CN107145665A - A kind of roadway surrounding rock answers force modeling and Forecasting Methodology - Google Patents
A kind of roadway surrounding rock answers force modeling and Forecasting Methodology Download PDFInfo
- Publication number
- CN107145665A CN107145665A CN201710308802.2A CN201710308802A CN107145665A CN 107145665 A CN107145665 A CN 107145665A CN 201710308802 A CN201710308802 A CN 201710308802A CN 107145665 A CN107145665 A CN 107145665A
- Authority
- CN
- China
- Prior art keywords
- stress
- elm
- strain
- matrix
- surrounding rock
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000011435 rock Substances 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 238000004445 quantitative analysis Methods 0.000 claims abstract description 13
- 238000010606 normalization Methods 0.000 claims abstract description 8
- 239000011159 matrix material Substances 0.000 claims description 36
- 238000012360 testing method Methods 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 16
- 238000005259 measurement Methods 0.000 claims description 12
- 230000006872 improvement Effects 0.000 claims description 8
- 210000002569 neuron Anatomy 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 230000003203 everyday effect Effects 0.000 claims description 2
- 238000006467 substitution reaction Methods 0.000 claims description 2
- 238000005457 optimization Methods 0.000 abstract description 13
- 238000004458 analytical method Methods 0.000 abstract description 5
- 238000001514 detection method Methods 0.000 abstract description 3
- 238000011017 operating method Methods 0.000 abstract description 3
- 230000000172 allergic effect Effects 0.000 abstract 1
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 241001269238 Data Species 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 229910052742 iron Inorganic materials 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000011439 discrete element method Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000002035 prolonged effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Architecture (AREA)
- Mathematical Analysis (AREA)
- Structural Engineering (AREA)
- Pure & Applied Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Civil Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Investigating Strength Of Materials By Application Of Mechanical Stress (AREA)
Abstract
The present invention relates to ground detection technique field, specifically disclose a kind of roadway surrounding rock and answer force modeling and Forecasting Methodology, this method is that the analysis mathematical modeling of roadway surrounding rock strain stress is set up using the ELM algorithms after optimization, utilizes set up mathematical modeling to obtain stress in the case where just knowing that strain.First, the actually detected value of several pieces roadway surrounding rock stress and strain is chosen, is normalized;Then, tradition ELM models are set up to a part of data in the strain stress data after normalization, using greedy algorithm, the quantitative analysis mathematical modeling after optimization ELM input layer weights foundation optimization;Then, stress is obtained to the prediction of known tunnel allergic effect based on the mathematical modeling.ELM models are applied to the modeling of roadway surrounding rock ess-strain by the present invention first, so as to realize the prediction in the case where just knowing that country rock strain to stress, using microcomputer modelling and calculate, its analytical cycle is short, cost is low, operating procedure is simple, improve operating efficiency, while reducing human error.
Description
Technical field
The present invention relates to ground detection technique field, and in particular to a kind of roadway surrounding rock answers force modeling and Forecasting Methodology.
Background technology
Roadway excavation, energy release of surrounding rock, while producing deviatoric stress in country rock.Surrouding rock stress is the stress of primary rock with partially should
The superposition of power, deviatoric stress can control rock mass damage.For workmen, roadway surrounding rock safety classification is important before constructing
Guilding principle.Correct classification can reduce construction cost, improve efficiency of construction, it is to avoid potential danger.At present, with quantitative target
Based on roadway surrounding rock safety status classification turn into developing direction.The strain of roadway surrounding rock and stress are two and important quantified
In index, Practical Project, the measurement of roadway surrounding rock strain is relatively easy to, but the measurement of the stress of roadway surrounding rock is but relatively more tired
Difficulty, and it is with high costs.
Often volume is larger for the existing sensor that can effectively measure stress, operation inconvenience, and price is high, it is difficult to real
Now extensive prolonged measurement, measurement mainly uses artificial continued observations and collection, and monitoring result is delayed.Existing numerical analysis
Modeling method, using numerical method, such as finite difference calculus, FInite Element, boundary element method, distinct element method etc., such as famous is imitative
True software for calculation FLAC3D is based on finite difference calculus establishment, according to the modulus of elasticity of ground, modulus of shearing, density, tension
A series of static parameters such as intensity, can simulate the stress and strain of roadway surrounding rock, acquisition for ideally data, but
In practical operation, the parameter that these methods needs are measured is more, and it is difficult to accurate stress is with answering in prediction acquisition reality
Variate.
At present, not yet there is the method that laneway stress can be accurately predicted in the case where just knowing that country rock strain, also do not have
The method that tunnel stress value and trend trend in tested part can be predicted.
The content of the invention
(1) technical problem to be solved
In order to solve the above mentioned problem of prior art, the present invention provides a kind of roadway surrounding rock and answers force modeling and Forecasting Methodology,
This method is to set up roadway surrounding rock strain and the quantitative analysis mathematical modulo of stress first with the ELM algorithms for improving input layer weights
Type, utilizes set up mathematical modeling quantitative forecast to go out the change of stress value.This method analytical cycle is short, easy to operate, operable
Property it is strong and to improve precision, accuracy by computer high.
(2) technical scheme
In order to achieve the above object, the main technical schemes that the present invention is used include:
A kind of roadway surrounding rock answers force modeling and Forecasting Methodology, and this method comprises the following steps:
S1, measurement obtains the stress value and strain value of the N number of point of roadway surrounding rock, record roadway surrounding rock of every day in M days
The strain value and the data of stress value each put;Wherein, the tunnel is horseshoe-shaped;
The data that the step one is obtained are normalized, the strain value and stress value are processed as into -1 by S2
Numerical value between to 1;
S3, the numerical value after being normalized using the S2 sets up the ELM algorithm quantitative analysis mathematical modelings by improvement, is based on
The mathematical modeling is tested the strain data for needing to predict, obtains the stress value being predicted in part and trend trend.
Method as described above, it is preferable that the S2 comprises the following steps:
S101, N number of point of roadway surrounding rock is made M × 2N's in strain Value Data vertically and horizontally respectively
Matrix A;And the matrix B of M × 1 is made in the stress Value Data of one of point;
S201, the matrix A that is obtained in the step S101, B are normalized;Respectively the matrix A,
Take that the data of maximum are designated as Cmax and minimum data are designated as Cmin in B, and both are made the difference obtain Cr, CtMeet formula one;
S202, with the numerical value in the matrix A, B replace C values respectively, the substitution formula one obtain respectively matrix A t,
Bt。
Method as described above, it is preferable that the S3 comprises the following steps:
S301, by the data of a portion after the normalization it is training sample, another part is used as test sample;
The matrix of the training sample is substituted into tradition ELM stress models to be modeled, the test sample is brought into and built
The stress model that mould is finished, obtains the test stress of prediction output, is compared with the data of actual measurement, obtains the prediction
The variance E of the test stress of output and the stress measured in practice;
S302, utilize the variance E, set up improvement ELM algorithm quantitative analysis mathematical modelings, according to country rock strain predict
Draw the stress of country rock.
Method as described above, it is preferable that described traditional ELM stress models are modeled, and are comprised the following steps:
Activation primitive g (x) takes sigmoid in S30101, ELM model;
S30102, the stochastic inputs layer matrix w for producing l × n:
Wherein, the l represents that hidden layer has l neuron, and the n, which is represented, has input layer to have n group input variables;Order output
Layer weights and hidden layer amount of bias are:
Wherein described m is m group output variables;
Input layer X is the training sample in the matrix A t, and output layer Y is the training sample in the matrix B t:
N, which is expressed as input layer X, in input layer X n groups vector, and each vector has Q element;In output layer Y, m is expressed as
Output layer Y has m groups vector, and each vector has p element,
Exporting T is:
Hidden layer exports H:
H=g (wX+b) formula seven
Wherein, g is sigmoid functions, and w is input layer weights, and X is input layer, and b is hidden layer amount of bias, according to described
And (7) (6), you can:
β=H+T
Wherein, the β is output layer weights;The H+Represent hidden layer output matrix H Moore-Penrose broad sense
It is inverse.
Method as described above, it is preferable that the ELM algorithm quantitative analysis mathematical modelings of the foundation improvement, specific steps
Including:
S30201, step-size in search t is defined first, take an input layer weight wijValue;
S30202, make wij'=wij+ t, updates output layer weights β ' according to traditional ELM methods, recalculates again
Variance E' is obtained, if the E'<E, retains new weight wij' and β ', and the E=E', continue executing with wij'=wij+t;
If S30203, the E'>E, selects another weights, performs S30201;If S30204, execution first time wij'
=wijDuring+t, the E'>E, then reversely search wij'=wij-t;If the E'<E, continues wij'=wij- t, until E'>E, choosing
Another weights is selected, S30201 is performed.
The method of the present invention is set up roadway surrounding rock stress and strain using ELM innovatory algorithm (modification of input layer weights) and determined
Amount analysis mathematical modeling, utilizes the quantitative value that stress is obtained in the case where just knowing that strain of set up mathematical modeling.
(3) beneficial effect
The beneficial effects of the invention are as follows:The modeling of roadway surrounding rock stress of the present invention and Forecasting Methodology can quick and precisely bases
Tunnel strain prediction laneway stress.Modeling of the present invention and Analysis of Prediction cycle are short, operating procedure is simple, cost
It is low, using microcomputer modelling and calculate, improve measuring accuracy, improve operating efficiency.In addition, the use of this method is reduced
The input and the input of a large amount of manpowers of instrument, working strength are small, the input cost of production have been saved, while reducing artificial mistake
Difference, and the degree of accuracy of prediction acquisition result is much higher than the method for being far longer than prior art.
Brief description of the drawings
Fig. 1 is roadway surrounding rock of the present invention strain and the modeling method flow chart of stress;
Fig. 2 is the flow chart of S2 specific steps described in Fig. 1;
The flow chart that Fig. 3 sets up for traditional ELM stress models;
Fig. 4 optimizes the flow chart of ELM input layer weights for the present invention using greedy algorithm;
Fig. 5 is the present invention is with ELM after optimization and is not optimised the comparison figure that ELM model trainings collection exports simulation result;
Fig. 6 is the position view of the tunnel section test point of the embodiment of the present invention 2;
Fig. 7 is the embodiment of the present invention 2 and conventional method test result figure.
Embodiment
In order to preferably explain the present invention, in order to understand, below in conjunction with the accompanying drawings, by embodiment, to this hair
It is bright to be described in detail.
Embodiment 1
The present invention establishes a kind of roadway surrounding rock and answers force modeling and Forecasting Methodology, and the modeling method is changed using inputting weights
ELM algorithms after entering set up roadway surrounding rock strain and the quantitative analysis mathematical modeling of stress, utilize set up mathematical modeling to exist
The stress of roadway surrounding rock is predicted in the case of the country rock strain just known that.
The modelling application of roadway surrounding rock stress and strain described by the embodiment of the present invention is in prediction roadway surrounding rock stress
Detailed process is as follows:
Preparation:Roadway surrounding rock internal strain and the stress value sample of the iron ore of mountain at the moment of Anshan iron and steel plant are gathered, record tunnel is enclosed
The value of the petrosa interface stress and strain of some days.
Specifically include following steps:As shown in figure 1,
S1, using convergence instrument, 3 D measuring instrument measures 5 points of roadway surrounding rock in strain value vertically and horizontally, prison
Survey 106 days, obtain 106 × 10 groups of sample datas;The stress value of a point is measured, 106 groups of sample datas are obtained.
S2, sample data is normalized, flow chart as shown in Figure 2, its implementation process is as follows:
Step S101, data normalization, incite somebody to action the data of resulting 106 × 10 groups of strain values, and 106 × 10 matrix A is made;
106 groups of stress value is obtained, 106 × 1 matrix B is made,
Step S201, by data normalization:
Obtained matrix A, matrix B are substituted into following equation one respectively to be normalized, wherein, respectively in the square
The data that maximum is obtained in battle array A, B are designated as Cmax and the data of minimum are designated as Cmin, and both are made the difference obtain Cr;CtMeet public
Formula one:
Step S202, by CtReplace original matrix A or B normalized after matrix A t, matrix B t.
The present embodiment is by normalization, by the less roadway surrounding rock strain value of actual value, and the larger tunnel of actual value
Surrouding rock stress value is compressed between -1 to 1, is so easy to set up the accurate model of description.
S3, normalized with S2 after numerical value set up by improvement ELM algorithm quantitative analysis mathematical modelings, based on the mathematics
Model is tested the strain data for needing to predict, obtains the stress value being predicted in part and trend trend;Use
Data after being normalized in S2, set up traditional ELM algorithm Quantitative Analysis Models, and optimize whole model using greedy algorithm,
The strain data for needing to predict is tested based on the model after the optimization, the stress value being predicted in part is obtained and becomes
Gesture is moved towards.
Specific steps:
S301, it regard random 90 groups of data of acquisition as training sample A90、B90, the data after normalization are to use At90、
Bt90Mark;Remaining 16 groups of data are obtained as test sample A16、B16, the data after normalization are to use At16、Bt16Mark,
It is stand-by;By the matrix training sample At90、Bt90Substitute into tradition ELM stress models to be modeled, the test sample is brought into
The model finished is modeled, the test stress of prediction output is obtained, is compared with the data of actual measurement, the prediction is obtained defeated
The variance of the test stress gone out and the stress measured in practice, is designated as E;
S302, the ELM algorithm quantitative analysis mathematical modelings improved using the variance E, foundation, are realized by Matlab and existed
In the case of knowing country rock strain, prediction draws the stress of country rock.
Greedy algorithm is mainly used to improve ELM models in the present invention, the input layer weights of traditional ELM networks are random productions
Raw, once after generation, their value is constant until training terminates, therefore, need to only determine hidden layer neuron activation primitive,
The number of hidden layer neuron number and network hidden layer, you can calculate output layer weights β.Because input layer weights are random,
Cause modeling it is unstable, variance is very big sometimes.In order to improve modeling accuracy, weighed using the random input layer of greedy algorithm amendment
Value so that whole system is advanced up towards the side of variance reduction.
Comprise the following steps that:
Activation primitive G (x) takes sigmoid functions in step S30101, ELM model, and node in hidden layer takes 20.
Step S30102, utilize the training sample after being normalized i.e. 90 group tunnel strain with stress data set up tradition
ELM models, are comprised the following steps that:
Produce l × n stochastic inputs layer matrix w:
Wherein, l is made to represent that hidden layer has l neuron, n, which is represented, has input layer there are n group input variables.
Output layer weights and the hidden layer amount of bias is made to be:
Wherein m is to refer to m group output variables, in the present embodiment m=n.
Input layer X is the training sample in matrix A t, and output layer Y is the training sample in matrix B t:
N, which is expressed as input layer X, in input layer X n groups vector, and each vector has Q in Q element, the present embodiment to be 10.It is defeated
Go out in layer Y, m, which is expressed as output layer Y, m groups vector, each vector has p in p element, the present embodiment to be 1;
Exporting T is:
T can regard as exporting Y;
Hidden layer exports H:
H=g (wX+b) formula seven
Wherein, g is sigmoid functions, and w is input layer weights, and x input layers, b is hidden layer amount of bias.According to formula six
With formula seven, you can:
H β=T formula eight
H exports for hidden layer in formula eight, and β is output layer weights.The output T of training sample is, it is known that finally calculate defeated
Go out weight computing output weights β:β=H+T, H here+Hidden layer output matrix H Moore-Penrose generalized inverses are represented, this
The traditional ELM stress models of one, sample, which are set up, to be finished.
Step S30103, rear 16 groups of data are brought into and set up the model that finishes, you can draw prediction output.
Step S30104, will draw prediction output with sensor gather data be compared, ELM calculate stress export
The variance of stress with measuring in practice is E variances, and its specific flow chart is as shown in Figure 3.
Traditional ELM modeling method is taught to formula 8 in formula 2.
Next, using greedy algorithm, further improving ELM and being modeled:Before modeling, first according to above-mentioned tradition
ELM is modeled, and input layer weights are wij, output layer weights are β, and error E, flow chart as shown in Figure 4, tool are obtained after calculating
Body step is:
Step S30201, defines step-size in search t first, takes an input layer weight wijValue.
Step S30202, makes wij'=wij+ t, wherein, i, j are exactly the i-th row in matrix, the number of jth row.According to traditional ELM
Method updates output layer weights β ' again, recalculates variance E', if E'<E, retains new weight wij' and β ', and E=
E', continues executing with wij'=wij+t。
Step S30203, if E'>E.If updated weights in S302 before, then it is assumed that find local minimum points, jump out choosing
Next weights are selected, are at this moment jumped out, another weights is selected, S30201 is performed;If the E' that S302 is calculated for the first time is greater than E,
Perform next step S30204.
Step S30204, if performing first time wij'=wijDuring+t, E'>E, then reversely search wij'=wij- t, if E'<
E, continues wij'=wij- t, until E'>E, selects another weights, that is, performs S30201.
Step S30205, the strain matrix for predicting needs are brought into ELM models after above-mentioned optimization, are realized by Matlab
Emulate and calculate the stress of country rock.
The stress and strain of roadway surrounding rock to be checked is modeled using the ELM after optimization, the modeling method is using greed calculation
Input layer weights in method, optimization ELM, so as to set up roadway surrounding rock strain and the quantitative analysis mathematical modeling of stress, using being built
The stress of country rock can be predicted out in the case where just knowing that country rock strain for vertical mathematical modeling.
Rear 16 groups of test datas are substituted into the emulation with Matlab implementation models, and image is made according to its degree of accuracy, are obtained
ELM is compared figure with tradition ELM to the prediction effect of roadway surrounding rock stress after to improvement, as shown in Figure 5.Wherein * desired values are profit
The actual value measured with sensor, zero is the predicted value of ELM after optimization, and is traditional ELM predicted value.Can be very bright from Fig. 5
Aobvious to find out, the ELM that the ELM ratios after optimization are not optimised is more nearly actual value, and the former degree of accuracy is far longer than the latter.
Embodiment 2
The method that embodiment 1 is set up is applied to the prediction that a tunnel dig through roadway surrounding rock stress, chooses different
In 4 sections in tunnel location, each section selection test point as shown in Figure 6, Fig. 6, summit has two points to constitute (ID
6061st, ID21461), take one, left side point ID21461 to be carried out as summit, obtain ID18941, ID20341, ID21461,
The strain both vertically as well as horizontally of this 5 points of ID4941, ID3541 is used as input.Summit (ID21461) vertical direction stress
It is used as input.Then it is brought into the ELM after traditional ELM and optimization to be trained, obtains model.Finally choose other sections 3
It is individual, the data of strain and stress are obtained using same method, as test sample, bring into the method that embodiment 1 is set up
Row test.Obtain result figure as shown in Figure 7.Wherein desired stress curve is that specific practical application sensor detects what is obtained
Actual value, the stress curve of traditional ELM predictions is the stress value that tradition ELM models are obtained, the stress curve that ELM is predicted after optimization
To predict the stress value obtained using the method for embodiment 1.
As a result the modeling and Forecasting Methodology for illustrating the roadway surrounding rock stress that the present invention is set up obtain predicted stresses value and reality
Difference is minimum between the stress value of measurement, predicted stresses value can be considered as into actual value and used.
Embodiment 3
The method that the present invention is set up and comparative analysis of the art methods to same roadway surrounding rock stress test, are adopted respectively
With method the following,
Control group 1:Surrouding rock stress is measured using 3 D measuring instrument on the market;
Control group 2:Utilization FLAC3D numerical methods of the prior art;
Experimental group 1:The traditional ELM modeling methods being not optimised;
Modeling method described in embodiment 1 is applied to prediction roadway surrounding rock stress.
The above method is measured obtain the difference size of the error after predicted stresses are compared with actual measurement, take length,
Consuming expense the data obtained see the table below 1.
The testing result of table 1.
Experimental subjects | Error difference | Measurement is time-consuming | Consuming expense |
Control group 1 | -- | >1 day | 3000000 yuan |
Control group 2 | It is ideally small | >1 hour | 10 yuan |
Experimental group 1 | Greatly | 2 seconds | 4000 yuan |
Embodiment 1 | It is small | 40 minutes | 4000 yuan |
As can be seen from Table 1:Surrouding rock stress is measured using 3 D measuring instrument on the market, can be considered for acquisition is true
Value, but required time is longer, and sensor price is up to millions of RMB.In addition, some chemical experimental instruments and manpower
Cost input is also essential.It is with low cost using existing numerical method analysis, but can only roadway surrounding rock preferable
Under state, numerical method could obtain preferable effect, and practicality is not enough.Using the traditional ELM modeling methods being not optimised, it is necessary to
The strain of roadway surrounding rock is obtained with strain transducer, its sensor total price is about 4000 yuan or so, and cost is not relatively high, tradition
ELM modeling method speed, but error is larger.Comparatively, herein described optimization ELM modeling methods, it is also desirable to use
The strain input cost that strain transducer obtains roadway surrounding rock is also 4000 yuan or so, testing result not high with respect to human cost
Error is small, predicts the outcome higher close to price, the result of the higher strain gauge detection of precision, considerable benefit.
Therefore, modeling method analytical cycle of the present invention is short, operating procedure is simple, using microcomputer modelling and counts
Calculate, improve measuring accuracy, improve operating efficiency.In addition, the use of this method reduces instrument input and a large amount of manpowers
Input, working strength is small, the input cost of production has been saved, while reducing human error.
It the above is only the preferred embodiment of the present invention, it is noted that above-mentioned preferred embodiment is not construed as pair
The limitation of the present invention, protection scope of the present invention should be defined by claim limited range.For the art
For those of ordinary skill, without departing from the spirit and scope of the present invention, some improvements and modifications can also be made, these change
Enter and retouch and also should be regarded as protection scope of the present invention.
Claims (5)
1. a kind of roadway surrounding rock answers force modeling and Forecasting Methodology, it is characterised in that it comprises the following steps:
S1, measurement obtains the stress value and strain value of the N number of point of roadway surrounding rock, and the roadway surrounding rock of every day is each in recording M days
The strain value of point and the data of stress value;Wherein, the tunnel is horseshoe-shaped;
The S1 data obtained are normalized, the strain value and stress value are processed as between -1 to 1 by S2
Numerical value;
S3, the numerical value after being normalized using the S2 sets up the ELM algorithm quantitative analysis mathematical modelings by improvement, based on the number
Learn model to test the strain data for needing to predict, obtain the stress value being predicted in part and trend trend.
2. the method as described in claim 1, it is characterised in that:The S2 comprises the following steps:
S101, the matrix that N number of point of roadway surrounding rock is made to M × 2N in strain Value Data vertically and horizontally respectively
A;And the matrix B of M × 1 is made in the stress Value Data of one of point;
S201, the matrix A that is obtained in the step S101, B are normalized;Respectively in the matrix A, B
Take that the data of maximum are designated as Cmax and minimum data are designated as Cmin, and both are made the difference obtain Cr, CtMeet formula one;
S202, with the numerical value in the matrix A, B C values are replaced respectively, the substitution formula one obtains matrix A t, Bt respectively.
3. the method as described in claim 1, it is characterised in that:The step S3 comprises the following steps:
S301, by the data of a portion after the normalization it is training sample, another part is used as test sample;
The matrix of the training sample is substituted into tradition ELM stress models to be modeled, the test sample is brought into and modeled
Complete stress model, obtains the test stress of prediction output, is compared with the data of actual measurement, obtains the prediction output
Test stress and the stress measured in practice variance E;
S302, the ELM algorithm quantitative analysis mathematical modelings improved using the variance E, foundation, are strained prediction according to country rock and drawn
The stress of country rock.
4. method as claimed in claim 3, it is characterised in that:Described traditional ELM stress models are modeled, including as follows
Step:
Activation primitive g (x) takes sigmoid in S30101, ELM model;
S30102, the stochastic inputs layer matrix w for producing l × n:
Wherein, the l represents that hidden layer has l neuron, and the n, which is represented, has input layer to have n group input variables;Output layer is made to weigh
Value β and hidden layer amount of bias b are:
Wherein, the m is m group output variables;
Input layer X is the training sample in the matrix A t, and output layer Y is the training sample in the matrix B t:
N, which is expressed as input layer X, in input layer X n groups vector, and each vector has Q element;In output layer Y, m is expressed as output
Layer Y has m groups vector, and each vector has p element,
Exporting T is:
Hidden layer exports H:
H=g (wX+b) formula seven
Wherein, g is sigmoid functions, and w is input layer weights, and X is input layer, and b is hidden layer amount of bias, according to the formula
Six and formula seven, you can:
β=H+T
Wherein, the β is output layer weights;The H+Represent hidden layer output matrix H Moore-Penrose generalized inverses.
5. method as claimed in claim 3, it is characterised in that:The ELM algorithm quantitative analysis mathematical modelings for setting up improvement,
Specific steps include:
S30201, step-size in search t is defined first, take an input layer weight wijValue;
S30202, make wij'=wij+ t, updates output layer weights β ' according to traditional ELM methods, recalculates acquisition side again
Poor E', if the E'<E, retains new weight wij' and β ', and the E=E', continue executing with wij'=wij+t;
If S30203, the E'>E, selects another weights, performs S30201;
If S30204, execution first time wij'=wijDuring+t, the E'>E, then reversely search wij'=wij-t;If the E'<
E, continues wij'=wij- t, until E'>E, selects another weights, performs S30201.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710308802.2A CN107145665B (en) | 2017-05-04 | 2017-05-04 | Roadway surrounding rock stress modeling and prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710308802.2A CN107145665B (en) | 2017-05-04 | 2017-05-04 | Roadway surrounding rock stress modeling and prediction method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107145665A true CN107145665A (en) | 2017-09-08 |
CN107145665B CN107145665B (en) | 2020-04-14 |
Family
ID=59774209
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710308802.2A Expired - Fee Related CN107145665B (en) | 2017-05-04 | 2017-05-04 | Roadway surrounding rock stress modeling and prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107145665B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112784331A (en) * | 2020-09-25 | 2021-05-11 | 汕头大学 | Soil stress-strain relation determination method based on improved LSTM deep learning method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104182622A (en) * | 2014-08-12 | 2014-12-03 | 大连海事大学 | Feedback analytical method and feedback analytical device during tunnel construction and based on extreme learning machine |
CN105260575A (en) * | 2015-11-17 | 2016-01-20 | 中国矿业大学 | Roadway surrounding rock deformation predicting method based on neural network |
-
2017
- 2017-05-04 CN CN201710308802.2A patent/CN107145665B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104182622A (en) * | 2014-08-12 | 2014-12-03 | 大连海事大学 | Feedback analytical method and feedback analytical device during tunnel construction and based on extreme learning machine |
CN105260575A (en) * | 2015-11-17 | 2016-01-20 | 中国矿业大学 | Roadway surrounding rock deformation predicting method based on neural network |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112784331A (en) * | 2020-09-25 | 2021-05-11 | 汕头大学 | Soil stress-strain relation determination method based on improved LSTM deep learning method |
Also Published As
Publication number | Publication date |
---|---|
CN107145665B (en) | 2020-04-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yu et al. | An intelligent displacement back-analysis method for earth-rockfill dams | |
CN107563067A (en) | Analysis of structural reliability method based on Adaptive proxy model | |
He et al. | Using genetic algorithms and finite element methods to detect shaft crack for rotor-bearing system | |
CN105259331B (en) | A kind of jointed rock mass uniaxial strengeth Forecasting Methodology | |
CN111209620B (en) | Method for predicting residual bearing capacity and crack propagation path of crack-containing structure | |
CN114117840B (en) | Structural performance prediction method based on simulation and test data hybrid drive | |
CN105260786A (en) | Comprehensive optimization method of simulation credibility evaluation model of electric propulsion system | |
CN102620939B (en) | Engine torque predicting method and engine torque predicting device | |
CN107977758A (en) | A kind of rainfall intensity forecasting procedure and relevant apparatus | |
CN104539601B (en) | Dynamic network attack process analysis method for reliability and system | |
CN114861542A (en) | Method, device and equipment for evaluating loss of direct current transmission project and storage medium | |
CN106568647A (en) | Nerve network-based concrete strength predication method | |
Chen et al. | Genetic programming for predicting aseismic abilities of school buildings | |
CN115270239A (en) | Bridge reliability prediction method based on dynamic characteristics and intelligent algorithm response surface method | |
YiFei et al. | Metamodel-assisted hybrid optimization strategy for model updating using vibration response data | |
CN107145665A (en) | A kind of roadway surrounding rock answers force modeling and Forecasting Methodology | |
CN116189794A (en) | Rammed earth water salt content measurement method | |
CN115577856A (en) | Method and system for predicting construction cost and controlling balance of power transformation project | |
CN116108745A (en) | Multi-parameter calibration method for water environment model, terminal equipment and storage medium | |
CN115310348A (en) | Stacking-based grouting amount integrated agent prediction model and prediction method | |
CN115048857A (en) | CNN-based transient stability limit transmission power determination method | |
CN112685900B (en) | Power load simulation method for representing impact load power characteristics | |
Lee et al. | Quick prediction of tunnel displacements using Artificial Neural Network and field measurement results | |
CN106934729A (en) | Building Testing and appraisal method and device | |
CN106709167A (en) | Similarity theory-based assembling validity assessment method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB03 | Change of inventor or designer information | ||
CB03 | Change of inventor or designer information |
Inventor after: Xiao Dong Inventor after: Zhang Qi Inventor after: Liu Xiaobo Inventor after: Mao Yachun Inventor before: Xiao Dong Inventor before: Li Beijing Inventor before: Liu Xiaobo Inventor before: Mao Yachun |
|
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200414 |