attitude data
Packet；
(2) training set data and test set data are obtained:
To in shield supplemental characteristic packet shield driving data and shield attitude data be normalized, and return what is obtained
For 80% shield driving data as training set data x-train, 80% shield driving data institute is right in one change shield data
The shield attitude data answered are as training set data y-train, and remainder data is as test set data；
(3) shield attitude Partial Linear Models are obtained:
Training set data x-train and training set data y-train are fitted using XGboost algorithm, obtained
Shield attitude Partial Linear Models；
(4) multiple groups are obtained and tunnel sample data:
(4a) uses wide method, using widened attitude parameter maximum correction amount is section span at double, to shield parameter number
Discretization is carried out according to the shield attitude data in packet, obtains multiple shield attitude parameters section；
(4b) is associated rule analysis to each shield attitude parameter section and shield driving data, obtains association rule
Then；
(4c) utilizes correlation rule, concludes the pick in shield data with each shield attitude parameter section with incidence relation
Into data, the multi-group data sample that incidence relation is respectively provided with multiple shield attitude parameters section is obtained；
(4d) uses wide method, right using the maximum correction amount of ring attitude parameter every during practice of construction as section span
Each shield attitude parameter section is divided again, obtains new shield attitude parameter section；
In (4e) induction step (4c) in obtained multi-group data sample with each new shield attitude parameter section
Tunneling data with incidence relation obtains every group of multiple groups tunneling data comprising a plurality of tunneling data；
(5) it rectifies a deviation to attitude of shield machine:
(5a) brings every group of driving sample data in shield attitude regression model into, is intended by shield attitude regression model
It is corresponding to calculate every tunneling data in every group of tunneling data for nonlinear function between the boring parameter and attitude parameter of conjunction
Attitude parameter value, obtain a plurality of attitude parameter value；
It is reference data that (5b), which chooses the shield data that frequency of occurrence is most in every group of driving sample data, obtains a plurality of shield
Structure reference data；
(5c) chooses a boring parameter equal with the attitude parameter departure corrected from a plurality of shield reference data
Data, and the parameter on shield machine is configured according to the boring parameter data of selection, when shield machine middle line is gradually revert to
After being mutually fitted with design axis, correction is completed.
Compared with the prior art, the invention has the following advantages:
1, for the present invention when tunneling correction, using notch, horizontal, perpendicular attitude straggling parameter is as deviation benchmark, to shield number
According to statistics division is carried out, recycles the regression model being fitted based on XGboost algorithm to calculate and obtain correction reference data, in shield
It is configured on structure machine according to reference data and then realizes correction, it is middle compared with the existing technology to calculate attitude of shield machine angle, displacement
Amount positions shield machine in real time, and positioning acquisition departure compared with desired trajectory is passed through drive system again and rectified a deviation,
Compared with prior art, the accuracy of shield driving correction is effectively increased；
2, attitude parameter is divided into multiple range formats, each interval computation goes out can be for reference in correction by the present invention
Boring parameter numerical value, be configured reference for operator when correction, rectified a deviation by point polycyclic, it is ensured that shield driving
Shi Anquan gap of the shield tail, and then ensure the safety of shield-tunneling construction, compared with the existing technology, pass through positioning and desired trajectory pair
Than, then departure is gone into driving mechanism and is directly rectified a deviation, compared with prior art, effectively improve shield-tunneling construction safety
Property；
Detailed description of the invention
Fig. 1 is implementation flow chart of the invention；
Fig. 2 is the box figure method schematic diagram of outlier identification of the present invention.
Specific embodiment
Below in conjunction with the drawings and specific embodiments, invention is further described in detail.
Referring to Fig.1, steps are as follows for realization of the invention:
Step 1) obtains shield supplemental characteristic packet:
Step 1a) missing values in shield machine shield data collected are filled, complete shield data are obtained, are picked
Except the exceptional value in complete shield data, or by the medians of complete shield data to the exceptional value in complete shield data into
Row replacement, obtains new shield data；
Shield data, including shield driving data and shield attitude data, wherein using tunneling direction as standard setting shield
Machine left and right directions, shield driving data include ring number, soil pressure, the jack thrust in upper and lower, left and right direction, cutterhead revolving speed, knife
Disk torque, driving speed, angle of revolution, pitch angle, gross thrust, total oil pressure, jack stroke, jack speed, screw machine revolving speed,
Screw machine torque, grouting pressure, grouting amount, X-coordinate, Y-coordinate, Z coordinate and gap of the shield tail, shield attitude data include notch
The horizontal departure and vertical missing of horizontal departure and vertical missing and shield tail.
In the present invention, the fillna function in pandas module is called to know missing values using python program language
It is filled not and with median.
Exceptional value is identified using " box figure method " in the present invention, and exceptional value is carried out using the median of shield data
Replacement.
Box figure method: as shown in Fig. 2, the numerical value for being more than or less than the bound of box figure setting is exceptional value, if on
Quartile is U, indicates the numerical value for being greater than the sample 3/4 in all samples, sets quartile and be set as L, indicate all
It is less than the numerical value of the sample 3/4 in sample, if the difference of upper quartile and lower quartile is IQR, it may be assumed that IQR=U-L, the then upper bound
For U+1.5IQR, lower bound L-1.5IQR.
Step 1b) mean value that calculates every ring shield data in new shield data, the shield data based on ring are obtained, and right
Shield data based on ring carry out Data Dimensionality Reduction, obtain the shield supplemental characteristic comprising shield driving data and shield attitude data
Packet；
It in shield driving, is tunneled as unit of ring, data of every ring length at 2 meters or so, in shield driving
It is to be acquired as unit of minute, and each ring driving is to need or so two hours, therefore have a plurality of pick in each ring
Into data.
In collected shield data attribute be very more, and many influences of the attribute for shield attitude deviation are micro-
Its is micro-, or even is not associated with shield attitude, and the data for retaining these attributes can largely increase data volume, seriously
The effect and speed of data analysis are influenced, therefore early period of the invention carries out dimensionality reduction to data, deletes attribute and its data useless.
In the present invention, Data Dimensionality Reduction is carried out using random forests algorithm.
The feature importance function that we can carry according to it in random forest carries out attribute selection.In
In decision tree, important attribute can be chosen one by one according to comentropy, i.e., using comentropy as classifying and dividing foundation.
Shield driving data after dimensionality reduction include the jack thrust and driving of ring number, upper and lower, left and right direction
Speed, shield attitude data include the horizontal departure of notch and the horizontal departure and vertical missing of vertical missing and shield tail
Step 2) obtains training set data and test set data:
To in shield supplemental characteristic packet shield driving data and shield attitude data be normalized, and return what is obtained
For 80% shield driving data as training set data x-train, 80% shield driving data institute is right in one change shield data
The shield attitude data answered are as training set data y-train, and remainder data is as test set data；
Normalization, i.e. min-max standardization also known as deviation standardization, are the linear transfors to initial data, formula is such as
Under:
Max is sample maximum, and min is sample minimum, and x is former data, x*For the data after normalization.
Step 3) obtains shield attitude Partial Linear Models:
Training set data x-train and training set data y-train are fitted using XGboost algorithm, obtained
Shield attitude Partial Linear Models；
Realize step are as follows:
Step 3a) iterative parameter of XGboost algorithm is set；
Step 3b) training set data x-train and training set data y-train is imported among XGboost algorithm, it is real
Now to the fitting of x-train and y-train, shield attitude Partial Linear Models are obtained.
The advantages of integrated study is relative to a body Model is integrated study by mode appropriate, integrates many " individuals
Model ", obtained final mask performance are more excellent than the performance of a body Model.XGboost is boosting in integrated study
One kind of race's algorithm is the different Weak Classifier boosting algorithm of training set, has many advantages, such as that regression fit precision is high, speed is fast.
Step 4) obtains multiple groups and tunnels sample data:
Step 4a) wide method is used, using widened attitude parameter maximum correction amount is section span at double, to shield parameter
Shield attitude data in data packet carry out discretization, obtain multiple shield attitude parameters section；
Wide method is one kind of the non-supervisory discretization of data, and the codomain of attribute is divided into same widths by wide method
Section, and the span in section is specified by user；
Widened attitude parameter maximum correction amount at double, acquisition methods are as follows:
Step 4a1) to be set as widened shield-tunneling construction maximum correction amount again be ma, a is shield-tunneling construction maximum correction amount, and m is
Coefficient, m > 1 and m round numbers；
Step 4a2) attitude parameter section is divided as unit of ma, and to obtained multiple attitude parameter sections with
Tunneling data is associated rule analysis, obtains the corresponding confidence level of correlation rule；
Apriori Algorithm for Association Rules:
Association analysis is usually used in the significant connection that discovery is hidden in data set, and the connection found can use association
Rule or the form of frequent item set indicate.
Confidence level:
Confidence level, that is, confidence level indicates the degree of reliability of a certain rule.
Support:
Minimum support is a preset threshold value, indicates importance of the item collection in statistical significance, some
The support of collection is greater than the minimum support of setting, then this item collection is referred to as frequent item set.
Step 4a3) gradually expand m numerical value, repeat step (4a2), reached by the corresponding confidence level of correlation rule
M when to preset requirement calculates widened attitude parameter maximum correction amount ma at double.
Step 4b) rule analysis is associated to each shield attitude parameter section and shield driving data, it is associated with
Rule；
Step 4c) correlation rule is utilized, concluding in shield data has incidence relation with each shield attitude parameter section
Tunneling data, obtain the multi-group data sample that incidence relation is respectively provided with multiple shield attitude parameters section；
Using correlation rule, the driving number in shield data with each shield attitude parameter section with incidence relation is concluded
According to realization step are as follows:
Step 4c1) from correlation rule choose confidence level be greater than pre-set min confidence and support be greater than it is preparatory
The correlation rule of the minimum support of setting obtains the incidence relation of shield attitude parameter section and tunneling data；
Step 4c2) according to the incidence relation between shield attitude parameter section and tunneling data, conclude in shield data with
Each shield attitude parameter section has the tunneling data of incidence relation, obtains being respectively provided with multiple shield attitude parameters section
The multi-group data sample of incidence relation.
Step (4d) use wide method, using the maximum correction amount of ring attitude parameter every during practice of construction as section across
Degree, divides each shield attitude parameter section again, obtains new shield attitude parameter section；
Each shield attitude parameter section is divided again, implementation method are as follows:
Using the tunneling direction of shield machine to the right as the forward direction of horizontal attitude parameter, upwardly direction tunneling direction is vertical appearance
The forward direction of state parameter, in the positive maximum deflection difference value M for the attitude parameter chosen from new shield data1With reversed maximum deviation
Value M2[- the M formed2, M1] in range, using wide method, to each shield attitude parameter section as unit of maximum correction amount a
It is divided again, obtains shield attitude parameter section: [- M2,-na), [- na ,-(n-1) a) ..., [(n-1) a, na),
[na, M1], wherein n is coefficient,And n round numbers.
Step 4e) induction step 4c) in obtained multi-group data sample with each new shield attitude parameter region
Between with incidence relation tunneling data, obtain every group include a plurality of tunneling data multiple groups tunneling data；
Step 5) rectifies a deviation to attitude of shield machine:
Step 5a) every group of driving sample data is brought into shield attitude regression model, pass through shield attitude regression model
The nonlinear function between boring parameter and attitude parameter being fitted, calculates every tunneling data in every group of tunneling data
Corresponding attitude parameter value obtains a plurality of attitude parameter value；
Step 5b) the shield data that frequency of occurrence is most in every group of driving sample data are chosen as reference data, it obtains more
Shield reference data；
Step 5c) a driving ginseng equal with the attitude parameter departure corrected is chosen from a plurality of shield reference data
Number data, and the parameter on shield machine is configured according to the boring parameter data of selection, when shield machine middle line gradually returns
To after being mutually fitted with design axis, correction is completed.
Effect of the invention can be further illustrated by following practice:
1) this example is in AMD Athlon (tm) X4 641Quad-Core Processor CPU@2.80GHz,
Under Windows 7 (× 64) system, on Anaconda spyder operation platform, modeling practice is completed；
2) modeling practice is carried out by using the data obtained in Ningbo subway " bright building --- gymnasium " section shield-tunneling construction；
3) in data prediction, missing values are filled using the median of shield data, using " box figure method "
Outlier identification is carried out, and exceptional value is replaced using median, obtains standardized Ningbo subway after normalization
Shield data；
4) dimensionality reduction is carried out to data using the feature extracting method of random forest；
5) XGboost based on integrated study is passed through using the key parameter of extraction according to shield-tunneling construction historical data
Regression model establishes shield attitude parametric prediction model, sets the number of iterations n_estimators=4000, and model accuracy is
90.19%；
6) Association Rule Analysis takes confidence level to be greater than 0.9 rule here.It obtains shield attitude parameter and shield driving is joined
Related law between number；
7) by statistical analysis technique, the section after discretization is refined according to the maximum correction amount of shield-tunneling construction,
And the corresponding small sample in minizone is obtained, small sample is taken back in the trained regression model of XGboost, is obtained corresponding new
Shield attitude deviation.It filters out compared with can be convenient for correction reference data of the parameter as this ring of setting.
Claims (5)
1. a kind of shield method for correcting error based on statistical analysis in conjunction with XGboost, it is characterised in that include the following steps:
(1) shield supplemental characteristic packet is obtained:
(1a) is filled the missing values in shield machine shield data collected, obtains complete shield data, rejects complete
Exceptional value in shield data, or the exceptional value in complete shield data is replaced by the median of complete shield data
It changes, obtains new shield data, shield data include shield driving data and shield attitude data, wherein are with tunneling direction
Standard setting shield machine left and right directions, shield driving data include ring number, soil pressure, upper and lower, left and right direction jack thrust,
Cutterhead revolving speed, cutter head torque, driving speed, angle of revolution, pitch angle, gross thrust, total oil pressure, jack stroke, jack speed,
Screw machine revolving speed, screw machine torque, grouting pressure, grouting amount, X-coordinate, Y-coordinate, Z coordinate and gap of the shield tail, shield attitude number
According to the horizontal departure and vertical missing of the horizontal departure and vertical missing and shield tail that include notch；
(1b) calculates the mean value of every ring shield data in new shield data, obtains the shield data based on ring, and to based on ring
Shield data carry out Data Dimensionality Reduction, obtain the shield supplemental characteristic packet comprising shield driving data and shield attitude data；
(2) training set data and test set data are obtained:
To the shield driving data and the normalization that is normalized, and will obtain of shield attitude data in shield supplemental characteristic packet
80% shield driving data are as training set data x-train in shield data, corresponding to 80% shield driving data
Shield attitude data are as training set data y-train, and remainder data is as test set data；
(3) shield attitude Partial Linear Models are obtained:
Training set data x-train and training set data y-train are fitted using XGboost algorithm, obtain shield appearance
State Partial Linear Models；
(4) multiple groups are obtained and tunnel sample data:
(4a) uses wide method, using widened attitude parameter maximum correction amount at double as section span, to shield supplemental characteristic packet
In shield attitude data carry out discretization, obtain multiple shield attitude parameters section；
(4b) is associated rule analysis to each shield attitude parameter section and shield driving data, obtains correlation rule；
(4c) utilizes correlation rule, concludes the driving number in shield data with each shield attitude parameter section with incidence relation
According to obtaining the multi-group data sample for being respectively provided with incidence relation with multiple shield attitude parameters section；
(4d) uses wide method, using the maximum correction amount of ring attitude parameter every during practice of construction as section span, to each
Shield attitude parameter section is divided again, obtains new shield attitude parameter section；
Have in obtained multi-group data sample with each new shield attitude parameter section in (4e) induction step (4c)
The tunneling data of incidence relation obtains every group of multiple groups tunneling data comprising a plurality of tunneling data；
(5) it rectifies a deviation to attitude of shield machine:
(5a) brings every group of driving sample data in shield attitude regression model into, is fitted by shield attitude regression model
Nonlinear function between boring parameter and attitude parameter calculates the corresponding appearance of every tunneling data in every group of tunneling data
State parameter value obtains a plurality of attitude parameter value；
It is reference data that (5b), which chooses the shield data that frequency of occurrence is most in every group of driving sample data, obtains a plurality of shield ginseng
Examine data；
(5c) chooses a boring parameter data equal with the attitude parameter departure corrected from a plurality of shield reference data,
And the parameter on shield machine is configured according to the boring parameter data of selection, when shield machine middle line gradually revert to and designs
After axis is mutually fitted, correction is completed.
2. the shield method for correcting error according to claim 1 based on statistical analysis in conjunction with XGboost, which is characterized in that
Training set data x-train and training set data y-train are fitted using XGboost algorithm described in step (3), it is real
Existing step are as follows:
The iterative parameter of (3a) setting XGboost algorithm；
(3b) imports training set data x-train and training set data y-train among XGboost algorithm, realizes to x-
The fitting of train and y-train obtains shield attitude Partial Linear Models.
3. the shield method for correcting error according to claim 1 based on statistical analysis in conjunction with XGboost, which is characterized in that
Widened attitude parameter maximum correction amount at double described in step (4a), acquisition methods are as follows:
It is ma that (4a1), which is set as widened shield-tunneling construction maximum correction amount again, and a is shield-tunneling construction maximum correction amount, and m is coefficient, m >
1 and m round numbers；
(4a2) divides attitude parameter section as unit of ma, and to obtained multiple attitude parameter sections and driving number
According to rule analysis is associated, the corresponding confidence level of correlation rule is obtained；
(4a3) gradually expands the numerical value of m, repeats step (4a2), reaches default by the corresponding confidence level of correlation rule and wants
M when asking calculates widened attitude parameter maximum correction amount ma at double.
4. the shield method for correcting error according to claim 1 based on statistical analysis in conjunction with XGboost, which is characterized in that
Correlation rule is utilized described in step (4c), concluding in shield data has incidence relation with each shield attitude parameter section
Tunneling data, realize step are as follows:
(4c1) from correlation rule choose confidence level be greater than pre-set min confidence and support be greater than it is pre-set most
The correlation rule of small support obtains the incidence relation of shield attitude parameter section and tunneling data；
(4c2) according to the incidence relation between shield attitude parameter section and tunneling data, conclude in shield data with each shield
Structure attitude parameter section has the tunneling data of incidence relation, obtains being respectively provided with multiple shield attitude parameters section and is associated with
The multi-group data sample of system.
5. the shield method for correcting error according to claim 1 based on statistical analysis in conjunction with XGboost, which is characterized in that
Each shield attitude parameter section is divided again described in step (4d), implementation method are as follows:
Using the tunneling direction of shield machine to the right as the forward direction of horizontal attitude parameter, upwardly direction tunneling direction is perpendicular attitude ginseng
Several forward direction, in the positive maximum deflection difference value M for the attitude parameter chosen from new shield data1With reversed maximum deflection difference value M2
[- the M formed2, M1] in range, using wide method, as unit of maximum correction amount a again to each shield attitude parameter section
It is divided, obtains shield attitude parameter section: [- M2,-na), [- na ,-(n-1) a) ..., [(n-1) a, na), [na, M1],
Wherein n is coefficient,And n round numbers.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810583493.4A CN108868805B (en) | 2018-06-08 | 2018-06-08 | Shield method for correcting error based on statistical analysis in conjunction with XGboost |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810583493.4A CN108868805B (en) | 2018-06-08 | 2018-06-08 | Shield method for correcting error based on statistical analysis in conjunction with XGboost |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108868805A CN108868805A (en) | 2018-11-23 |
| CN108868805B true CN108868805B (en) | 2019-11-15 |
Family
ID=64338527
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810583493.4A Active CN108868805B (en) | 2018-06-08 | 2018-06-08 | Shield method for correcting error based on statistical analysis in conjunction with XGboost |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108868805B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109766331A (en) * | 2018-12-06 | 2019-05-17 | 中科恒运股份有限公司 | Method for processing abnormal data and device |
| CN109558697A (en) * | 2019-01-16 | 2019-04-02 | 天津大学 | A kind of characteristic parameter selection method and system for the airborne parameter of shield-tunneling construction |
| CN109779649A (en) * | 2019-02-13 | 2019-05-21 | 上海大学 | Shield driving axis real-time deviation correcting system and method based on big data |
| CN110195592B (en) * | 2019-04-30 | 2021-02-05 | 华中科技大学 | Intelligent shield tunneling pose prediction method and system based on hybrid deep learning |
| CN110067568B (en) * | 2019-05-30 | 2020-06-23 | 上海隧道工程有限公司 | Self-adaptive control method and system for shield deviation-correcting oil pressure output |
| 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 |
| CN110067566B (en) * | 2019-05-30 | 2020-06-05 | 上海隧道工程有限公司 | Method and system for predicting shield deviation rectifying moment |
| CN110185463B (en) * | 2019-07-01 | 2020-10-09 | 西安电子科技大学 | Control method for shield tunneling attitude |
| CN110533065A (en) * | 2019-07-18 | 2019-12-03 | 西安电子科技大学 | Based on the shield attitude prediction technique from coding characteristic and deep learning regression model |
| CN111365015A (en) * | 2020-03-05 | 2020-07-03 | 中建交通建设集团有限公司 | Shield tunneling parameter feature extraction and attitude deviation prediction method based on XGboost |
| CN111636891B (en) * | 2020-06-08 | 2022-04-08 | 中铁高新工业股份有限公司 | Real-time shield attitude prediction system and construction method of prediction model |
| CN112100841B (en) * | 2020-09-09 | 2024-04-19 | 中铁二十局集团有限公司 | Method and device for predicting attitude of shield tunneling machine, terminal equipment and storage medium |
| CN112364917B (en) * | 2020-11-10 | 2024-02-13 | 辽宁石油化工大学 | Geological type identification method and device, storage medium and computer equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07189587A (en) * | 1993-12-27 | 1995-07-28 | Sumitomo Heavy Ind Ltd | Method and device for controlling grease supplying pressure for sealed part of shield excavating machine |
| CN1800583A (en) * | 2005-12-05 | 2006-07-12 | 上海市第二市政工程有限公司 | Method and apparatus for real-time automatic correction of shield attitude deviation |
| CN102518446A (en) * | 2012-01-12 | 2012-06-27 | 北京众策创晟掘进装备有限公司 | Automatic deviation rectifying method of shield tunneling machine |
| CN102996137A (en) * | 2012-12-31 | 2013-03-27 | 上海同岩土木工程科技有限公司 | S-shaped curve based method for control track of shield tunneling machine |
| CN105736007A (en) * | 2016-03-18 | 2016-07-06 | 济南轨道交通集团有限公司 | Shield tunneling machine positioning and deviation correcting system and method based on stratum information integration |
| CN106815369A (en) * | 2017-01-24 | 2017-06-09 | 中山大学 | A kind of file classification method based on Xgboost sorting algorithms |
-
2018
- 2018-06-08 CN CN201810583493.4A patent/CN108868805B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH07189587A (en) * | 1993-12-27 | 1995-07-28 | Sumitomo Heavy Ind Ltd | Method and device for controlling grease supplying pressure for sealed part of shield excavating machine |
| CN1800583A (en) * | 2005-12-05 | 2006-07-12 | 上海市第二市政工程有限公司 | Method and apparatus for real-time automatic correction of shield attitude deviation |
| CN102518446A (en) * | 2012-01-12 | 2012-06-27 | 北京众策创晟掘进装备有限公司 | Automatic deviation rectifying method of shield tunneling machine |
| CN102996137A (en) * | 2012-12-31 | 2013-03-27 | 上海同岩土木工程科技有限公司 | S-shaped curve based method for control track of shield tunneling machine |
| CN105736007A (en) * | 2016-03-18 | 2016-07-06 | 济南轨道交通集团有限公司 | Shield tunneling machine positioning and deviation correcting system and method based on stratum information integration |
| CN106815369A (en) * | 2017-01-24 | 2017-06-09 | 中山大学 | A kind of file classification method based on Xgboost sorting algorithms |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108868805A (en) | 2018-11-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108868805B (en) | Shield method for correcting error based on statistical analysis in conjunction with XGboost | |
| US11162349B2 (en) | Systems and methods for geosteering during well drilling | |
| WO2019042483A2 (en) | Tbm boring rock state real-time sensing system and method | |
| CN106437512B (en) | A kind of shale gas horizontal well landing path drilling tracking regulation and control method | |
| CN105631753B (en) | Horizontal well oil reservoir profile modeling method based on stratum attitude | |
| CN116522692B (en) | Underground engineering surrounding rock structural feature in-situ detection and classification method | |
| CN105484735A (en) | Method for evaluating coincidence rate of actual drilling borehole trajectory and design track | |
| CN108763752A (en) | The method that water channel determines construction boring parameter is worn under a kind of shield tunnel | |
| CN116341294B (en) | Three-dimensional stress field construction method and device | |
| CN104899928A (en) | Three-directional geology modeling method based on sparse borehole points | |
| CN108952657A (en) | Horizontal well platform fracturing fracture length determination method and device | |
| CN111192359B (en) | Method for extracting borehole lithology data by using geological section borehole space track morphology | |
| CN103590828B (en) | Method for evaluating formation pressure by logging dck index method | |
| CN106285479B (en) | Three-dimensional GIS technology-based borehole trajectory control method | |
| CN112489162B (en) | Method for forecasting coal seam geology and drawing sectional view of large-scale micro-unit coal seam | |
| CN107784437B (en) | Stress concentration-based coal roadway driving face outburst danger identification method | |
| CN107103165B (en) | Method for calculating density of steep dip shear crack line of oil field reservoir | |
| CN108242069B (en) | Oil reservoir profile compiling and drawing method based on seismic profile and time-depth relation | |
| CN107481298B (en) | Oil reservoir profile compiling method comprising inclined well | |
| CN107917692A (en) | A kind of block identification method based on unmanned plane | |
| CN113742962A (en) | Shale reservoir attribute three-dimensional modeling method based on horizontal well | |
| CN105626045A (en) | Method and device for determining shape of well bore | |
| CN105804723A (en) | Downhole tool depth measurement method for resource exploration and development | |
| CN206960672U (en) | A kind of wireless controlled passes the Tunnel prediction system of acceleration transducer | |
| CN105484732B (en) | Processing method for horizontal well drilling geosteering work progress well depth |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200730 Address after: Room 1506, No. 69, west section of south 2nd Ring Road, Beilin District, Xi'an City, Shaanxi Province Patentee after: Xi'an Qigong Data Technology Co.,Ltd. Address before: 710071, No. 2 Taibai South Road, Yanta District, Shaanxi, Xi'an Patentee before: XIDIAN University |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221220 Address after: 710000 No. 702-14, Building 106, Xi'an Jiawei Science and Technology Industrial Park, No. 99, Hangtian East Road, Xi'an National Civil Aerospace Industrial Base, Shaanxi Province Patentee after: Shaanxi Hanguang Digital Technology Co.,Ltd. Address before: 710068 Room 1506, No. 69, West Section of South Second Ring Road, Beilin District, Xi'an, Shaanxi Patentee before: Xi'an Qigong Data Technology Co.,Ltd. |