CN112364422B - MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation - Google Patents
MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation Download PDFInfo
- Publication number
- CN112364422B CN112364422B CN202011268712.3A CN202011268712A CN112364422B CN 112364422 B CN112364422 B CN 112364422B CN 202011268712 A CN202011268712 A CN 202011268712A CN 112364422 B CN112364422 B CN 112364422B
- Authority
- CN
- China
- Prior art keywords
- shield
- construction
- deformation
- segment ring
- construction area
- 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
- 238000010276 construction Methods 0.000 title claims abstract description 451
- 238000000034 method Methods 0.000 title claims abstract description 133
- 238000003062 neural network model Methods 0.000 claims abstract description 67
- 230000008569 process Effects 0.000 claims abstract description 51
- 238000012549 training Methods 0.000 claims abstract description 50
- 238000013461 design Methods 0.000 claims abstract description 8
- 238000005259 measurement Methods 0.000 claims description 76
- 238000012544 monitoring process Methods 0.000 claims description 62
- 238000012360 testing method Methods 0.000 claims description 39
- 239000002689 soil Substances 0.000 claims description 38
- 230000005641 tunneling Effects 0.000 claims description 30
- 238000007781 pre-processing Methods 0.000 claims description 18
- 230000003068 static effect Effects 0.000 claims description 16
- 230000006835 compression Effects 0.000 claims description 15
- 238000007906 compression Methods 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 12
- 238000010586 diagram Methods 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 abstract description 6
- 238000004364 calculation method Methods 0.000 description 5
- 230000004913 activation Effects 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 2
- 238000007596 consolidation process Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 241000512668 Eunectes Species 0.000 description 1
- 101100523065 Hydrogenovibrio crunogenus (strain DSM 25203 / XCL-2) pyrF gene Proteins 0.000 description 1
- 208000033641 Ring chromosome 5 syndrome Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 230000001808 coupling effect Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000001125 extrusion Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000004062 sedimentation Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH 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
- E21D9/0607—Making by using a driving shield, i.e. advanced by pushing means bearing against the already placed lining the shield being provided with devices for lining the tunnel, e.g. shuttering
-
- 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
- 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—Systems or methods specially adapted for 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Abstract
The invention discloses a dynamic prediction method for shield construction earth surface deformation based on MIC-LSTM, which comprises the following steps: 1. in the shield construction process, obtaining the earth surface deformation of the construction area where the I shield segment rings are positioned and the earth surface deformation influence factors; 2. training an LSTM neural network model; 3. and predicting the earth surface deformation of each shield segment in the construction process of the construction area where the shield segment is positioned after the I shield segments according to the trained LSTM neural network model. The method has simple steps and reasonable design, realizes the weight assignment of each influence factor of the deformation of the earth surface of the shield construction so as to distinguish the importance degree of each influence factor on the prediction model, adopts the LSTM neural network to carry out modeling dynamic prediction, and improves the prediction precision and efficiency of the deformation of the earth surface of the shield construction.
Description
Technical Field
The invention belongs to the technical field of shield construction, and particularly relates to a dynamic prediction method for shield construction earth surface deformation based on MIC-LSTM.
Background
The shield construction is a construction technology widely applied in the tunnel excavation process, and the construction technology is easy to cause extrusion or relaxation of surrounding soil bodies due to stratum excavation in the construction process, is easy to cause stratum loss, and further causes the problem of surface deformation. After the surface deformation exceeds a certain safety value, uneven deformation of buildings along the line and damage of underground pipelines are easily caused, and unsafe accidents such as collapse and the like of excavated tunnels can be even caused. Therefore, it is important to grasp the deformation of the earth surface in real time in the shield tunneling process, which makes the problem of predicting the deformation of the earth surface an important issue to be concerned.
At present, the prediction method for the surface deformation mainly comprises a numerical method, a formula method, an analysis method, an intelligent algorithm and the like. With the continuous development of computer level and various deep learning algorithms, the intelligent algorithm is superior to other methods in accuracy and timeliness in solving the problem of earth surface deformation caused by shield construction. However, when the earth surface deformation caused by shield construction is predicted by using some intelligent algorithms, only main factors influencing the earth surface deformation are considered, and the influence of the deformation at the current moment on the earth surface deformation prediction is not considered yet, so that the method belongs to the category of static prediction. In addition, some intelligent algorithms do not distinguish importance degrees of factors when considering main factors influencing the surface deformation, and the coupling relation among the factors can possibly cause information overlapping, so that the calculation amount of the prediction model is increased.
Therefore, a dynamic prediction method for deformation of the shield construction earth surface based on MIC-LSTM is needed at present, weight assignment of each influence factor of deformation of the shield construction earth surface is realized, importance degree of each influence factor to a prediction model is distinguished, modeling dynamic prediction is carried out by adopting an LSTM neural network, and prediction precision and efficiency of deformation of the shield construction earth surface are improved.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides the dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM, which has the advantages of simple steps and reasonable design, realizes the weight assignment of each influence factor of the deformation of the shield construction earth surface, distinguishes the importance degree of each influence factor on a prediction model, adopts an LSTM neural network to carry out modeling dynamic prediction, and improves the prediction precision and efficiency of the deformation of the shield construction earth surface.
In order to solve the technical problems, the invention adopts the following technical scheme: the dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of:
step one, in the shield construction process, obtaining the earth surface deformation of the construction area where the I shield segment rings are positioned and earth surface deformation influence factors; the earth surface deformation influencing factors comprise geological parameters of shield construction, coverage ratio of a tunnel formed by the shield construction and tunneling parameters of a shield machine; i represents the total number of shield segment rings, wherein I is a positive integer, and I is not less than 50;
training an LSTM neural network model:
Step 201, respectively preprocessing the final value of the earth surface deformation and the earth surface deformation influence factors of the construction area where the I shield segment rings are located to obtain L influence factor sets after preprocessing and a deformation set after preprocessing; wherein the first influence factor set after pretreatment is that The pretreatment data of the first influencing factor in the construction of the construction area where the ith shield segment ring is positioned are represented, I represents the sequence number of the shield segment ring, I is a positive integer, and I is more than or equal to 1 and less than or equal to I; l is a positive integer, and L is more than or equal to 1 and less than or equal to L, wherein L represents the total number of earth surface deformation influencing factors;
the deformation set after pretreatment is { Y } ′1 ,Y ′2 ,...,Y ′i ,...,Y ′I}; wherein ,Y′i The pretreated data of the final value of the earth surface deformation of the construction area where the ith shield segment ring is positioned are represented;
step 202, processing the L preprocessed influence factor sets and the preprocessed deformation set by adopting an MIC method to obtain the weight of the surface deformation influence factors;
step 203, constructing an LSTM neural network model;
step 204, multiplying the weight of the surface deformation influencing factors obtained in the step 202 and the surface deformation influencing factors at 10 adjacent measurement moments as an input layer, taking the surface deformation at 3 subsequent measurement moments as an output layer, and inputting the constructed LSTM neural network model for training to obtain a trained LSTM neural network model;
And thirdly, predicting the earth surface deformation of each shield segment in the construction process of the construction area where the shield segment is located after the I shield segments are predicted according to the trained LSTM neural network model.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: the geological parameters of the shield construction in the first step comprise cohesive force, internal friction angle, compression modulus, static side pressure coefficient and poisson ratio, and the tunneling parameters of the shield machine comprise tunneling speed of the shield machine, cutter head torque of the shield machine, soil bin pressure of the shield machine, total thrust of the shield machine, grouting pressure of the shield machine, grouting quantity of the shield machine and shield tail clearance.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: the method for obtaining the earth surface deformation influence factors of the construction area where the I shield segment rings are located in is the same, wherein the earth surface deformation influence factors of the construction area where the I shield segment rings are located are obtained, and the concrete process is as follows:
step 101, obtaining geological parameters of shield construction according to a geological profile and a geological survey report;
102, acquiring a coverage ratio of a tunnel formed by shield construction according to a shield construction design drawing;
And 103, acquiring tunneling parameters of the shield machine according to a PLC data acquisition module on the shield machine.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: the method for obtaining the earth surface deformation influence factors of the construction area where the I shield segment rings are located in is the same, wherein the earth surface deformation influence factors of the construction area where the I shield segment rings are located are obtained, and the concrete process is as follows:
step A01, the data processor is according to the formulaObtaining the cohesive force of stratum of the construction area where the ith shield segment ring is positioned +.>Inner friction angle of stratum of construction area where ith shield segment ring is positioned>Compression modulus of stratum of construction area where ith shield segment ring is located>Static lateral pressure coefficient of stratum of construction area where ith shield segment ring is positioned +.>Poisson ratio of earth surface of construction area where ith shield segment ring is positioned> wherein ,Hi Represents the thickness of stratum of the construction area where the ith shield segment ring is positioned, N i The stratum representing the construction area of the ith shield segment ring is composed of the total number of soil layers from top to bottom,/for the shield segment ring>N in stratum representing construction area of ith shield segment ring i Thickness of soil layer, n i and Ni Are all positive integers, and n is more than or equal to 1 i ≤N i ,/>N in stratum representing construction area of ith shield segment ring i Cohesive force of individual soil layers->N in stratum representing construction area of ith shield segment ring i Inner friction angle of each soil layer->N in stratum representing construction area of ith shield segment ring i Compression modulus of individual soil layers->N in stratum representing construction area of ith shield segment ring i Static side pressure coefficient of each soil layer, < > is->N in stratum representing construction area of ith shield segment ring i Poisson ratio of individual soil layers;
a02, the data processor acquires the earth surface deformation influence factors of the construction area of the ith shield segment ring, and records the earth surface deformation influence factors of the construction area of the ith shield segment ring as an ith shield construction influence factor data set wherein ,/>The tunneling speed of the shield machine when the ith shield segment ring is constructed in the construction area is represented by +.>The cutter torque of the shield machine when the construction area of the ith shield segment ring is constructed is represented, and the cutter torque is +.>Representing the soil bin pressure of the shield machine when the ith shield segment ring is in the construction area, and +.>Representing the total thrust of the shield machine when the ith shield segment ring is positioned in the construction area and the +. >The grouting pressure of the shield machine when the construction area of the ith shield segment ring is constructed is represented by +.>Indicating the position of the ith shield segment ringGrouting amount of shield machine during construction of construction area, < ->Represents the gap between the shield tail and the shield tail of the construction area where the ith shield segment ring is positioned, and the shield tail is filled with the shield>And (5) representing the coverage ratio of the tunnel formed by construction of the construction area where the ith shield segment ring is positioned.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: the method for obtaining the surface deformation of the construction area where the I shield segment ring is located in the first step is the same, wherein the method for obtaining the surface deformation of the construction area where the I shield segment ring is located in the second step comprises the following specific processes:
a1, arranging a plurality of deformation monitoring points on the ground surface of a construction area where an ith shield segment ring is positioned;
a2, according to preset measurement moments, obtaining deformation of each measurement moment of each deformation monitoring point on the earth surface of a construction area where the ith shield segment ring is located, and obtaining the maximum deformation of each measurement moment on the earth surface of the construction area where the ith shield segment ring is located; wherein the maximum deformation of the earth surface of the construction area where the ith shield segment ring is positioned at the d measurement moment is recorded as the earth surface deformation h of the construction area where the ith shield segment ring is positioned at the d measurement moment i (d);
A3, measuring the deformation monitoring point at the D measurement time after the construction of the ith shield segment ring is completed for 10 days, obtaining the maximum deformation of the ground surface of the construction area where the ith shield segment ring is located at the D measurement time, obtaining the ground surface deformation of the construction area where the ith shield segment ring is located at the D measurement time, and taking the ground surface deformation of the construction area where the ith shield segment ring is located at the D measurement time as the ground surface deformation final value Y of the construction area where the ith shield segment ring is located i The method comprises the steps of carrying out a first treatment on the surface of the Wherein D and D are positive integers, D is more than or equal to 1 and less than or equal to D, and D represents the total number of measurement.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: in step 201, the final value of the surface deformation of the construction area where the I shield segment rings are located and the surface deformation influencing factors are preprocessed respectively, and the specific process is as follows:
the data processor preprocesses the earth surface deformation influence factors of the I shield segment rings, and the specific process is as follows:
step 2011, the data processor records the cohesive force influence factor set in the I group shield construction influence factor data set as the 1 st influence factor set wherein ,/>Data representing the cohesive force of the 1 st influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
The inner friction angle influence factor set in the I group shield construction influence factor data set is recorded as the 2 nd influence factor set wherein ,/>Data representing the internal friction angle of the 2 nd influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the compression modulus influence factor set in the I group shield construction influence factor data set is recorded as the 3 rd influence factor set wherein ,/>Data representing compression modulus of the 3 rd influencing factor in construction of a construction area where the ith shield segment ring is positioned;
recording static side pressure coefficient influence factor set in I group shield construction influence factor data setMake the 4 th set of influencing factors wherein ,/>Data representing the static lateral pressure coefficient of the 4 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the Poisson ratio influence factor set in the I group shield construction influence factor data set is recorded as the 5 th influence factor set wherein ,/>Data representing the Poisson ratio of the 5 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
recording a tunneling speed influence factor set in the I-group shield construction influence factor data set as a 6 th influence factor set wherein ,/>Data representing tunneling speed of a 6 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
Recording a cutter disc torque influence factor set in the I-group shield construction influence factor data set as a 7 th influence factor set wherein ,/>Data representing cutter torque of a 7 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
centralizing I group shield construction influence factor dataThe bin pressure influence factor set is recorded as the 8 th influence factor set wherein ,/>Data representing the pressure of an 8 th influencing factor soil bin in the construction of a construction area where an i shield segment ring is positioned;
the total thrust influence factor set in the I group shield construction influence factor data set is recorded as the 9 th influence factor set wherein ,/>Data representing total thrust of a 9 th influence factor in construction of a construction area where an i shield segment ring is positioned;
recording the grouting pressure influence factor set in the I-group shield construction influence factor data set as the 10 th influence factor set wherein ,/>Data representing grouting pressure of a 10 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
the grouting quantity influence factor set in the I group shield construction influence factor data set is recorded as the 11 th influence factor set wherein ,/>Data representing the grouting amount of the 11 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
Influence of I group shield constructionThe shield tail clearance influence factor set in the factor data set is recorded as the 12 th influence factor set wherein ,/>Data representing a 12 th influence factor shield tail gap in the construction of a construction area where an i shield segment ring is positioned;
the tunnel span ratio influence factor set in the I group shield construction influence factor data set is recorded as the 13 th influence factor set wherein ,/>Data representing the 13 th influence factor tunnel span ratio when the ith shield segment ring is constructed in the construction area;
step 2012, the data processor marks the 1 st set of influencing factors to the 13 st set of influencing factors as the 1 st set of influencing factors wherein ,/>Data representing a first influencing factor when the ith shield segment ring is constructed in a construction area, wherein L is a positive integer, L is more than or equal to 1 and less than or equal to L, and L=13;
in step 2013, the method for preprocessing the 13 influence factor sets by the data processor is the same, wherein the preprocessing of the first influence factor set is as follows:
step 20131, when i > 1,when in use, then->Is an abnormal value, then ∈>Replaced by->And is also provided withOtherwise, go (L)>Replaced by->And-> wherein ,/>Data representing the first influencing factor of the construction area of the i-1 shield segment ring, and +. >Data representing the first influencing factor in construction of the construction area where the (i+1) th shield segment ring is located, mu l Mean value and sigma of the first influence factor set l Representing the standard deviation of the first set of influencing factors;
step 20132, until the first set of influencing factors is completedTo obtain the first influence factor set after pretreatment +.>
2014, deforming the earth surface of the construction area where the I shield segment rings are positioned by the data processorThe final value of the quantity is recorded as a deformation quantity set { Y ] 1 ,Y 2 ,...,Y i ,...,Y I}; wherein ,Yi The final value of the earth surface deformation of the construction area where the ith shield segment ring is positioned is represented as Y i-1 The final value of the surface deformation of the construction area where the (i+1) th shield segment ring is positioned is Y i+1 ;
Step 2015, when i > 1, |Y i -μ y |>3σ y When then Y i Is an abnormal value, Y i Replaced by Y' i And (2) andotherwise, Y i Replaced by Y' i And Y is i =Y′ i; wherein ,μy Mean value and sigma of deformation collection y Representing standard deviation of deformation amount set;
step 2016, until Y in the set of deformations is completed I To obtain a preprocessed deformation set { Y' 1 ,Y′ 2 ,...,Y′ i ,...,Y′ I }。
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: in step 202, the MIC method is adopted to process the preprocessed L influence factor sets and the preprocessed deformation set to obtain the weight of the surface deformation influence factor, and the specific process is as follows:
2021, taking the preprocessed first influence factor set as an X-axis coordinate, and taking the preprocessed deformation set as a Y-axis coordinate to obtain a scatter diagram of the first influence factor and the deformation under the I shield segment ring; wherein,is a data scatter point;
step 2022, the data processor performs grid division on the scatter diagram of the first influencing factor and the deformation to obtain a maximum normalized mutual information value, where the maximum normalized mutual information value is obtained by the first influencing factorCorresponding maximum information coefficient value MIC l ;
Step 2023, repeating step 2021 and step 2022 for multiple times to obtain maximum information coefficient values corresponding to 13 influencing factors;
step 2024, the data processor follows the formulaObtaining the weight corresponding to the first influence factor
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: the trained LSTM neural network model is obtained in step 204, and the specific acquisition process is as follows:
2041, dividing the maximum deformation of the earth surface of the construction area where the I shield segment rings are located at D measuring moments into a training set and a testing set in the shield construction process; the training set comprises a training set group number F, a testing set group number F', wherein the training set accounts for 80%, and the testing set accounts for 20%;
Step 2042, the data processor sets the training set and the testing set to comprise the products of the earth surface deformation of adjacent 10 measurement moments and the weights corresponding to 13 influence factors of 13 influence factor data respectively as input layers, and the earth surface deformation of the next adjacent 3 measurement moments as output layers;
2043, inputting the F group training set into the LSTM neural network model for training to obtain an initial LSTM neural network model; wherein the mean square error average value of the training set is less than 0.001;
step 2044, inputting the F' group test set into an initial LSTM neural network model for testing to obtain a trained LSTM neural network model; the mean square error average value of the test set is smaller than 0.07, and the correlation coefficient of the test set is larger than 90%.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: the trained LSTM neural network model predicts the earth surface deformation of the construction area where each shield segment is located after I shield segments, and the concrete process is as follows:
step 301, when the I+g shield segment ring is constructed, obtaining a product set X of the I+g shield construction influence factor data and the weight corresponding to 13 influence factors I+g And (2) andwherein g is a positive integer,
and g is 1, 2;
302, setting the d ' measurement time of the earth surface of the construction area where the I+g shield segment ring is located as the current time, and recording the maximum deformation of the d ' measurement time of the earth surface of the construction area where the I+g shield segment ring is located as the earth surface deformation h of the d ' measurement time of the construction area where the I+g shield segment ring is located I+g (d') using the maximum deformation and X of the current time and the previous 9 measurement times of the earth surface of the construction area where the (I+g) th shield segment ring is positioned I+g Input layer { H ] as LSTM neural network model I+g (d′),X I+g}; wherein ,HI+g (d′)=[h I+g (d′-9),h I+g (d′-8),h I+g (d′-7),h I+g (d′-6),h I+g (d′-5),h I+g (d′-4),h I+g (d′-3),h I+g (d′-2),h I+g (d′-1),h I+g (d′)]
The method comprises the steps of carrying out a first treatment on the surface of the Wherein d 'is a positive integer and d' is greater than 9;
step 303, combining { H } I+g (d′),X I+g Inputting the trained LSTM neural network model in the step 204, and predicting to obtain the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +1 measurement moment, the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +2 measurement moment and the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +3 measurement moment.
The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of: in step 204, a grid search method may be further used to obtain an optimal value of the number of neurons in the LSTM layer and the number of neurons in the Dense layer, and an optimal value of the learning rate lr and the training iteration number te.
Compared with the prior art, the invention has the following advantages:
1. the method has simple steps and reasonable design, and realizes the prediction of the deformation of the earth surface of the shield construction.
2. The method comprises the steps of firstly obtaining the earth surface deformation of a construction area where I shield segment rings are located and earth surface deformation influence factors, then training an LSTM neural network model, in the training process of the LSTM neural network model, firstly respectively preprocessing the earth surface deformation final value of the construction area where the I shield segment rings are located and the earth surface deformation influence factors, then adopting an MIC method to process the preprocessed L influence factor sets and the preprocessed deformation sets to obtain the weight of the earth surface deformation influence factors, finally multiplying the weight of the earth surface deformation influence factors and the earth surface deformation of the adjacent 10 measurement moments are used as input layers, the earth surface deformation of the next adjacent 3 measurement moments is used as output layers, inputting the LSTM neural network model for training, and finally predicting the earth surface deformation of the construction area where each shield segment is located after the I shield segments are located according to the trained LSTM neural network model.
3. According to the method, the data of each influence factor influencing the deformation of the earth surface of the shield construction and the final value of the deformation of the earth surface of the shield construction are preprocessed before the weights corresponding to the influence factors are obtained by adopting the maximum information coefficient MIC method, the 13 th influence factor set after preprocessing and the deformation set after preprocessing are obtained through judging, removing and replacing the influence factor set and the deformation set, and the accuracy of calculating the weights corresponding to the influence factors by adopting the maximum information coefficient method is improved.
4. According to the method, the maximum information coefficient MIC method is adopted to obtain the maximum normalized mutual information value corresponding to the first influence factor under different grid division in the weights corresponding to the influence factors, so that the weights corresponding to the first influence factor are obtained, the correlation degree of the linear and nonlinear influence factors and the deformation is effectively adapted, the calculation complexity is low, and the robustness is good.
5. According to the invention, a trained LSTM neural network model is adopted, the weight of the surface deformation influence factors and the surface deformation influence factors are effectively utilized to multiply and the surface deformation quantity at the adjacent 10 measurement moments are taken as an input layer, and the surface deformation quantity at the subsequent adjacent 3 measurement moments is taken as an output layer, so that the calculation efficiency and the calculation precision of the LSTM neural network model are improved, and the gradient attenuation and explosion problems of the LSTM neural network model are reduced.
6. The invention can distinguish the importance degree of each influence factor of the input layer of the LSTM neural network model, is beneficial to saving the calculation of the LSTM neural network model and improves the prediction precision of the model.
7. According to the invention, the data of the construction areas where the I shield segment rings are located are acquired and trained in the shield construction process, so that the earth surface deformation of the next 3 adjacent measurement moments after the current measurement can be predicted in advance in the construction process of the construction areas where the I shield segment rings are located, the earth surface deformation data can be acquired in advance by constructors conveniently, the constructors can take measures in advance or reduce the expansion of accidents conveniently, and the prediction precision and the operation efficiency of the earth surface deformation prediction of the shield construction are higher.
In conclusion, the method has simple steps and reasonable design, realizes the weight assignment of each influence factor of the deformation of the earth surface of the shield construction so as to distinguish the importance degree of each influence factor on the prediction model, adopts the LSTM neural network to carry out modeling dynamic prediction, and improves the prediction precision and efficiency of the deformation of the earth surface of the shield construction.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
FIG. 1 is a block flow diagram of the method of the present invention.
Detailed Description
The technical solutions of the present invention will be clearly and completely described below with reference to the drawings and the present embodiment of the present invention, and it should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present invention, and unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs.
As shown in FIG. 1, a dynamic prediction method for shield construction earth surface deformation based on MIC-LSTM is provided, which specifically comprises the following steps:
step one, in the shield construction process, obtaining the earth surface deformation of the construction area where the I shield segment rings are positioned and earth surface deformation influence factors; the earth surface deformation influencing factors comprise geological parameters of shield construction, coverage ratio of a tunnel formed by the shield construction and tunneling parameters of a shield machine; i represents the total number of shield segment rings, wherein I is a positive integer, and I is not less than 50;
training an LSTM neural network model:
step 201, respectively preprocessing the final value of the earth surface deformation and the earth surface deformation influence factors of the construction area where the I shield segment rings are located to obtain L influence factor sets after preprocessing and a deformation set after preprocessing; wherein the first influence factor set after pretreatment is that The pretreatment data of the first influencing factor in the construction of the construction area where the ith shield segment ring is positioned are represented, I represents the sequence number of the shield segment ring, I is a positive integer, and I is more than or equal to 1 and less than or equal to I; l is a positive integer, and L is more than or equal to 1 and less than or equal to L, wherein L represents the total number of earth surface deformation influencing factors;
the deformation amount set after pretreatment is { Y' 1 ,Y′ 2 ,...,Y′ i ,...,Y′ I -a }; wherein Y' i The pretreated data of the final value of the earth surface deformation of the construction area where the ith shield segment ring is positioned are represented;
step 202, processing the L preprocessed influence factor sets and the preprocessed deformation set by adopting an MIC method to obtain the weight of the surface deformation influence factors;
step 203, constructing an LSTM neural network model;
step 204, multiplying the weight of the surface deformation influencing factors obtained in the step 202 and the surface deformation influencing factors at 10 adjacent measurement moments as an input layer, taking the surface deformation at 3 subsequent measurement moments as an output layer, and inputting the constructed LSTM neural network model for training to obtain a trained LSTM neural network model;
and thirdly, predicting the earth surface deformation of each shield segment in the construction process of the construction area where the shield segment is located after the I shield segments are predicted according to the trained LSTM neural network model.
In this embodiment, geological parameters of the shield construction in the first step include cohesive force, internal friction angle, compression modulus, static side pressure coefficient and poisson ratio, and tunneling parameters of the shield machine include tunneling speed of the shield machine, cutter head torque of the shield machine, soil bin pressure of the shield machine, total thrust of the shield machine, grouting pressure of the shield machine, grouting amount of the shield machine and shield tail gap.
In this embodiment, the method for obtaining the earth surface deformation influence factor of the construction area where the I shield segment ring is located in the first step is the same, where the specific process of obtaining the earth surface deformation influence factor of the construction area where the I shield segment ring is located is as follows:
step 101, obtaining geological parameters of shield construction according to a geological profile and a geological survey report;
102, acquiring a coverage ratio of a tunnel formed by shield construction according to a shield construction design drawing;
and 103, acquiring tunneling parameters of the shield machine according to a PLC data acquisition module on the shield machine.
In this embodiment, the method for obtaining the earth surface deformation influence factor of the construction area where the I shield segment ring is located in the first step is the same, where the specific process of obtaining the earth surface deformation influence factor of the construction area where the I shield segment ring is located is as follows:
Step A01, the data processor is according to the formulaObtaining the cohesive force of stratum of the construction area where the ith shield segment ring is positioned +.>Inner friction angle of stratum of construction area where ith shield segment ring is positioned>Compression modulus of stratum of construction area where ith shield segment ring is located>Static lateral pressure coefficient of stratum of construction area where ith shield segment ring is positioned +.>Poisson ratio of earth surface of construction area where ith shield segment ring is positioned> wherein ,Hi Represents the thickness of stratum of the construction area where the ith shield segment ring is positioned, N i The stratum representing the construction area of the ith shield segment ring is composed of the total number of soil layers from top to bottom,/for the shield segment ring>N in stratum representing construction area of ith shield segment ring i Thickness of soil layer, n i and Ni Are all positive integers, and n is more than or equal to 1 i ≤N i ,/>N in stratum representing construction area of ith shield segment ring i Cohesive force of individual soil layers->N in stratum representing construction area of ith shield segment ring i Inner friction angle of each soil layer->N in stratum representing construction area of ith shield segment ring i Compression modulus of individual soil layers->N in stratum representing construction area of ith shield segment ring i Static side pressure coefficient of each soil layer, < > is- >N in stratum representing construction area of ith shield segment ring i Poisson ratio of individual soil layers;
a02, the data processor acquires the earth surface deformation influence factors of the construction area of the ith shield segment ring, and records the earth surface deformation influence factors of the construction area of the ith shield segment ring as an ith shield construction influence factor data set wherein ,/>The tunneling speed of the shield machine when the ith shield segment ring is constructed in the construction area is represented by +.>The cutter torque of the shield machine when the construction area of the ith shield segment ring is constructed is represented, and the cutter torque is +.>Representing the soil bin pressure of the shield machine when the ith shield segment ring is in the construction area, and +.>Represents the ith shieldTotal thrust of shield machine in construction area where the segment ring is located, +.>The grouting pressure of the shield machine when the construction area of the ith shield segment ring is constructed is represented by +.>The grouting quantity of the shield machine when the construction area of the ith shield segment ring is constructed is represented by +.>Represents the gap between the shield tail and the shield tail of the construction area where the ith shield segment ring is positioned, and the shield tail is filled with the shield>And (5) representing the coverage ratio of the tunnel formed by construction of the construction area where the ith shield segment ring is positioned.
In this embodiment, the method for obtaining the surface deformation of the construction area where the I shield segment ring is located in the first step is the same, where the method for obtaining the surface deformation of the construction area where the I shield segment ring is located includes the following specific processes:
A1, arranging a plurality of deformation monitoring points on the ground surface of a construction area where an ith shield segment ring is positioned;
a2, according to preset measurement moments, obtaining deformation of each measurement moment of each deformation monitoring point on the earth surface of a construction area where the ith shield segment ring is located, and obtaining the maximum deformation of each measurement moment on the earth surface of the construction area where the ith shield segment ring is located; wherein the maximum deformation of the earth surface of the construction area where the ith shield segment ring is positioned at the d measurement moment is recorded as the earth surface deformation h of the construction area where the ith shield segment ring is positioned at the d measurement moment i (d);
A3, measuring the D measuring moment of the deformation monitoring point after the construction of the ith shield segment ring is completed for 10 days, and obtaining the maximum deformation of the ground surface of the construction area where the ith shield segment ring is positioned at the D measuring moment to obtain the ith shieldConstructing the earth surface deformation of the construction area of the segment ring at the D measuring moment, and taking the earth surface deformation of the construction area of the shield segment ring at the D measuring moment as the earth surface deformation final value Y of the construction area of the shield segment ring at the i i The method comprises the steps of carrying out a first treatment on the surface of the Wherein D and D are positive integers, D is more than or equal to 1 and less than or equal to D, and D represents the total number of measurement.
In this embodiment, in step 201, the final value of the surface deformation and the surface deformation influencing factors of the construction area where the I shield segment rings are located are preprocessed respectively, and the specific process is as follows:
the data processor preprocesses the earth surface deformation influence factors of the I shield segment rings, and the specific process is as follows:
step 2011, the data processor records the cohesive force influence factor set in the I group shield construction influence factor data set as the 1 st influence factor set wherein ,/>Data representing the cohesive force of the 1 st influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the inner friction angle influence factor set in the I group shield construction influence factor data set is recorded as the 2 nd influence factor set wherein ,/>Data representing the internal friction angle of the 2 nd influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the compression modulus influence factor set in the I group shield construction influence factor data set is recorded as the 3 rd influence factor set wherein ,/>Data representing compression modulus of the 3 rd influencing factor in construction of a construction area where the ith shield segment ring is positioned;
recording a static side pressure coefficient influence factor set in the I group shield construction influence factor data set as a 4 th influence factor set wherein ,/>Data representing the static lateral pressure coefficient of the 4 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the Poisson ratio influence factor set in the I group shield construction influence factor data set is recorded as the 5 th influence factor set wherein ,/>Data representing the Poisson ratio of the 5 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
recording a tunneling speed influence factor set in the I-group shield construction influence factor data set as a 6 th influence factor set wherein ,/>Data representing tunneling speed of a 6 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
recording a cutter disc torque influence factor set in the I-group shield construction influence factor data set as a 7 th influence factor set wherein ,/>Data representing cutter torque of a 7 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
recording a soil bin pressure influence factor set in the I-group shield construction influence factor data set as an 8 th influence factor set wherein ,/>Data representing the pressure of an 8 th influencing factor soil bin in the construction of a construction area where an i shield segment ring is positioned;
the total thrust influence factor set in the I group shield construction influence factor data set is recorded as the 9 th influence factor set wherein ,/>Data representing total thrust of a 9 th influence factor in construction of a construction area where an i shield segment ring is positioned;
recording the grouting pressure influence factor set in the I-group shield construction influence factor data set as the 10 th influence factor set wherein ,/>Data representing grouting pressure of a 10 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
the grouting quantity influence factor set in the I group shield construction influence factor data set is recorded as the 11 th influence factor set wherein ,/>Data representing the grouting amount of the 11 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
recording a shield tail clearance influence factor set in the I-group shield construction influence factor data set as a 12 th influence factor set wherein ,/>Data representing a 12 th influence factor shield tail gap in the construction of a construction area where an i shield segment ring is positioned;
the tunnel span ratio influence factor set in the I group shield construction influence factor data set is recorded as the 13 th influence factor set wherein ,/>Data representing the 13 th influence factor tunnel span ratio when the ith shield segment ring is constructed in the construction area;
step 2012, the data processor marks the 1 st set of influencing factors to the 13 st set of influencing factors as the 1 st set of influencing factors wherein ,/>Data representing a first influencing factor when the ith shield segment ring is constructed in a construction area, wherein L is a positive integer, L is more than or equal to 1 and less than or equal to L, and L=13;
in step 2013, the method for preprocessing the 13 influence factor sets by the data processor is the same, wherein the preprocessing of the first influence factor set is as follows:
step 20131, when i > 1,when in use, then->Is an abnormal value, then ∈>Replaced by->And is also provided withOtherwise, go (L)>Replaced by->And-> wherein ,/>Data representing the first influencing factor of the construction area of the i-1 shield segment ring, and +.>Data representing the first influencing factor in construction of the construction area where the (i+1) th shield segment ring is located, mu l Mean value and sigma of the first influence factor set l Representing the standard deviation of the first set of influencing factors;
step 20132, until the first set of influencing factors is completedIs judged to obtain a prognosisTreated first influencing factor set +.>
2014, the data processor records the final value of the surface deformation of the construction area where the I shield segment rings are located as a deformation set { Y } 1 ,Y 2 ,...,Y i ,...,Y I}; wherein ,Yi The final value of the earth surface deformation of the construction area where the ith shield segment ring is positioned is represented as Y i-1 The final value of the surface deformation of the construction area where the (i+1) th shield segment ring is positioned is Y i+1 ;
Step 2015, when i > 1, |Y i -μ y |>3σ y When then Y i Is an abnormal value, Y i Replaced by Y' i And (2) andotherwise, Y i Replaced by Y' i And Y is i =Y′ i; wherein ,μy Mean value and sigma of deformation collection y Representing standard deviation of deformation amount set; />
Step 2016, until Y in the set of deformations is completed I To obtain a preprocessed deformation set { Y' 1 ,Y′ 2 ,...,Y′ i ,...,Y′ I }。
In this embodiment, in step 202, the MIC method is used to process the preprocessed L impact factor sets and the preprocessed deformation set to obtain the weight of the surface deformation impact factor, and the specific process is as follows:
2021, taking the preprocessed first influence factor set as an X-axis coordinate, and taking the preprocessed deformation set as a Y-axis coordinate to obtain a scatter diagram of the first influence factor and the deformation under the I shield segment ring; wherein,is a data scatter point;
step 2022, the data processor performs grid division on the scatter diagram of the first influencing factor and the deformation to obtain a maximum normalized mutual information value, where the maximum normalized mutual information value is the maximum information coefficient value MIC corresponding to the first influencing factor l ;
Step 2023, repeating step 2021 and step 2022 for multiple times to obtain maximum information coefficient values corresponding to 13 influencing factors;
step 2024, the data processor follows the formulaObtaining the weight corresponding to the first influence factor
In this embodiment, the trained LSTM neural network model is obtained in step 204, and the specific acquisition process is as follows:
2041, dividing the maximum deformation of the earth surface of the construction area where the I shield segment rings are located at D measuring moments into a training set and a testing set in the shield construction process; the training set comprises a training set group number F, a testing set group number F', wherein the training set accounts for 80%, and the testing set accounts for 20%;
step 2042, the data processor sets the training set and the testing set to comprise the products of the earth surface deformation of adjacent 10 measurement moments and the weights corresponding to 13 influence factors of 13 influence factor data respectively as input layers, and the earth surface deformation of the next adjacent 3 measurement moments as output layers;
2043, inputting the F group training set into the LSTM neural network model for training to obtain an initial LSTM neural network model; wherein the mean square error average value of the training set is less than 0.001;
step 2044, inputting the F' group test set into an initial LSTM neural network model for testing to obtain a trained LSTM neural network model; the mean square error average value of the test set is smaller than 0.07, and the correlation coefficient of the test set is larger than 90%.
In this embodiment, the surface deformation of the construction area where each shield segment is located after predicting the I shield segments by using the trained LSTM neural network model in the third step is specifically as follows:
step 301, when the I+g shield segment ring is constructed, obtaining a product set X of the I+g shield construction influence factor data and the weight corresponding to 13 influence factors I+g And (2) andwherein g is a positive integer and g is 1, 2;
302, setting the d ' measurement time of the earth surface of the construction area where the I+g shield segment ring is located as the current time, and recording the maximum deformation of the d ' measurement time of the earth surface of the construction area where the I+g shield segment ring is located as the earth surface deformation h of the d ' measurement time of the construction area where the I+g shield segment ring is located I+g (d') using the maximum deformation and X of the current time and the previous 9 measurement times of the earth surface of the construction area where the (I+g) th shield segment ring is positioned I+g Input layer { H ] as LSTM neural network model I+g (d′),X I+g}; wherein ,HI+g (d′)=[h I+g (d′-9),h I+g (d′-8),h I+g (d′-7),h I+g (d′-6),h I+g (d′-5),h I+g (d′-4),h I+g (d′-3),h I+g (d′-2),h I+g (d′-1),h I+g (d′)]
The method comprises the steps of carrying out a first treatment on the surface of the Wherein d 'is a positive integer and d' is greater than 9;
step 303, combining { H } I+g (d′),X I+g Inputting the trained LSTM neural network model in the step 204, and predicting to obtain the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +1 measurement moment, the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +2 measurement moment and the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +3 measurement moment.
In this embodiment, in step 204, a grid search method may be further used to obtain the optimal values of the number of neurons in the LSTM layer, the number of neurons in the Dense layer, the learning rate lr, and the training iteration number te.
In this embodiment, it should be noted that, the numerical range of the total number I of shield segment rings may be appropriately adjusted according to the need, so as to increase the accuracy of training of the LSTM neural network model.
In this embodiment, the area to be constructed is divided into a plurality of shield segment ring construction areas according to the tunneling direction of the shield construction from back to front.
In this embodiment, it should be noted that, in step a02, the deformation of the ith deformation monitoring point at the d measurement time on the earth surface of the construction area where the ith shield segment ring is located is obtained, and the deformation of the ith deformation monitoring point at the d measurement time is ordered from small to large, so as to obtain the maximum deformation of the ith deformation of the earth surface of the construction area where the ith shield segment ring is located.
In this example, the interval between two adjacent measurement times is 12 to 24 hours.
In this embodiment, it is further preferable that the interval between two adjacent measurement times is 12 hours.
In this embodiment, the thickness H of the stratum of the construction area where the ith shield segment ring is located i Refers to the bottom layer of the region where the outer diameter of the ith shield segment ring is positioned. Here, H i Is equal to the outer diameter of the shield segment ring.
In the embodiment, in the shield construction process, deformation data of each deformation monitoring point are obtained in real time according to preset measurement time.
In this embodiment, the coverage ratio refers to a ratio between a depth of a designed shield tunneling axis of a tunnel and a diameter of the tunnel.
In this embodiment, during actual construction, the distance between two adjacent shield segment rings is 1.2m.
In this embodiment, it should be noted that a plurality of deformation monitoring points are uniformly distributed on the earth surface of the construction area where each shield segment ring is located.
Further, the quantity of deformation monitoring points on the earth surface of the construction area where any shield segment ring is located is 11, wherein 11 deformation monitoring points are respectively 1 st deformation monitoring point, 2 nd deformation monitoring point, 3 rd deformation monitoring point, 4 th deformation monitoring point, 5 th deformation monitoring point, 6 th deformation monitoring point, 7 th deformation monitoring point, 8 th deformation monitoring point, 9 th deformation monitoring point, 10 th deformation monitoring point, 11 th deformation monitoring point.
In this embodiment, further, the connection lines of 11 deformation monitoring points on the earth surface of the construction area where any shield segment ring is located are vertically arranged with the shield tunneling direction, and the connection line projection of 11 deformation monitoring points coincides with the center of any shield segment ring in the width direction;
in this embodiment, further, 11 deformation monitoring points are located on the same section.
In this embodiment, further, the 1 st deformation monitoring point and the 11 th deformation monitoring point are symmetrically arranged, the 2 nd deformation monitoring point and the 10 th deformation monitoring point are symmetrically arranged, the 3 rd deformation monitoring point and the 9 th deformation monitoring point are symmetrically arranged, the 4 th deformation monitoring point and the 8 th deformation monitoring point are symmetrically arranged, the 5 th deformation monitoring point and the 7 th deformation monitoring point are symmetrically arranged, and the vertical projection of the 6 th deformation monitoring point is located on the shield tunneling axis;
in this embodiment, further, the horizontal distance between the 1 st deformation monitoring point and the 2 nd deformation monitoring point is 6m, the horizontal distance between the 2 nd deformation monitoring point and the 3 rd deformation monitoring point is 6m, the horizontal distance between the 3 rd deformation monitoring point and the 4 th deformation monitoring point is 3m, the horizontal distance between the 4 th deformation monitoring point and the 5 th deformation monitoring point is 3m, and the horizontal distance between the 5 th deformation monitoring point and the 6 th deformation monitoring point is 4.5m.
In this embodiment, in step a02, the deformation of each deformation monitoring point on the earth surface of the construction area where the ith shield segment ring is located is obtained according to the preset measurement time, and the specific process is as follows:
step A, adopting a total station or a precise level to perform shield segment ring on the ith shield segment ringMeasuring the initial elevation of each deformation monitoring point on the ground surface of the construction area, and acquiring the initial elevation of each deformation monitoring point on the ground surface of the construction area where the ith shield segment ring is positioned; wherein, the initial Gao Chengji of the c deformation monitoring point on the earth surface of the construction area where the i shield segment ring is positioned is taken as
B, measuring the elevation of each deformation monitoring point on the ground surface of the construction area where the ith shield segment ring is positioned by adopting a total station or a precise level gauge at the d measurement moment, and acquiring the elevation of each deformation monitoring point at the d measurement moment on the ground surface of the construction area where the ith shield segment ring is positioned; wherein d is a positive integer, the value of two adjacent measuring moments is 12-24 hours, and Gao Chengji of the (d) th measuring moment of the (c) th deformation monitoring point on the ground surface of the construction area where the (i) th shield segment ring is positioned is taken as
Step C, obtaining the deformation of each deformation monitoring point on the earth surface of the construction area of the ith shield segment ring according to the initial elevation of each deformation monitoring point on the earth surface of the construction area of the ith shield segment ring and the elevation of the d measurement moment of each deformation monitoring point on the earth surface of the construction area of the ith shield segment ring; wherein the deformation of the ith deformation monitoring point on the earth surface of the construction area where the ith shield segment ring is positioned at the d measuring moment is recorded asAnd is also provided withc is a positive integer, c is more than or equal to 1 and less than or equal to 11, and d is a positive integer;
step D, sorting the deformation of 11 deformation monitoring points at the D-th measuring moment on the earth surface of the construction area where the i shield segment ring is positioned from small to large to obtain the maximum deformation of the D-th measuring moment on the earth surface of the construction area where the i shield segment ring is positioned;
and E, in the shield construction process, according to the methods from the step B to the step D, obtaining the earth surface deformation of the construction area where the ith shield segment ring is positioned at the D measurement moment after the construction of the ith shield segment ring is completed for 10 days.
In this example, it is further explained that, whenWhen the value is larger than zero, the bulge deformation is indicated; when->When the pressure is less than zero, the sedimentation deformation is indicated; when- >Equal to zero, the undeformed amount is indicated.
In this embodiment, the total station may employ the TCR 1201 as the standard card total station.
In this embodiment, the precise level may be a Tianbao DINI03 precise level.
In this embodiment, it should be noted that, in actual use, according to the stratum to be tunneled by the shield, finite element software may be further adopted to simulate and build a finite element model of the shield tunnel, and simulate the shield construction, so as to obtain factors affecting the deformation of the earth surface of the shield construction, including geological parameters of the shield construction, coverage ratio of the tunnel formed by the shield construction, and tunnelling parameters of the shield machine.
In this embodiment, the deformation of the earth's surface during the shield construction is mainly caused by the stratum loss. Geological parameters of the shield penetrating through the stratum can reflect consolidation and sub-consolidation conditions of the soil body after disturbance in the shield construction process. Therefore, the geological parameter is one of important factors affecting the deformation of the earth's surface in the shield construction. As the earth surface deformation caused by the shield construction is gradually smaller along with the increase of the buried depth of the tunnel, the coverage area of the shield tunnel can exactly reflect the earth covering condition and the shield tunneling characteristic of the tunnel, and the influence of the earth surface deformation of the shield is larger. The stability of the tunnel face of the shield construction is controlled by the pressure of the soil bin of the shield machine, the total thrust and the cutter torque of the shield machine can influence the tunneling speed, the tunneling speed can influence the disturbance degree of surrounding soil, and therefore, the mutual coupling effect among tunneling parameters can influence the deformation of the shield earth surface. Synchronous grouting construction after the shield is assembled with the segment ring can well reduce stratum loss and control earth surface deformation. Therefore, grouting pressure and grouting amount in the shield synchronous grouting process can have a great influence on deformation of the shield construction ground surface. The shield tail gap refers to a gap between the outer edge of the segment and the inner wall of the shield tail. The shield tail gap generated in the shield construction process can cause the soil body to move into the gap so as to loosen and collapse the soil body, thereby causing the deformation of the earth surface, and the method is one of key factors influencing the deformation of the earth surface of the shield.
In this embodiment, the data processor in step 2022 performs grid division on the scatter plot of the first influencing factor and the deformation amount to obtain the maximum normalized mutual information value, which specifically includes the following steps:
step I, setting the number of grids divided in the X-axis direction as X when dividing the alpha-th grid of the data processor a Setting the grid number divided in the Y-axis direction as Y a; wherein ,xa ×y a <I 0.6 A is a positive integer;
step II, the data processor obtains x a and ya Maximum mutual information value under different grid division and is recorded as if max (x a ,y a );
Step III, the data processor is according to the formulaObtaining x a and ya Normalized mutual information value if 'under grid division' max (x a ,y a );
Step IV, when the data processor is divided into the (a+1) th grids, setting the number of grids divided in the X-axis direction as X a+1 Setting the grid number divided in the Y-axis direction as Y a+1; wherein ,xa+1 ×y a+1 <I 0.6 And x is a+1 =x a +1,y a+1 =y a +1, and x at grid division 1 1 =1,y 1 =1;
Step V, obtaining x according to the methods described in the step II and the step III a+1 and ya+1 Normalized mutual information value if 'under grid division' max (x a+1 ,y a+1 );
Step VI, repeating the step II and the step III for A times to obtain A normalized mutual information values, and sequencing the A normalized mutual information values from small to large to obtain the maximum normalized mutual information value; wherein A represents the total times of grid division, A is a positive integer, a is more than or equal to 1 and less than or equal to A, and x A ×y A <I 0.6 ,x A Represents the number of grids divided in the X-axis direction in the A-th grid division, y A The number of grids divided in the Y-axis direction at the time of the a-th grid division is represented.
In this embodiment, it should be noted that dropout=0.4 is set in the LSTM layer, so that over-fitting can be effectively avoided and model performance can be improved.
In this embodiment, by considering the weights corresponding to the 13 influencing factors and multiplying the 13 influencing factors with the weights corresponding to the 13 influencing factors respectively to be used as the input item of the LSTM neural network model, the parameters of the LSTM neural network model are accurate, and the performance of the LSTM neural network is improved.
In this embodiment, in step 203, an LSTM neural network model is constructed; the LSTM neural network model comprises 1 input layer, 3 LSTM layers, 1 Dense layer and 1 output layer; the LSTM neural network model is provided with an s igmoid activation function and a tanh activation function, the number of neurons of an input layer is 23, and the number of neurons of an output layer is 3;
in this embodiment, the training of the LSTM neural network model in step 204 further includes the following steps:
the data processor sets the learning rate lr to be 0.001-1, the training iteration number te is 100-250, and the weight parameter initial value of the forgetting gate, the bias parameter initial value of the forgetting gate, the weight parameter initial value of the input gate, the bias parameter initial value of the input gate, the weight parameter initial value of the output gate, the bias parameter initial value of the output gate, the weight parameter initial value of the unit state input at the current moment and the weight parameter initial value of the unit state input at the current moment are random numbers in standard normal distribution;
The data processor optimizes the weight parameters of the forgetting gate, the bias parameters of the forgetting gate, the weight parameters of the input gate, the bias parameters of the input gate, the weight parameters of the output gate, the bias parameters of the output gate, the weight parameters of the unit state input at the current moment and the weight parameters of the unit state input at the current moment in the LSTM neural network model by adopting a back propagation method.
In this embodiment, the mean square error average value of the training set in step 2043 is obtained as follows:
step 20431, the data processor sets the f training set to input the LSTM neural network model to obtain three predicted deformation amounts corresponding to the f training set and />And the measured earth surface deformation of the adjacent 3 measuring moments corresponding to the f training set are respectively +.> and />Wherein F is more than or equal to 1 and less than or equal to F, and F and F are positive integers;
step 20432, the data processor is based on the formulaObtaining the mean square error MSE of the f training set f ;
Step 20433, the data processor is based on the formulaThe mean square error mean MSE of the training set is obtained.
In this embodiment, the method for obtaining the mean square error average value of the test set is the same as the method for obtaining the mean square error average value of the training set.
In this embodiment, the mean square error average value of the test set in step 2044 is obtained as follows:
Step 20441, the data processor sets the f 'test set to input the LSTM neural network model to obtain three predicted deformation corresponding to the f' test set and />And the measured earth surface deformation amounts at the adjacent 3 measuring moments corresponding to the f' th test set are +.> and />Wherein, F 'is more than or equal to 1 and less than or equal to F', and F 'and F' are positive integers;
step 20442, the data processor is based on the formulaObtaining the mean square error MSE ' of the f ' test set ' f′ ;
Step 20443, the data processor is based on the formulaThe mean square error mean MSE' of the test set is obtained.
In this embodiment, the obtaining of the correlation coefficient of the test set in step 2044 specifically includes the following steps:
20444, the data processor composes 3F 'predicted deformation amounts from the three predicted deformation amounts corresponding to the F' group test valuesForming 3F 'measured earth surface deformation by three measured deformation corresponding to the F' group test values; wherein the f "th predicted deformation is denoted as Y f The measured surface deformation corresponding to the f' th predicted deformation is recorded asF ' is a positive integer, and F ' is more than or equal to 1 and less than or equal to 3F '; />
Step 20445, the data processor is according to the formulaObtaining the correlation coefficient R of the test set 2 ;
Step 20445, the data processor determines that the mean square error mean value MSE' of the test set is less than 0.07, and the correlation coefficient R of the test set 2 Whether or not > 0.9 holds true when MSE' < 0.07 and R 2 When more than 0.9 is established, a trained LSTM neural network model is obtained.
In this embodiment, it should be noted that,the 1 st influencing factor of the construction area where the I+g shield segment ring is positioned is expressed, and the I+g shield segment ring is->The 2 nd influencing factors of the construction area where the I+g shield segment ring is positioned are expressed,indicating the 3 rd influencing factor of the construction area where the I+g shield segment ring is located, < ->Indicating the 4 th influencing factor of the construction area where the I+g shield segment ring is positioned, < ->Representing the construction area of the I+g shield segment ring5 th influencing factor in domain construction, < ->The 6 th influencing factor of the construction area where the I+g shield segment ring is positioned is expressed, and the I+g shield segment ring is added>Represents the 7 th influencing factor of the construction area where the I+g shield segment ring is positioned, < ->Indicating the 8 th influencing factor of the construction area where the I+g shield segment ring is positioned, < ->Indicating the 9 th influencing factor of the construction area where the I+g shield segment ring is positioned, < ->The 10 th influencing factor of the construction area where the I+g shield segment ring is positioned is expressed, and the I+g shield segment ring is added>Indicating the 11 th influencing factor of the construction area where the I+g shield segment ring is positioned, The 12 th influencing factor of the construction area where the I+g shield segment ring is positioned is expressed, and the I+g shield segment ring is added>And (3) representing the coverage ratio of the tunnel formed in shield construction of the construction area where the I+g shield segment ring is positioned.
In this embodiment, it should be noted that,respectively represent the 1 st influencing factorsWeight-13 th influence factor.
In this embodiment, the data of this embodiment is derived from the left line construction data of the shield section of the Kunming subway five-line six-standard Yi-Kangfu-way station, and the computer configuration and software environment used for completing this embodiment are: intel (R) Core (TM) i5-7200U CPU,16.0GB memory. The system is Windows10 (64 bits), the program language version is Python3.7.8, the integrated development environment is spyder 4.1.4 version in an Anaconda package, and Tensorflow is used as a back end in a Keras library.
In this embodiment, the total number of training sets and test sets is 200, the number of groups of training sets is f=160, and the number of groups of test sets is F' =40.
In this embodiment, in step 202, the weights corresponding to the influencing factors are obtained by using a maximum information coefficient MIC method, as shown in table 1 below.
Table 1 weights corresponding to influencing factors
In this embodiment, the number of neurons of the first LSTM layer in the super parameters of the trained LSTM neural network model is Unit1, the number of neurons of the second LSTM layer is Unit2, the number of neurons of the third LSTM layer is Unit3, and the number of neurons of the response layer is Unit4, as shown in table 2.
Table 2 super parameter range and optimal value
In the embodiment, the training of the F group to obtain MSE=0.0008 < 0.001 shows that the LSTM neural network model is trained well to obtain an initial LSTM neural network model;
in this embodiment, F' groupTraining to obtain MSE' =0.064 < 0.07, R 2 =0.945 > 90%, the LSTM neural network model is well trained, and the corresponding model accuracy requirement can be met.
In this embodiment, the MIC-LSTM model of the earth surface deformation caused by the shield construction constructed by the present invention is compared with the predicted results of the LSTM neural network model and the BP neural network model, and the results are shown in table 3. As can be seen from the table, the prediction model provided by the invention has the advantages of minimum prediction error, highest fitting degree of the prediction result and the actual measurement result, and the validity of the constructed MIC-LSTM prediction model is verified, so that the prediction result of the ground surface deformation with higher precision can be provided, decision support can be provided for the prediction control of the ground surface deformation in the shield construction process, and the safety of the shield construction is guaranteed.
Table 3 model predictive outcome comparison
In conclusion, the method has simple steps and reasonable design, realizes the weight assignment of each influence factor of the deformation of the earth surface of the shield construction so as to distinguish the importance degree of each influence factor on the prediction model, adopts the LSTM neural network to carry out modeling dynamic prediction, and improves the prediction precision and efficiency of the deformation of the earth surface of the shield construction.
The foregoing description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and any simple modification, variation and equivalent structural changes made to the above embodiment according to the technical substance of the present invention still fall within the scope of the technical solution of the present invention.
Claims (9)
1. The dynamic prediction method for the deformation of the shield construction earth surface based on the MIC-LSTM is characterized by comprising the following steps of:
step one, in the shield construction process, obtaining the earth surface deformation of the construction area where the I shield segment rings are positioned and earth surface deformation influence factors; the earth surface deformation influencing factors comprise geological parameters of shield construction, coverage ratio of a tunnel formed by the shield construction and tunneling parameters of a shield machine; i represents the total number of shield segment rings, wherein I is a positive integer, and I is not less than 50;
training an LSTM neural network model:
step 201, respectively preprocessing the final value of the earth surface deformation and the earth surface deformation influence factors of the construction area where the I shield segment rings are located to obtain L influence factor sets after preprocessing and a deformation set after preprocessing; wherein the first influence factor set after pretreatment is that The pretreatment data of the first influencing factor in the construction of the construction area where the ith shield segment ring is positioned are represented, I represents the sequence number of the shield segment ring, I is a positive integer, and I is more than or equal to 1 and less than or equal to I; l is a positive integer, and L is more than or equal to 1 and less than or equal to L, wherein L represents the total number of earth surface deformation influencing factors;
the deformation amount set after pretreatment is { Y' 1 ,Y′ 2 ,...,Y′ i ,...,Y′ I -a }; wherein Y' i The pretreated data of the final value of the earth surface deformation of the construction area where the ith shield segment ring is positioned are represented;
step 202, processing the L preprocessed influence factor sets and the preprocessed deformation set by adopting an MIC method to obtain the weight of the surface deformation influence factors;
step 203, constructing an LSTM neural network model;
step 204, multiplying the weight of the surface deformation influencing factors obtained in the step 202 and the surface deformation influencing factors at 10 adjacent measurement moments as an input layer, taking the surface deformation at 3 subsequent measurement moments as an output layer, and inputting the constructed LSTM neural network model for training to obtain a trained LSTM neural network model;
predicting the earth surface deformation of each shield segment in the construction process of the construction area where the shield segment is located after I shield segments according to the trained LSTM neural network model;
In step 202, the MIC method is adopted to process the preprocessed L influence factor sets and the preprocessed deformation set to obtain the weight of the surface deformation influence factor, and the specific process is as follows:
2021, taking the preprocessed first influence factor set as an X-axis coordinate, and taking the preprocessed deformation set as a Y-axis coordinate to obtain a scatter diagram of the first influence factor and the deformation under the I shield segment ring; wherein,is a data scatter point;
step 2022, the data processor performs grid division on the scatter diagram of the first influencing factor and the deformation to obtain a maximum normalized mutual information value, where the maximum normalized mutual information value is the maximum information coefficient value MIC corresponding to the first influencing factor l ;
Step 2023, repeating step 2021 and step 2022 for multiple times to obtain maximum information coefficient values corresponding to 13 influencing factors;
2. The MIC-LSTM-based dynamic prediction method for the deformation of the shield construction earth surface is characterized by comprising the following steps of: the geological parameters of the shield construction in the first step comprise cohesive force, internal friction angle, compression modulus, static side pressure coefficient and poisson ratio, and the tunneling parameters of the shield machine comprise tunneling speed of the shield machine, cutter head torque of the shield machine, soil bin pressure of the shield machine, total thrust of the shield machine, grouting pressure of the shield machine, grouting quantity of the shield machine and shield tail clearance.
3. The MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation according to claim 1 or 2, wherein the method comprises the following steps: the method for obtaining the earth surface deformation influence factors of the construction area where the I shield segment rings are located in is the same, wherein the earth surface deformation influence factors of the construction area where the I shield segment rings are located are obtained, and the concrete process is as follows:
step 101, obtaining geological parameters of shield construction according to a geological profile and a geological survey report;
102, acquiring a coverage ratio of a tunnel formed by shield construction according to a shield construction design drawing;
and 103, acquiring tunneling parameters of the shield machine according to a PLC data acquisition module on the shield machine.
4. The MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation according to claim 1 or 2, wherein the method comprises the following steps: the method for obtaining the earth surface deformation influence factors of the construction area where the I shield segment rings are located in is the same, wherein the earth surface deformation influence factors of the construction area where the I shield segment rings are located are obtained, and the concrete process is as follows:
step A01, the data processor is according to the formulaObtaining the cohesive force of stratum of the construction area where the ith shield segment ring is positioned +. >Inner friction angle of stratum of construction area where ith shield segment ring is positioned>Compression modulus of stratum of construction area where ith shield segment ring is located>Static lateral pressure coefficient of stratum of construction area where ith shield segment ring is positioned +.>Poisson ratio of earth surface of construction area where ith shield segment ring is positioned> wherein ,Hi Represents the thickness of stratum of the construction area where the ith shield segment ring is positioned, N i The stratum representing the construction area of the ith shield segment ring is composed of the total number of soil layers from top to bottom,/for the shield segment ring>N in stratum representing construction area of ith shield segment ring i Thickness of soil layer, n i and Ni Are all positive integers, and n is more than or equal to 1 i ≤N i ,/>N in stratum representing construction area of ith shield segment ring i Cohesive force of individual soil layers->N in stratum representing construction area of ith shield segment ring i Inner friction angle of each soil layer->N in stratum representing construction area of ith shield segment ring i Compression modulus of individual soil layers->N in stratum representing construction area of ith shield segment ring i Static side pressure coefficient of each soil layer, < > is->N in stratum representing construction area of ith shield segment ring i Poisson ratio of individual soil layers;
a02, the data processor acquires the earth surface deformation influence factors of the construction area of the ith shield segment ring, and records the earth surface deformation influence factors of the construction area of the ith shield segment ring as an ith shield construction influence factor data set wherein ,/>The tunneling speed of the shield machine when the ith shield segment ring is constructed in the construction area is represented by +.>The cutter torque of the shield machine when the construction area of the ith shield segment ring is constructed is represented, and the cutter torque is +.>Representing the soil bin pressure of the shield machine when the ith shield segment ring is in the construction area, and +.>Representing the total thrust of the shield machine when the ith shield segment ring is positioned in the construction area and the +.>The grouting pressure of the shield machine when the construction area of the ith shield segment ring is constructed is represented by +.>The grouting quantity of the shield machine when the construction area of the ith shield segment ring is constructed is represented by +.>Represents the gap between the shield tail and the shield tail of the construction area where the ith shield segment ring is positioned, and the shield tail is filled with the shield>And (5) representing the coverage ratio of the tunnel formed by construction of the construction area where the ith shield segment ring is positioned.
5. The MIC-LSTM-based dynamic prediction method for the deformation of the shield construction earth surface is characterized by comprising the following steps of: the method for obtaining the surface deformation of the construction area where the I shield segment ring is located in the first step is the same, wherein the method for obtaining the surface deformation of the construction area where the I shield segment ring is located in the second step comprises the following specific processes:
a1, arranging a plurality of deformation monitoring points on the ground surface of a construction area where an ith shield segment ring is positioned;
A2, according to preset measurement moments, obtaining deformation of each measurement moment of each deformation monitoring point on the earth surface of a construction area where the ith shield segment ring is located, and obtaining the maximum deformation of each measurement moment on the earth surface of the construction area where the ith shield segment ring is located; wherein the maximum deformation of the earth surface of the construction area where the ith shield segment ring is positioned at the d measurement moment is recorded as the earth surface deformation h of the construction area where the ith shield segment ring is positioned at the d measurement moment i (d);
A3, measuring the deformation monitoring point at the D measurement time after the construction of the ith shield segment ring is completed for 10 days, obtaining the maximum deformation of the ground surface of the construction area where the ith shield segment ring is located at the D measurement time, obtaining the ground surface deformation of the construction area where the ith shield segment ring is located at the D measurement time, and taking the ground surface deformation of the construction area where the ith shield segment ring is located at the D measurement time as the ground surface deformation final value Y of the construction area where the ith shield segment ring is located i The method comprises the steps of carrying out a first treatment on the surface of the Wherein D and D are positive integers, D is more than or equal to 1 and less than or equal to D, and D represents the total number of measurement.
6. The MIC-LSTM-based dynamic prediction method for the deformation of the shield construction earth surface is characterized by comprising the following steps of: in step 201, the final value of the surface deformation of the construction area where the I shield segment rings are located and the surface deformation influencing factors are preprocessed respectively, and the specific process is as follows:
The data processor preprocesses the earth surface deformation influence factors of the I shield segment rings, and the specific process is as follows:
step 2011, the data processor records the cohesive force influence factor set in the I group shield construction influence factor data set as the 1 st influence factor set wherein ,/>Data representing the cohesive force of the 1 st influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the inner friction angle influence factor set in the I group shield construction influence factor data set is recorded as the 2 nd influence factor set wherein ,/>Data representing the internal friction angle of the 2 nd influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the compression modulus influence factor set in the I group shield construction influence factor data set is recorded as the 3 rd influence factor set wherein ,/>Data representing compression modulus of the 3 rd influencing factor in construction of a construction area where the ith shield segment ring is positioned;
recording a static side pressure coefficient influence factor set in the I group shield construction influence factor data set as a 4 th influence factor set wherein ,/>Data representing the static lateral pressure coefficient of the 4 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
the Poisson ratio influence factor set in the I group shield construction influence factor data set is recorded as the 5 th influence factor set wherein ,/>Data representing the Poisson ratio of the 5 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
recording a tunneling speed influence factor set in the I-group shield construction influence factor data set as a 6 th influence factor set wherein ,/>Data representing tunneling speed of a 6 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
recording a cutter disc torque influence factor set in the I-group shield construction influence factor data set as a 7 th influence factor set wherein ,/>Data representing cutter torque of a 7 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
recording a soil bin pressure influence factor set in the I-group shield construction influence factor data set as an 8 th influence factor set wherein ,/>Data representing the pressure of an 8 th influencing factor soil bin in the construction of a construction area where an i shield segment ring is positioned;
the total thrust influence factor set in the I group shield construction influence factor data set is recorded as the 9 th influence factor set wherein ,/>Data representing total thrust of a 9 th influence factor in construction of a construction area where an i shield segment ring is positioned;
recording the grouting pressure influence factor set in the I-group shield construction influence factor data set as the 10 th influence factor set wherein ,/>Data representing grouting pressure of a 10 th influencing factor in construction of a construction area where an i shield segment ring is positioned;
the grouting quantity influence factor set in the I group shield construction influence factor data set is recorded as the 11 th influence factor set wherein ,/>data representing the grouting amount of the 11 th influencing factor in the construction of the construction area where the ith shield segment ring is positioned;
recording a shield tail clearance influence factor set in the I-group shield construction influence factor data set as a 12 th influence factor set wherein ,/>Data representing a 12 th influence factor shield tail gap in the construction of a construction area where an i shield segment ring is positioned;
the tunnel span ratio influence factor set in the I group shield construction influence factor data set is recorded as the 13 th influence factor set wherein ,/>Data representing the 13 th influence factor tunnel span ratio when the ith shield segment ring is constructed in the construction area;
step 2012, the data processor marks the 1 st set of influencing factors to the 13 st set of influencing factors as the 1 st set of influencing factors wherein ,/>Data representing a first influencing factor when the ith shield segment ring is constructed in a construction area, wherein L is a positive integer, L is more than or equal to 1 and less than or equal to L, and L=13;
In step 2013, the method for preprocessing the 13 influence factor sets by the data processor is the same, wherein the preprocessing of the first influence factor set is as follows:
step 20131, when i > 1,when in use, then->Is an abnormal value, then ∈>Replaced by->And is also provided withOtherwise, go (L)>Replaced by->And-> wherein ,/>Data representing the first influencing factor of the construction area of the i-1 shield segment ring, and +.>Data representing the first influencing factor in construction of the construction area where the (i+1) th shield segment ring is located, mu l Mean value and sigma of the first influence factor set l Representing the standard deviation of the first set of influencing factors;
step 20132, until the first set of influencing factors is completedTo obtain the first influence factor set after pretreatment +.>
2014, the data processor records the final value of the surface deformation of the construction area where the I shield segment rings are located as a deformation set { Y ] 1 ,Y 2 ,...,Y i ,...,Y I}; wherein ,Yi The final value of the earth surface deformation of the construction area where the ith shield segment ring is positioned is represented as Y i-1 The final value of the surface deformation of the construction area where the (i+1) th shield segment ring is positioned is Y i+1 ;
Step 2015, when i > 1, |Y i -μ y |>3σ y When then Y i Is an abnormal value, Y i Replaced by Y' i And (2) andotherwise, Y i Replaced by Y' i And Y is i =Y′ i; wherein ,μy Mean value and sigma of deformation collection y Representing standard deviation of deformation amount set;
step 2016, until Y in the set of deformations is completed I To obtain a preprocessed deformation set { Y' 1 ,Y′ 2 ,...,Y′ i ,...,Y′ I }。
7. The MIC-LSTM-based dynamic prediction method for the deformation of the shield construction earth surface is characterized by comprising the following steps of: the trained LSTM neural network model is obtained in step 204, and the specific acquisition process is as follows:
2041, dividing the maximum deformation of the earth surface of the construction area where the I shield segment rings are located at D measuring moments into a training set and a testing set in the shield construction process; the training set comprises a training set group number F, a testing set group number F', wherein the training set accounts for 80%, and the testing set accounts for 20%;
step 2042, the data processor sets the training set and the testing set to comprise the products of the earth surface deformation of adjacent 10 measurement moments and the weights corresponding to 13 influence factors of 13 influence factor data respectively as input layers, and the earth surface deformation of the next adjacent 3 measurement moments as output layers;
2043, inputting the F group training set into the LSTM neural network model for training to obtain an initial LSTM neural network model; wherein the mean square error average value of the training set is less than 0.001;
step 2044, inputting the F' group test set into an initial LSTM neural network model for testing to obtain a trained LSTM neural network model; the mean square error average value of the test set is smaller than 0.07, and the correlation coefficient of the test set is larger than 90%.
8. The MIC-LSTM-based dynamic prediction method for the deformation of the shield construction earth surface is characterized by comprising the following steps of: the trained LSTM neural network model predicts the earth surface deformation of the construction area where each shield segment is located after I shield segments, and the concrete process is as follows:
step 301, when the I+g shield segment ring is constructed, obtaining a product set X of the I+g shield construction influence factor data and the weight corresponding to 13 influence factors I+g And (2) andwherein g is a positive integer and g is 1, 2;
302, setting the d ' measurement time of the earth surface of the construction area where the I+g shield segment ring is located as the current time, and recording the maximum deformation of the d ' measurement time of the earth surface of the construction area where the I+g shield segment ring is located as the earth surface deformation h of the d ' measurement time of the construction area where the I+g shield segment ring is located I+g (d') then at the I+g thMaximum deformation and X of current earth surface moment and previous 9 measurement moments of construction area where shield segment ring is located I+g Input layer { H ] as LSTM neural network model I+g (d′),X I+g}; wherein ,HI+g (d′)=[h I+g (d′-9),h I+g (d′-8),h I+g (d′-7),h I+g (d′-6),h I+g (d′-5),h I+g (d′-4),h I+g (d′-3),h I+g (d′-2),h I+g (d′-1),h I+g (d′)]
The method comprises the steps of carrying out a first treatment on the surface of the Wherein d 'is a positive integer and d' is greater than 9;
step 303, combining { H } I+g (d′),X I+g Inputting the trained LSTM neural network model in the step 204, and predicting to obtain the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +1 measurement moment, the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +2 measurement moment and the earth surface deformation of the construction area where the I+g shield segment ring is located at the d ' +3 measurement moment.
9. The MIC-LSTM-based dynamic prediction method for the deformation of the shield construction earth surface is characterized by comprising the following steps of: in step 204, a grid search method may be further used to obtain an optimal value of the number of neurons in the LSTM layer and the number of neurons in the Dense layer, and an optimal value of the learning rate lr and the training iteration number te.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011268712.3A CN112364422B (en) | 2020-11-13 | 2020-11-13 | MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011268712.3A CN112364422B (en) | 2020-11-13 | 2020-11-13 | MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112364422A CN112364422A (en) | 2021-02-12 |
CN112364422B true CN112364422B (en) | 2023-06-13 |
Family
ID=74515541
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011268712.3A Active CN112364422B (en) | 2020-11-13 | 2020-11-13 | MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112364422B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113128106B (en) * | 2021-04-06 | 2023-03-21 | 汕头大学 | Method for determining surface subsidence caused by shield construction of karst stratum |
CN113251914A (en) * | 2021-04-06 | 2021-08-13 | 兰州交通大学 | Surface deformation prediction method combining InSAR technology and long-term memory neural network |
CN117235874B (en) * | 2023-11-14 | 2024-02-20 | 中电建铁路建设投资集团有限公司 | Track deformation prediction method and system based on shield tunneling |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109978226A (en) * | 2019-01-24 | 2019-07-05 | 同济大学 | Shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network |
WO2019140794A1 (en) * | 2018-01-16 | 2019-07-25 | 天津大学 | Dimensional analysis-based method for calculating total thrust of shield tunneling |
CN110147875A (en) * | 2019-05-07 | 2019-08-20 | 西安交通大学 | A kind of shield machine auxiliary cruise method based on LSTM neural network |
-
2020
- 2020-11-13 CN CN202011268712.3A patent/CN112364422B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019140794A1 (en) * | 2018-01-16 | 2019-07-25 | 天津大学 | Dimensional analysis-based method for calculating total thrust of shield tunneling |
CN109978226A (en) * | 2019-01-24 | 2019-07-05 | 同济大学 | Shield construction ground settlement prediction method based on Recognition with Recurrent Neural Network |
CN110147875A (en) * | 2019-05-07 | 2019-08-20 | 西安交通大学 | A kind of shield machine auxiliary cruise method based on LSTM neural network |
Non-Patent Citations (1)
Title |
---|
地铁隧道盾构施工地面沉降风险评价与沉降量预警研究;朱倩倩;《中国优秀硕士学位论文全文数据库》;第A011-40页 * |
Also Published As
Publication number | Publication date |
---|---|
CN112364422A (en) | 2021-02-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112364422B (en) | MIC-LSTM-based dynamic prediction method for shield construction earth surface deformation | |
CN104989456B (en) | A kind of Large Span Underground engineering excavation surrounding rock stability monitoring and pre-alarming method | |
CN103866769B (en) | Based on 3D geological model and real-time monitoring dam grouting engineering analysis control method | |
AU2019312330A1 (en) | Method for predicting mine strata pressure behavior data of stoping tunnel | |
CN110298107B (en) | Working face impact risk evaluation method based on incremental stacking | |
CN115828374A (en) | Method and system for predicting non-uniform deformation of composite stratum tunnel in construction period | |
CN103205972A (en) | Method for analyzing relationship between deformation of foundation pit and ground subsidence outside of foundation pit | |
CN107368463B (en) | Roadway nonlinear deformation prediction method based on fiber bragg grating sensor network data | |
CN112989262A (en) | Method for predicting maximum horizontal displacement of adjacent tunnel by considering excavation process of foundation pit | |
CN111119902B (en) | Tunnel dynamic construction method based on BP neural network | |
CN115423167A (en) | Subway deep foundation pit construction safety early warning and decision-making assisting method and system | |
CN112380757B (en) | Analysis method for offset distance of surface subsider peak point during construction of curved tunnel | |
Dong et al. | Inclination prediction of a giant open caisson during the sinking process using various machine learning algorithms | |
CN116776553A (en) | Method and device for controlling deformation of shield construction earth surface based on digital twin | |
Wang et al. | Multiparameter inversion early warning system of tunnel stress-seepage coupling based on IA-BP algorithm | |
CN106777556A (en) | A kind of spacial analytical method for assessing slope excavating phase stable state | |
CN115640996A (en) | Evaluation method for water inrush disaster of tunnel in water-rich complex stratum | |
CN115455791A (en) | Method for improving landslide displacement prediction accuracy rate based on numerical simulation technology | |
CN112861208B (en) | Analysis method for converting row pile supporting system into underground continuous wall supporting structure | |
CN112379082B (en) | MIC-based method for determining influence factors of shield construction surface deformation | |
CN111767323A (en) | Method for predicting length of stratum fracture in real time based on kernel SMO algorithm | |
CN111413732B (en) | Method for predicting collapse instability of high side wall of stratified rock mass underground engineering in excavation process | |
CN116878577B (en) | Method and system for monitoring tunnel drilling and blasting in-situ reconstruction and expansion engineering | |
CN113569327B (en) | Method and system for optimizing posture of shield tunneling machine and method and system for training model | |
US20230146355A1 (en) | Constructing derivative profiles from discrete pressure data of transient well tests |
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 |