CN109978226A - Shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network - Google Patents
Shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network Download PDFInfo
- Publication number
- CN109978226A CN109978226A CN201910068462.XA CN201910068462A CN109978226A CN 109978226 A CN109978226 A CN 109978226A CN 201910068462 A CN201910068462 A CN 201910068462A CN 109978226 A CN109978226 A CN 109978226A
- Authority
- CN
- China
- Prior art keywords
- layer
- recognition
- neural network
- recurrent neural
- network
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 42
- 238000010276 construction Methods 0.000 title claims abstract description 41
- 230000000306 recurrent effect Effects 0.000 title claims abstract description 41
- 238000004062 sedimentation Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000005259 measurement Methods 0.000 claims description 9
- 210000002569 neuron Anatomy 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 4
- 239000003673 groundwater Substances 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 2
- 239000000284 extract Substances 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 description 12
- 238000009412 basement excavation Methods 0.000 description 6
- 239000011435 rock Substances 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 230000000644 propagated effect Effects 0.000 description 4
- 241001269238 Data Species 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 239000002689 soil Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007596 consolidation process Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005538 encapsulation Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000001125 extrusion Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 239000011178 precast concrete Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 230000005641 tunneling Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
-
- 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
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Tourism & Hospitality (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Excavating Of Shafts Or Tunnels (AREA)
Abstract
The shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network that the present invention relates to a kind of, choose the principal element that shield construction causes surface subsidence, and merge the sedimentation situation at current time, collectively as input data, surface subsidence is effectively predicted in the middle deep-neural-network based on Recognition with Recurrent Neural Network established through the invention.The present invention gives detailed model foundation and training methods, situation data will be settled first to propagate by the time channel of Recognition with Recurrent Neural Network, then again by major influence factors with its it is organic merge, pass through deep layer network layer extracted in self-adaptive feature, finally obtain settlement prediction result.Therefore, present invention precision of prediction with higher, and generalization ability is strong, can put into the practical application of shield tunnel excavated earth settlement prediction.
Description
Technical field
The present invention relates to shield-tunneling construction field of intelligent monitoring, and in particular to a kind of shield-tunneling construction based on Recognition with Recurrent Neural Network
Surface subsidence prediction technique.
Background technique
Shield method is the Fully-mechanized construction method of one of shallow mining method, it is to promote Shield-type machinery in ground,
Surrounding country rock is supported by shield shell and section of jurisdiction and prevents the collapsing in tunnel, while with cutting dress in front of excavation face
Carry out soil excavation is set, by being unearthed, machinery is transported outside hole, by jack in rear pressurization jacking, and assembled precast concrete pipe
Piece forms a kind of mechanized construction method of tunnel structure.
Ground settlement refers to because excavation causes the extrusion of surrounding rock in excavation, or because collapsing causes country rock relaxation, excavates back wall
Gap between rock and supporting, country rock and lining cutting causes to consolidate because level of ground water declines, and makes the reasons such as supporting sinking in weak surrounding rock,
Caused surface subsidence phenomenon.
Surface subsidence caused by shield driving includes five stages: sedimentation, shield machine in front of initial sedimentation, excavation face
Sedimentation, the sedimentation and final consolidation settlement in shield tail gap when passing through.In Analysis on Shield Tunnel Driven Process, always inevitably draw
The loosening and depression for playing construction tunnel surrounding soil, cause the loss on stratum, so as to cause the sedimentation of earth's surface.When sedimentation is more than one
When determining range, will affect along building stabilization, more serious meeting so that tunnel cave, cause it is inestimable it is serious after
Fruit.As it can be seen that being very significant to the real-time Predicting Technique expansion research of surface subsidence in shield tunnel mining process.
Chinese invention CN107239599A is with disclosing caused by a kind of shield-tunneling construction based on neural fuzzy inference system
Table settlement prediction method, this method choose the major influence factors of Surface Settlement Resulted by Shield Tunneling, and these are mainly influenced
Factor forms effective data set, by the Ground surface settlement model built using neural fuzzy inference system, to the significant figure
It is calculated according to collection, to obtain prediction result and carry out model verifying according to the prediction result.But the invention only accounts for shadow
The factor of sound, without considering influence of the ground settlement occurred to subsequent ground settlement variation tendency.
Chinese invention CN107092990A discloses a kind of shield construction ground settlement prediction system based on big data analysis
System and method, use big data platform module for data collection, data prediction, feature extraction, establish prediction model and model
Using etc. modules mass data storage and parallel computation service are provided, form surface subsidence forecasting system.Prediction technique step packet
It includes: building big data platform；Collect shield construction ground settling data；Data prediction；Feature extraction；It is pre- to establish surface subsidence
Survey model；The encapsulation of prediction model functional interface；Surface subsidence prediction is carried out in work progress.But the invention only proposes to establish pre-
The step of examining system, is without going into seriously the algorithm of specific implementation.
Summary of the invention
The shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network that the object of the present invention is to provide a kind of,
By the real time data and construction parameter generated in Analysis on Shield Tunnel Driven Process, based on Recognition with Recurrent Neural Network technology to surface subsidence
Predicted, quickly to handle mass data and obtain prediction result, there is very high practicability.
In order to achieve the above objectives, the present invention provides the following technical scheme that
A kind of shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network, is summarized as follows:
(1), truthful data is collected, data set is established；
(2), the shield construction ground Settlement Model based on Recognition with Recurrent Neural Network is established；
(3), training pattern obtains all hyper parameters；
(4), model, investment actual use are verified.
During practice of construction, there are many influence factor of surface subsidence, and tunnel location, geological conditions, shield parameter all can
Surface subsidence situation is largely influenced, this is the result of one and its complicated each combined factors effect.Recognition with Recurrent Neural Network
As one of the algorithm having outstanding performance in depth learning technology, there is very powerful nonlinear fitting ability, to time series
The processing capacity of data is even more that other models are incomparable.Moreover, the training process of model is exactly the determination process of hyper parameter,
By the study to a large amount of truthful datas, each hyper parameter of model can constantly be optimized, to provide more accurate ground
Face settlement prediction value.
Further, overall plan of the invention is described in detail below:
(1) truthful data is collected, data set is established.
This is the basis of the training patterns of a large amount of complete truthful datas, this method propose based on Recognition with Recurrent Neural Network
Shield construction ground settlement prediction method further includes the tunnel of each time point other than collecting the sedimentation value changed over time
Position, geological conditions and shield parameter.Wherein, tunnel location includes tunnel depth and driving distance；Geological conditions includes
Tunnel top geology, tunnel bottom geology and level of ground water；Shield parameter include face pressure, fltting speed, pitch angle,
Grouting at the tail of the shield machine pressure and grouting at the tail of the shield machine filling rate.These data collections divide data set by the requirement of mode input after, often
A sample data further includes cut-off newest ten settlement measurement data of current time, altogether other than including above-mentioned ten attributes
With the input data as model.
(2) the shield construction ground Settlement Model based on Recognition with Recurrent Neural Network is established.
The model shares five layers, and each layer of output is exactly next layer of input.First layer is Recognition with Recurrent Neural Network layer,
Input is ten settlement measurement data in step (1) sample data, and what is obtained is the result after time shaft acts on；Second
Layer realizes the dual function of straight articulamentum and input layer, after the output of first layer is handled with ten attribute values in sample data
Merge, collectively as the output of this layer；Third layer to layer 5 is straight articulamentum, and number of nodes is respectively 20,10 and 3, passes through this
Three layers are gradually compressed feature quantity, until the output of last layer 5 is the predicted value of rear three ground settlement values.
(3) training pattern obtains all hyper parameters.
The training of shield construction ground Settlement Model based on Recognition with Recurrent Neural Network described in step (2), need to use based on when
Between back-propagation algorithm and back-propagation algorithm；Its training process is exactly to constantly update hyper parameter in the hope of obtaining higher prediction
The process of precision, comprising the following steps:
1. a sample data is inputted, by the established network of shield construction ground Settlement Model based on Recognition with Recurrent Neural Network
Propagated forward is primary, and output valve is calculated；
2. it is primary by the established network backpropagation of shield construction ground Settlement Model based on Recognition with Recurrent Neural Network, wherein
Straight articulamentum uses back-propagation algorithm, and first layer Recognition with Recurrent Neural Network layer uses time-based back-propagation algorithm, calculates
Obtain the gradient value of each hyper parameter；
3. updating the value of each hyper parameter, renewal process uses Adam optimization algorithm；
4. judging whether model accuracy meets the requirements at this time, if then training finishes, otherwise returns to the and 1. walk.
(4) model, investment actual use are verified.
After training, with mark off come test the set pair analysis model tested, with determine gained model in training set
On whether still have a satisfactory performance.After the model obtained by determination has certain generalization really, reality can be put into
Shield construction ground settlement prediction task in.
Compared with prior art, the beneficial effects of the present invention are:
The present invention devises a kind of shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network, and this method is chosen
Shield tunnel excavate in tunnel location, geological conditions and shield parameter, and sedimentation measurement by current time is
Input data continues to optimize model parameter, passes through test by the structure of circulation layer and multiple straight articulamentums voluntarily learning characteristic
After the verifying of collection, the Forecasting Model of Land Subsidence that Practical Project uses can be put by finally obtaining.Compared to existing conventional machines
Learning method, this method have merged the great influence parameter in shield-tunneling construction based on more massive truthful data, in advance
Survey precision is higher, and generalization ability is stronger.
Detailed description of the invention
Implement sample or technical solution in the prior art in order to illustrate more clearly of the present invention, it below will be to implementation sample
Or attached drawing needed to be used in the description of the prior art is briefly described, it should be apparent that, the accompanying drawings in the following description is this
The implementation sample of invention for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the shield construction ground settlement prediction method flow provided in an embodiment of the present invention based on Recognition with Recurrent Neural Network
Figure；
Fig. 2 is the signal of the shield construction ground Settlement Model provided in an embodiment of the present invention based on Recognition with Recurrent Neural Network
Figure.
Specific embodiment
To keep technical solution and the advantage of present invention implementation sample clearer, implement in sample below in conjunction with the present invention
Attached drawing, to the present invention implement sample in technical solution be clearly and completely described.
As shown in Figure 1, a kind of shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network, including walk as follows
It is rapid:
(1) truthful data is collected, data set is established.
It chooses to the key factor for causing surface subsidence in construction, and sedimentation measured value, collectively constitutes more efficiently
Data set, specifically, these key factors are tunnel depth, driving distance, tunnel top geology, tunnel bottom geology, underground
Water level, face pressure, fltting speed, pitch angle, grouting at the tail of the shield machine pressure and grouting at the tail of the shield machine filling rate, measured settlement are cut-off
Nearest ten sedimentation measurements of current point in time.
It is assumed that nearest ten sedimentation measurements for ending current time are used respectively in t momentIt indicates, on
The key factor for stating ten influence surface subsidence is used respectivelyIt indicates, t moment backward distinguish three times by sedimentation measurement
WithIt indicates, then sample set X and tally set Y comprising n sample are equal to:
(2) the shield construction ground Settlement Model based on Recognition with Recurrent Neural Network is established.
As shown in Fig. 2, the model shares five layers, each layer of output is exactly next layer of input.First layer is circulation mind
Through network structure, input as nearly ten sedimentation measurements, export for it is after time shaft is propagated as a result, its calculation formula is:
Wherein,
o10=0
In formula,
o1i-- the output valve of i-th of neuron of first layer；
xi-- the input value of i-th of neuron；
w11-- along the weight of time channel；
w12-- the weight of the first layer network.
The output of first layer Recognition with Recurrent Neural Network structure is applied ReLU activation primitive by second layer network structure, while by shadow
Ring ten key factors output in combination as this layer of surface subsidence, it may be assumed that
In formula,
o2i-- the output valve of i-th of neuron of the second layer；
w2The weight of-the second layer network.
Third layer to layer 5 is straight articulamentum, gradually extracts main feature by this three-layer network, reduces neuron number
Amount, finally obtains the predicted value of subsequent surface subsidence three times.This three layers calculating process is as follows:
In formula,
o3The output valve of-third layer；
o4- the four layer of output valve；
o5The output valve of-layer 5；
w3The weight of-third layer network；
b3The biasing of-third layer network.
Wherein,
In formula,
xijThe i-th row jth column element of-matrix x.
(3) training pattern obtains all hyper parameters.
1. after obtaining the output valve o of propagated forward, using mean square error loss function come the prediction effect of Evaluation model,
Its calculation formula is as follows:
In formula,
L-mean square error penalty values.
2. be utilized respectively back-propagation algorithm and time-based back-propagation algorithm calculate each layer hyper parameter gradient it is big
It is small.Back propagation is applied to straight articulamentum, and time-based back-propagation algorithm is applied to Recognition with Recurrent Neural Network layer.It is direct-connected to connect
Layer passes through the gradient value of the available parameters of chain ruleAndCirculation layer gradient calculation formula is as follows:
It enablesSo second layer gradient are as follows:
First layer gradient are as follows:
Wherein,
3. parameter update is carried out using Adam optimization algorithm, wherein directly after obtaining all gradient values of each network layer
What articulamentum needed to update has weighted value and bias term, and it is weighted value that Recognition with Recurrent Neural Network layer needs, which update, and Adam optimization is calculated
The calculating process of method is as follows:
Vdw=β1Vdw+(1-β1)dw,Vdb=β1Vdb+(1-β1)db
Sdw=β2Sdw+(1-β2)(dw)2,Sdb=β2Sdb+(1-β2)(db)2
Wherein,
β1=0.9
β2=0.999
ε=10-8
In formula,
The gradient value of dw-- weight w；
The gradient value of db-- biasing b；
Vdw,Sdw-- the exponent-weighted average number of dw；
Vdb,Sdb-- the exponent-weighted average number of db；
Exponent-weighted average number amendment；
Exponent-weighted average number amendment；
So far primary complete propagated forward, back-propagating and parameter renewal process are just completed, later with next group
Training data continues iteration to update all hyper parameters, until completing a wheel iteration of all trained training datas.Here instruction
Practice data and account for about the 90% of total number of samples, remainder 10% is tested as test the set pair analysis model.
4. updating when parameter is repeated with training data until after meeting preset required precision, stop iteration,
The parameter obtained at this time is the model parameter after the completion of training.
(4) model, investment actual use are verified.
It is tested later with test the set pair analysis model, update of the test process without parameter, only judgment models precision is
It is no to meet the requirements, guarantee that model has good generalization with this, can put into practical shield tunnel excavation and carry out
The prediction work of surface subsidence.
Foregoing description is only the description to present pre-ferred embodiments, is not any restriction to the scope of the invention.Appoint
Any change or modification what those skilled in the art makes according to the technology contents of the disclosure above should all regard
For equivalent effective embodiment, the range of technical solution of the present invention protection is belonged to.
Claims (9)
1. a kind of shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network, which is characterized in that using tunnel location,
Geological conditions, shield parameter and the comprehensive consideration for settling situation realize ground using deep layer network model extracted in self-adaptive feature
The prediction of face sedimentation, including following procedure:
(1), truthful data is collected, data set is established；
(2), the shield construction ground Settlement Model based on Recognition with Recurrent Neural Network is established；
(3), training pattern obtains all hyper parameters；
(4), model, investment actual use are verified.
2. the shield construction ground settlement prediction method according to claim 1 based on Recognition with Recurrent Neural Network, feature exist
In: data set described in step (1), from the principal element for influencing surface subsidence, including the sedimentation situation changed over time
And tunnel location, geological conditions and the shield parameter of each time point；
The tunnel location includes tunnel depth and driving distance；
The geological conditions includes tunnel top geology, tunnel bottom geology and level of ground water；
The shield parameter includes face pressure, fltting speed, pitch angle, grouting at the tail of the shield machine pressure and grouting at the tail of the shield machine filling
Rate；
The sedimentation situation includes cut-off newest ten settlement measurement data of current time.
3. the shield construction ground settlement prediction method according to claim 2 based on Recognition with Recurrent Neural Network, feature exist
In the shield construction ground Settlement Model described in step (2) based on Recognition with Recurrent Neural Network includes:
First layer is Recognition with Recurrent Neural Network layer, is inputted to settle situation, i.e. cut-off newest ten settlement measurement numbers of current time
According to extracting its characteristic value by the propagation of time channel；
The second layer, for after the output of first layer is handled with the tunnel depth, driving distance, tunnel top geology, tunnel
Bottom geology, level of ground water, face pressure, fltting speed, pitch angle, grouting at the tail of the shield machine pressure and grouting at the tail of the shield machine filling rate ten
Attribute value merges, collectively as the output of this layer；
Third layer, is straight articulamentum, and number of nodes 20 extracts the characteristic value completely inputted after merging；
4th layer, be straight articulamentum, number of nodes 10, compressive features quantity；
Layer 5 is straight articulamentum, and number of nodes 3 further compresses feature quantity, obtains required surface subsidence prediction
Value.
4. the shield construction ground settlement prediction method according to claim 3 based on Recognition with Recurrent Neural Network, feature exist
In the Recognition with Recurrent Neural Network layer of the first layer is incorporated in processing sedimentation situation data, and propagates along time channel, function
Are as follows:
In formula,
o1iFor the output valve of i-th of neuron of first layer,
For ReLU activation primitive,
xi-- the input value of i-th of neuron；
w11-- along the weight of time channel；
w12-- the weight of the first layer network.
5. the shield construction ground settlement prediction method according to claim 3 based on Recognition with Recurrent Neural Network, feature exist
In will be by the processed sedimentation situation data of Recognition with Recurrent Neural Network layer and surface subsidence major influence factors in the second layer
Ten attributes merge, form new input data, function are as follows:
In formula,
o2i-- the output valve of i-th of neuron of the second layer；
w2-- the weight of the second layer network；
xi-- the input value of i-th of neuron.
6. the shield construction ground settlement prediction method according to claim 3 based on Recognition with Recurrent Neural Network, feature exist
In, last Three Tiered Network Architecture is straight articulamentum, feature quantity is gradually reduced, the surface subsidence predicted value after obtaining three times,
Its function are as follows:
In formula,
o3The output valve of-third layer；
o4- the four layer of output valve；
o5The output valve of-layer 5；
w3The weight of-third layer network；
w4The weight of-the four-layer network network；
w5The weight of-layer 5 network；
b3The biasing of-third layer network；
b4The biasing of-the four-layer network network；
b5The biasing of-layer 5 network.
7. the shield construction ground settlement prediction method according to claim 3 based on Recognition with Recurrent Neural Network, feature exist
In, in model training, prediction effect is judged using mean square error loss function, and back-propagation algorithm is used to straight articulamentum,
Time-based back-propagation algorithm, mean square error loss function are used to Recognition with Recurrent Neural Network layer are as follows:
In formula,
L-mean square error penalty values.
8. the shield construction ground settlement prediction method according to claim 7 based on Recognition with Recurrent Neural Network, feature exist
Adam optimization algorithm is used when, parameter updates, wherein parameter beta1=0.9, β2=0.999, ε=10-8。
9. the shield construction ground settlement prediction method according to claim 8 based on Recognition with Recurrent Neural Network, feature exist
In being tested with test set gained model after completing model training, determine that it is real with putting into after good generalization
Border engineering uses.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910068462.XA CN109978226B (en) | 2019-01-24 | 2019-01-24 | Shield construction ground settlement prediction method based on cyclic neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910068462.XA CN109978226B (en) | 2019-01-24 | 2019-01-24 | Shield construction ground settlement prediction method based on cyclic neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109978226A true CN109978226A (en) | 2019-07-05 |
| CN109978226B CN109978226B (en) | 2023-02-28 |
Family
ID=67076686
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910068462.XA Active CN109978226B (en) | 2019-01-24 | 2019-01-24 | Shield construction ground settlement prediction method based on cyclic neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109978226B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110617074A (en) * | 2019-09-20 | 2019-12-27 | 西安电子科技大学 | Incidence relation method for ground settlement and tunneling parameters in shield construction |
| CN111119902A (en) * | 2019-12-16 | 2020-05-08 | 北京科技大学 | Tunnel dynamic construction method based on BP neural network |
| CN111460737A (en) * | 2020-04-09 | 2020-07-28 | 昆山阳翎机器人科技有限公司 | Intelligent settlement prediction method and system for slurry air pressure balance shield |
| CN111832223A (en) * | 2020-06-29 | 2020-10-27 | 上海隧道工程有限公司 | Neural network-based shield construction surface subsidence prediction method |
| CN112364422A (en) * | 2020-11-13 | 2021-02-12 | 中铁二十局集团有限公司 | Shield construction earth surface deformation dynamic prediction method based on MIC-LSTM |
| CN113128106A (en) * | 2021-04-06 | 2021-07-16 | 汕头大学 | Method for determining surface subsidence caused by shield construction of karst stratum |
| CN113204824A (en) * | 2021-05-21 | 2021-08-03 | 上海大学 | Multi-model fusion shield construction settlement prediction method and system |
| CN113239439A (en) * | 2021-05-21 | 2021-08-10 | 上海大学 | Shield construction ground surface settlement prediction system and method |
| CN113298220A (en) * | 2021-05-31 | 2021-08-24 | 中铁十六局集团北京轨道交通工程建设有限公司 | Neural network optimization-based shield tunneling machine tunneling speed prediction method |
| CN113486818A (en) * | 2021-07-09 | 2021-10-08 | 吉林大学 | Full fighting rate prediction system and method based on machine vision |
| CN113931636A (en) * | 2021-10-22 | 2022-01-14 | 广州地铁设计研究院股份有限公司 | Tunnel underpass existing operation subway line grouting construction parameter selection method and application thereof |
| CN118070567A (en) * | 2024-04-18 | 2024-05-24 | 安徽省交通控股集团有限公司 | Shield speed control system and method based on quantitative simulation of tunnel construction environment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107239599A (en) * | 2017-05-16 | 2017-10-10 | 五邑大学 | Based on Ground surface settlement method caused by the shield-tunneling construction of neural fuzzy inference system |
| CN109242171A (en) * | 2018-08-28 | 2019-01-18 | 河南省豫晋高速公路建设有限公司 | A kind of shield-tunneling construction Ground surface settlement method based on BIM and RS-SVR |
-
2019
- 2019-01-24 CN CN201910068462.XA patent/CN109978226B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107239599A (en) * | 2017-05-16 | 2017-10-10 | 五邑大学 | Based on Ground surface settlement method caused by the shield-tunneling construction of neural fuzzy inference system |
| CN109242171A (en) * | 2018-08-28 | 2019-01-18 | 河南省豫晋高速公路建设有限公司 | A kind of shield-tunneling construction Ground surface settlement method based on BIM and RS-SVR |
Non-Patent Citations (1)
| Title |
|---|
| 许宁等: "改进型LSTM变形预测模型研究", 《江西理工大学学报》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110617074A (en) * | 2019-09-20 | 2019-12-27 | 西安电子科技大学 | Incidence relation method for ground settlement and tunneling parameters in shield construction |
| CN111119902A (en) * | 2019-12-16 | 2020-05-08 | 北京科技大学 | Tunnel dynamic construction method based on BP neural network |
| CN111119902B (en) * | 2019-12-16 | 2021-04-06 | 北京科技大学 | Tunnel dynamic construction method based on BP neural network |
| CN111460737A (en) * | 2020-04-09 | 2020-07-28 | 昆山阳翎机器人科技有限公司 | Intelligent settlement prediction method and system for slurry air pressure balance shield |
| CN111460737B (en) * | 2020-04-09 | 2023-12-29 | 昆山阳翎机器人科技有限公司 | Intelligent settlement prediction method and system for slurry air pressure balance shield |
| CN111832223A (en) * | 2020-06-29 | 2020-10-27 | 上海隧道工程有限公司 | Neural network-based shield construction surface subsidence prediction method |
| CN112364422B (en) * | 2020-11-13 | 2023-06-13 | 中铁二十局集团有限公司 | MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation |
| CN112364422A (en) * | 2020-11-13 | 2021-02-12 | 中铁二十局集团有限公司 | Shield construction earth surface deformation dynamic prediction method based on MIC-LSTM |
| CN113128106A (en) * | 2021-04-06 | 2021-07-16 | 汕头大学 | Method for determining surface subsidence caused by shield construction of karst stratum |
| CN113128106B (en) * | 2021-04-06 | 2023-03-21 | 汕头大学 | Method for determining surface subsidence caused by shield construction of karst stratum |
| CN113239439A (en) * | 2021-05-21 | 2021-08-10 | 上海大学 | Shield construction ground surface settlement prediction system and method |
| CN113204824A (en) * | 2021-05-21 | 2021-08-03 | 上海大学 | Multi-model fusion shield construction settlement prediction method and system |
| CN113239439B (en) * | 2021-05-21 | 2022-04-05 | 上海大学 | Shield construction ground surface settlement prediction system and method |
| CN113298220A (en) * | 2021-05-31 | 2021-08-24 | 中铁十六局集团北京轨道交通工程建设有限公司 | Neural network optimization-based shield tunneling machine tunneling speed prediction method |
| CN113298220B (en) * | 2021-05-31 | 2023-08-04 | 中铁十六局集团北京轨道交通工程建设有限公司 | Neural network optimization-based shield tunneling speed prediction method |
| CN113486818A (en) * | 2021-07-09 | 2021-10-08 | 吉林大学 | Full fighting rate prediction system and method based on machine vision |
| CN113486818B (en) * | 2021-07-09 | 2022-05-20 | 吉林大学 | Full fighting rate prediction system and method based on machine vision |
| CN113931636A (en) * | 2021-10-22 | 2022-01-14 | 广州地铁设计研究院股份有限公司 | Tunnel underpass existing operation subway line grouting construction parameter selection method and application thereof |
| CN113931636B (en) * | 2021-10-22 | 2024-05-07 | 广州地铁设计研究院股份有限公司 | Tunnel underpass existing operation subway line grouting construction parameter selection method and application thereof |
| CN118070567A (en) * | 2024-04-18 | 2024-05-24 | 安徽省交通控股集团有限公司 | Shield speed control system and method based on quantitative simulation of tunnel construction environment |
| CN118070567B (en) * | 2024-04-18 | 2024-07-23 | 安徽省交通控股集团有限公司 | Shield speed control system and method based on quantitative simulation of tunnel construction environment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109978226B (en) | 2023-02-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109978226A (en) | Shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network | |
| Li et al. | Evolution characteristics and displacement forecasting model of landslides with stair-step sliding surface along the Xiangxi River, three Gorges Reservoir region, China | |
| Xue et al. | Analysis of factors influencing tunnel deformation in loess deposits by data mining: a deformation prediction model | |
| Xue et al. | A two-step comprehensive evaluation model for rockburst prediction based on multiple empirical criteria | |
| Yoo et al. | Tunneling performance prediction using an integrated GIS and neural network | |
| Reddy | An empirical study on the best prediction model of water inflow into drill and blast tunnels among several machine learning techniques | |
| CN102561330B (en) | Integrated construction method for artificial dug pile and steel pipe concrete column | |
| CN111119902B (en) | Tunnel dynamic construction method based on BP neural network | |
| CN108842821B (en) | Calculation method for reasonable buried depth of submarine tunnel constructed by drilling and blasting method | |
| Sundell et al. | A probabilistic approach to soil layer and bedrock-level modeling for risk assessment of groundwater drawdown induced land subsidence | |
| CN105421335B (en) | The anti-liquifying method of cement mixing pile composite foundation based on place excess pore water pressure ratio | |
| Nguyen et al. | Prediction of ground subsidence due to underground mining through time using multilayer feed-forward artificial neural networks and back-propagation algorithm–case study at Mong Duong underground coal mine (Vietnam) | |
| Polomčić et al. | Application of Groundwater modeling for designing a dewatering system: Case study of the Buvač Open Cast Mine, Bosnia and Herzegovina | |
| CN116882023A (en) | Method for predicting transverse settlement of upper earth covering layer in underground excavation construction of subway tunnel | |
| CN116756807A (en) | Linear underground engineering water inflow numerical prediction method in coupling construction process | |
| CN115099111B (en) | Shield construction ground surface settlement prediction method based on machine learning and finite element | |
| CN110851991A (en) | Underground water flow numerical simulation method | |
| Javadi | Estimation of air losses in compressed air tunneling using neural network | |
| Van Hoa | Rational Pile Design using Computer-based Program Coding in Matlab: A Case Study | |
| CN110080197B (en) | Lanzhou red sandstone classification method and corresponding engineering measure selection method | |
| CN114066271A (en) | Tunnel water inrush disaster monitoring and management system | |
| Zhao et al. | A modified back analysis method for deep excavation with multi-objective optimization procedure | |
| Mobarra et al. | Application of artificial neural networks to the prediction of TBM penetration rate in TBM-driven golab water transfer tunnel | |
| Yu et al. | An artificial intelligence optimization method of back analysis of unsteady-steady seepage field for the dam site under complex geological condition | |
| CN113627761B (en) | Parallel evaluation method for geotechnical engineering water inrush probability prediction |
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 |