CN105353427A - Tunnel surrounding rock dynamic grading method and device thereof - Google Patents
Tunnel surrounding rock dynamic grading method and device thereof Download PDFInfo
- Publication number
- CN105353427A CN105353427A CN201510925540.5A CN201510925540A CN105353427A CN 105353427 A CN105353427 A CN 105353427A CN 201510925540 A CN201510925540 A CN 201510925540A CN 105353427 A CN105353427 A CN 105353427A
- Authority
- CN
- China
- Prior art keywords
- tunnel
- communication module
- learning machine
- individuality
- extreme learning
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V11/00—Prospecting or detecting by methods combining techniques covered by two or more of main groups G01V1/00 - G01V9/00
Abstract
The invention discloses a tunnel surrounding rock dynamic grading method and a device thereof. The device comprises a surrounding rock geological parameter acquisition device which is arranged at a tunnel face and used to obtain a surrounding rock geological parameter, a first wireless communication module connected with the surrounding rock geological parameter acquisition device, a transmitting box arranged outside a tunnel portal, and a relay station arranged between the surrounding rock geological parameter acquisition device and the transmitting box; an installation position of the transmitting box is within a coverage scope of a mobile phone signal, and the transmitting box comprises a second wireless communication module and a GPRS communication module connected with the second wireless communication module; the relay station is connected with the first wireless communication module and the second wireless communication module; and a server of the GPRS communication module is connected through a GPRS network. Dynamic grading of tunnel surrounding rocks is completed, and the surrounding rock grading is relative rapid and reliable.
Description
Technical field
The present invention relates to a kind of tunnel surrounding dynamic classification method and device thereof.
Background technology
Fender graded refers to and unlimited rock mass sequence is divided into limited the classification with different degree of stability according to the index such as rock mass completeness and rock strength, incorporating into by some country rocks that stability is similar is a class, whole country rocks is divided into some classes, and then the degree of stability again according to each class surrounding rock on the basis of rocvk classification provides best construction method and support structure design, fender graded is the foundation selecting construction method.In recent years, multiple computational intelligence method is applied in fender graded by domestic and international tunnel related researcher, as BP neural network, support vector machine (SVM) etc., achieve good effect, but BP neural network and support vector machine method are in application process, there is parameter to determine the shortcomings such as difficulty, iterative computation speed is slow can not meet the needs of Rapid Construction of Tunnels.And extreme learning machine (extremelearningmachine, ELM) only need to arrange network hidden layer node number before training, the input layer weights and the hidden layer side-play amount that adjust network is not needed in algorithm implementation, and unique optimum solution can be produced, there is Selecting parameter easy, fast and the advantage that Generalization Capability is good of pace of learning, the mode that in prior art, limits of application learning machine carries out classification of tunnel surroun ding rock cannot realize the remote transmission of advance geologic information, long-range country rock dynamic classification can not be completed, the network of country rock information is not more had to share and terminal inquiry function.
Summary of the invention
The present invention is directed to the proposition of above problem, and develop a kind of tunnel surrounding dynamic classification method and device thereof.
Technological means of the present invention is as follows:
A kind of tunnel surrounding dynamic classification method, comprises the steps:
Step 1: lay wall rock geology parameter acquisition devices in tunnel tunnel face place; Described wall rock geology parameter acquisition devices is connected with the first wireless communication module;
Step 2: lay firing box in tunnel portal and ensure that the riding position of described firing box is within the coverage of mobile phone signal outward; The GPRS communication module that described firing box comprises the second wireless communication module and is connected with described second wireless communication module; Described GPRS communication module is connected with server via GPRS network;
Step 3: lay relay station between described wall rock geology parameter acquisition devices and described firing box; Described relay station connects described first wireless communication module and described second wireless communication module;
Step 4: wall rock geology parameter acquisition devices obtains the wall rock geology parameter at tunnel tunnel face place, and this wall rock geology parameter transfers to described server through the first wireless communication module, relay station and firing box;
Step 5: described server using described wall rock geology parameter as input, fender graded result is exported by fender graded evolution extreme learning machine model, and corresponding for described fender graded result tunnel face surface information is stored in database, be distributed on network simultaneously;
Further, described wall rock geology parameter at least comprises: Rock Mass Integrality, groundwater parameter, rock mass discontinuity, the displacement of country rock Vault settlement and country rock convergence displacement; Described wall rock geology parameter acquisition devices comprises:
For obtaining geology detecting instrument and the ground penetrating radar of Rock Mass Integrality and rock mass discontinuity;
For obtaining the pore pressure gauge of the groundwater parameter of country rock;
For obtaining the displacement meter of country rock Vault settlement displacement;
And for obtaining the convergence gauge of country rock convergence displacement;
Further, described fender graded evolution extreme learning machine model draws as follows:
Steps A: the sample set building fender graded;
Step B: described sample set is divided into training sample set and test sample book collection two parts;
Step C: using the input layer weights of extreme learning machine and the hidden layer side-play amount individuality as improvement of differential evolution algorithm, by described improvement of differential evolution algorithm stochastic generation initial population;
Step D: the adaptive value calculating each individuality of current population;
Step e: judge whether the individuality of current population meets evolution termination condition, is perform step G, otherwise perform step F;
Step F: usage variance evolution algorithm carries out mutation operation and interlace operation successively to the individuality in parent population, obtain new progeny population, for obtained new progeny population and its parent population, perform and select operation, select individuality that in two generation populations, adaptive value is outstanding as population of future generation, return step D;
Step G: the individuality exporting adaptive value optimum in current population, and obtain corresponding optimum input layer weights and hidden layer side-play amount, perform step H;
Step H: the input layer weights and the hidden layer side-play amount that utilize described optimum, is trained training sample set respectively by extreme learning machine and learns, and obtains fender graded evolution extreme learning machine model;
Further,
Described steps A is specially: collect the wall rock geology parameter in constructing tunnel process, and set up with wall rock geology parameter for input, Grades of Surrounding Rock is the sample set of the fender graded exported;
Described step D is specially:
For the Different Individual of current population, by extreme learning machine described training sample set trained and learn, obtaining fender graded extreme learning machine model;
Adopt described fender graded extreme learning machine model to predict described test sample book collection, obtain predicated error using this predicated error as adaptive value;
Evolution termination condition in described step e is that the adaptive value of a certain individuality in current population is less than preset value;
Further, described relay station is laid in knee, tunnel and/or turnoff, tunnel.
A kind of tunnel surrounding dynamic classification device, comprising:
Be laid in tunnel tunnel face place, for obtaining the wall rock geology parameter acquisition devices of wall rock geology parameter;
The first wireless communication module be connected with described wall rock geology parameter acquisition devices;
Be laid in the firing box outside tunnel portal, the riding position of this firing box is within the coverage of mobile phone signal; The GPRS communication module that described firing box comprises the second wireless communication module and is connected with described second wireless communication module;
Be laid in the relay station between described wall rock geology parameter acquisition devices and described firing box; Described relay station connects described first wireless communication module and described second wireless communication module;
The server of described GPRS communication module is connected by GPRS network; The wall rock geology parameter that described wall rock geology parameter acquisition devices obtains transfers to described server through the first wireless communication module, relay station and firing box, this server using described wall rock geology parameter as input, fender graded result is exported by fender graded evolution extreme learning machine model, and corresponding for described fender graded result tunnel face surface information is stored in database, be distributed on network simultaneously;
Further, described wall rock geology parameter at least comprises: Rock Mass Integrality, groundwater parameter, rock mass discontinuity, the displacement of country rock Vault settlement and country rock convergence displacement; Described wall rock geology parameter acquisition devices comprises:
For obtaining geology detecting instrument and the ground penetrating radar of Rock Mass Integrality and rock mass discontinuity;
For obtaining the pore pressure gauge of the groundwater parameter of country rock;
For obtaining the displacement meter of country rock Vault settlement displacement;
And for obtaining the convergence gauge of country rock convergence displacement;
Further, described device also comprises:
Sample builds module, for building the sample set of fender graded;
Sample divides module, for described sample set being divided into training sample set and test sample book collection two parts;
Improvement of differential evolution algorithm module, for using the input layer weights of extreme learning machine and the hidden layer side-play amount individuality as improvement of differential evolution algorithm, stochastic generation initial population, and when the individuality of current population does not meet evolution termination condition, successively mutation operation and interlace operation are carried out to the individuality in parent population, obtains new progeny population, for obtained new progeny population and its parent population, perform and select operation, select individuality that in two generation populations, adaptive value is outstanding as population of future generation;
Computing module, for calculating the adaptive value of each individuality of current population;
Judge module, for judging whether the individuality of current population meets evolution termination condition;
Output module, for when the individuality of current population meets evolution termination condition, exports the individuality of adaptive value optimum in current population, and obtains corresponding optimum input layer weights and hidden layer side-play amount;
Evolution extreme learning machine study module, for utilizing input layer weights and the hidden layer side-play amount of described optimum, to be trained training sample set by extreme learning machine and learning, obtaining fender graded evolution extreme learning machine model;
Further, described evolution termination condition is that the adaptive value of a certain individuality in current population is less than preset value; Described computing module comprises:
Extreme learning machine study module, for the Different Individual for current population, to be trained described training sample set by extreme learning machine and learns, and obtains fender graded extreme learning machine model;
With adaptive value acquisition module, for adopting described fender graded extreme learning machine model to predict described test sample book collection, obtain predicated error using this predicated error as adaptive value;
Further, described relay station is laid in knee, tunnel and/or turnoff, tunnel.
Owing to have employed technique scheme, tunnel surrounding dynamic classification method provided by the invention and device thereof, by arranging wall rock geology parameter acquisition devices, and lay the wireless communication networks being applicable to tunnel and using, achieve and the wall rock geology parameter of acquisition is directly transferred to server by wireless communication networks, server by utilizing fender graded evolution extreme learning machine model carries out computational analysis to determine country rock grade fast, then country rock class information is stored to database and sharing distribution to network, complete the dynamic classification of tunnel surrounding, fender graded is fast and reliable more, solve the problem could carrying out classification after carrying out the technical scheme needs artificial detection geologic parameter of fender graded based on extreme learning machine in prior art further, the wall rock geology parameter acquiring in tunnel is convenient, equipment is flexible for installation easy, improve the efficiency of fender graded process.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the method for the invention;
Fig. 2 is the structural representation of device of the present invention;
Fig. 3 is the structured flowchart of device of the present invention;
Fig. 4 is the schematic diagram of fender graded evolution extreme learning machine model process of establishing of the present invention;
Fig. 5 is the schematic diagram of fender graded extreme learning machine model of the present invention;
Fig. 6 is the process flow diagram that the present invention draws fender graded evolution extreme learning machine model.
In figure: 1, tunnel, 2, face, 3, camera, 4, detecting head, 5, server, 6, detection case, 7, firing box, 8, hole.
Embodiment
A kind of tunnel surrounding dynamic classification method as shown in figures 1 to 6, comprises the steps:
Step 1: lay wall rock geology parameter acquisition devices in face 2 place, tunnel 1; Described wall rock geology parameter acquisition devices is connected with the first wireless communication module;
Step 2: lay firing box 7 outer and ensure that the riding position of described firing box 7 is within the coverage of mobile phone signal in hole, tunnel 18; The GPRS communication module that described firing box 7 comprises the second wireless communication module and is connected with described second wireless communication module; Described GPRS communication module is connected with server 5 via GPRS network;
Step 3: lay relay station between described wall rock geology parameter acquisition devices and described firing box 7; Described relay station connects described first wireless communication module and described second wireless communication module;
Step 4: wall rock geology parameter acquisition devices obtains the wall rock geology parameter at face 2 place, tunnel 1, and this wall rock geology parameter transfers to described server 5 through the first wireless communication module, relay station and firing box 7;
Step 5: described server 5 using described wall rock geology parameter as input, fender graded result is exported by fender graded evolution extreme learning machine model, and described fender graded result corresponding tunnel 1 face 2 information is stored in database, be distributed on network simultaneously;
Further, described wall rock geology parameter at least comprises: Rock Mass Integrality, groundwater parameter, rock mass discontinuity, the displacement of country rock Vault settlement and country rock convergence displacement; Described wall rock geology parameter acquisition devices comprises:
For obtaining geology detecting instrument and the ground penetrating radar of Rock Mass Integrality and rock mass discontinuity;
For obtaining the pore pressure gauge of the groundwater parameter of country rock;
For obtaining the displacement meter of country rock Vault settlement displacement;
And for obtaining the convergence gauge of country rock convergence displacement;
Further, described fender graded evolution extreme learning machine model draws as follows:
Steps A: the sample set building fender graded;
Step B: described sample set is divided into training sample set and test sample book collection two parts;
Step C: using the input layer weights of extreme learning machine and the hidden layer side-play amount individuality as improvement of differential evolution algorithm, by described improvement of differential evolution algorithm stochastic generation initial population;
Step D: the adaptive value calculating each individuality of current population;
Step e: judge whether the individuality of current population meets evolution termination condition, is perform step G, otherwise perform step F;
Step F: usage variance evolution algorithm carries out mutation operation and interlace operation successively to the individuality in parent population, obtain new progeny population, for obtained new progeny population and its parent population, perform and select operation, select individuality that in two generation populations, adaptive value is outstanding as population of future generation, return step D;
Step G: the individuality exporting adaptive value optimum in current population, and obtain corresponding optimum input layer weights and hidden layer side-play amount, perform step H;
Step H: the input layer weights and the hidden layer side-play amount that utilize described optimum, is trained training sample set respectively by extreme learning machine and learns, and obtains fender graded evolution extreme learning machine model;
Further,
Described steps A is specially: collect the wall rock geology parameter in tunnel 1 work progress, and set up with wall rock geology parameter for input, Grades of Surrounding Rock is the sample set of the fender graded exported;
Described step D is specially:
For the Different Individual of current population, by extreme learning machine described training sample set trained and learn, obtaining fender graded extreme learning machine model;
Adopt described fender graded extreme learning machine model to predict described test sample book collection, obtain predicated error using this predicated error as adaptive value;
Evolution termination condition in described step e is that the adaptive value of a certain individuality in current population is less than preset value;
Further, described relay station is laid in knee, tunnel 1 and/or turnoff, tunnel 1.
A kind of tunnel surrounding dynamic classification device as shown in Figures 2 and 3, comprising: be laid in face 2 place, tunnel 1, for obtaining the wall rock geology parameter acquisition devices of wall rock geology parameter; The first wireless communication module be connected with described wall rock geology parameter acquisition devices; Be laid in the firing box 7 outside hole, tunnel 18, the riding position of this firing box 7 is within the coverage of mobile phone signal; The GPRS communication module that described firing box 7 comprises the second wireless communication module and is connected with described second wireless communication module; Be laid in the relay station between described wall rock geology parameter acquisition devices and described firing box 7; Described relay station connects described first wireless communication module and described second wireless communication module; The server 5 of described GPRS communication module is connected by GPRS network; The wall rock geology parameter that described wall rock geology parameter acquisition devices obtains transfers to described server 5 through the first wireless communication module, relay station and firing box 7, this server 5 using described wall rock geology parameter as input, fender graded result is exported by fender graded evolution extreme learning machine model, and described fender graded result corresponding tunnel 1 face 2 information is stored in database, be distributed on network simultaneously; Further, described wall rock geology parameter at least comprises: Rock Mass Integrality, groundwater parameter, rock mass discontinuity, the displacement of country rock Vault settlement and country rock convergence displacement; Described wall rock geology parameter acquisition devices comprises: for obtaining geology detecting instrument and the ground penetrating radar of Rock Mass Integrality and rock mass discontinuity; For obtaining the pore pressure gauge of the groundwater parameter of country rock; For obtaining the displacement meter of country rock Vault settlement displacement; And for obtaining the convergence gauge of country rock convergence displacement; Further, described device also comprises: sample builds module, for building the sample set of fender graded; Sample divides module, for described sample set being divided into training sample set and test sample book collection two parts; Improvement of differential evolution algorithm module, for using the input layer weights of extreme learning machine and the hidden layer side-play amount individuality as improvement of differential evolution algorithm, stochastic generation initial population, and when the individuality of current population does not meet evolution termination condition, successively mutation operation and interlace operation are carried out to the individuality in parent population, obtains new progeny population, for obtained new progeny population and its parent population, perform and select operation, select individuality that in two generation populations, adaptive value is outstanding as population of future generation; Computing module, for calculating the adaptive value of each individuality of current population; Judge module, for judging whether the individuality of current population meets evolution termination condition; Output module, for when the individuality of current population meets evolution termination condition, exports the individuality of adaptive value optimum in current population, and obtains corresponding optimum input layer weights and hidden layer side-play amount; Evolution extreme learning machine study module, for utilizing input layer weights and the hidden layer side-play amount of described optimum, to be trained training sample set by extreme learning machine and learning, obtaining fender graded evolution extreme learning machine model; Further, described evolution termination condition is that the adaptive value of a certain individuality in current population is less than preset value; Described computing module comprises: extreme learning machine study module, for the Different Individual for current population, to be trained and learn by extreme learning machine to described training sample set, obtains fender graded extreme learning machine model; With adaptive value acquisition module, for adopting described fender graded extreme learning machine model to predict described test sample book collection, obtain predicated error using this predicated error as adaptive value; Further, described relay station is laid in knee, tunnel 1 and/or turnoff, tunnel 1.
Tunnel surrounding dynamic classification method provided by the invention and device thereof, by arranging wall rock geology parameter acquisition devices, and lay the wireless communication networks being applicable to tunnel 1 and using, achieve and the wall rock geology parameter of acquisition is directly transferred to server 5 by wireless communication networks, server 5 utilizes fender graded evolution extreme learning machine model to carry out computational analysis to determine country rock grade fast, then country rock class information is stored to database and sharing distribution to network, complete the dynamic classification of tunnel surrounding, fender graded is fast and reliable more, solve the problem could carrying out classification after carrying out the technical scheme needs artificial detection geologic parameter of fender graded based on extreme learning machine in prior art further, the wall rock geology parameter acquiring in tunnel is convenient, equipment is flexible for installation easy, improve the efficiency of fender graded process.Fender graded method of the present invention based on extreme learning machine, strong adaptability, have that Selecting parameter is easy, the fast and advantage that Generalization Capability is good of pace of learning.After fender graded result is distributed to network by server 5, the network terminal can realize the dynamic queries of fender graded result, be convenient to the concrete condition that associated user understands constructing tunnel in time, tunnel tunnel face is the excavated surface of constantly pushing ahead in Tunnel Engineering, tunnel tunnel face for each excavation all can carry out the process of corresponding country rock dynamic classification, the result of country rock dynamic classification process and the relevant information corresponding stored of corresponding excavated surface (tunnel tunnel face).
Wall rock geology parameter transfers to described server 5 through the first wireless communication module, relay station, the second wireless communication module and GPRS communication module; Geology detecting instrument and ground penetrating radar are used for the tectonic structure in face 2 front, forward probe tunnel 1, comprise tomography, joint, fractured weak zone etc., can feed back tunnel 1 in time and excavate front geological condition; Described geology detecting instrument and ground penetrating radar all have camera 3 and detecting head 4, detecting head 4 and camera 3 point to tunnel 1 face 2, directivity and distance is should be noted during installation, ensure that wall rock geology parameter acquisition devices is as much as possible near face 2, in addition, construction damages wall rock geology parameter acquisition devices possibly, necessarily need protect riding position and device; Described device also comprises the detection case 6 being laid in face 2 place, tunnel 1, and the remainder that geology detecting instrument and ground penetrating radar remove camera 3 and detecting head 4 is placed in described detection case 6, and described first wireless communication module is arranged in described detection case 6; By the advance geologic information of geology detecting instrument and ground penetrating radar detection, can determine the wall rock geology parameters such as Rock-mass integrity index, ROCK MASS JOINT face ductility, ORIENTATION OF DISCONTINUITY IN ROCK MASS more accurately, the determination for country rock grade provides foundation; It is inner that described pore pressure gauge can be laid in tunnel 1 country rock, detects the pore water pressure in country rock; Described wall rock geology parameter can also comprise: country rock rebound strength and terrestrial stress, and further, described wall rock geology parameter acquisition devices can also comprise: for obtain country rock rebound strength reisilometer, for obtaining the soil pressure cell of terrestrial stress size; The emitting antenna of described first wireless communication module needs to expose; The installation site of firing box 7 should ensure within the coverage of mobile phone signal, and particularly, mobile phone signal intensity percent should be not less than 15, and signal intensity specifically can obtain by sending note: AT+CSQ; In unobstructed situation, communication distance between wireless communication module is 1000 ~ 1500m to the maximum, distance between wireless communication module should be shortened according to actual conditions in inside, tunnel 1, and bend in tunnel 1 or turnoff interpolation relay station, and the antenna of each wireless communication module is exposed to ensure signal transmitting and receiving, between each wireless communication module and relay station, adopts RS485 host-host protocol; Described detection case 6, relay station and firing box 7 also comprise a 12V direct supply for powering respectively.
Fig. 4 shows the schematic diagram of fender graded evolution extreme learning machine model process of establishing of the present invention, and as shown in Figure 4, the process of establishing of fender graded evolution extreme learning machine model of the present invention comprises the steps:
1. training sample and test sample book is determined;
2. initial population is produced;
3. extreme learning machine parameter value;
4. extreme learning machine study prediction;
5. adaptive value calculates;
6. whether adaptive value meets the demands, and is, stops, otherwise performs step 7.;
7. mutation operation;
8. interlace operation;
9. select operation, return step 3..
The present invention is using the input layer weights of extreme learning machine and the hidden layer side-play amount optimized variable as improvement of differential evolution algorithm, simultaneously to train predicated error as the adaptive value of improvement of differential evolution algorithm, simultaneously according to the problem of required feedback analysis, there is representational sample set, this sample set comprises training sample set and test sample book collection, then the parameter setting improvement of differential evolution algorithm comprises population quantity, evolutionary generation, intersection factor CR and amplification factor F, and produce initial population at random, the input layer weights of the corresponding extreme learning machine of each individuality and hidden layer side-play amount, train, obtain and export weights, and then obtain the topological structure of extreme learning machine, the extreme learning machine of test sample book to training is utilized to carry out forecast test, using predicated error as the adaptive value of improvement of differential evolution algorithm, when adaptive value is not less than preset value (evolution termination condition), preset value can get 0, improvement of differential evolution algorithm makes a variation, the iterative operations such as intersection and selection, until meet evolution termination condition, now in current population the individuality of adaptive value optimum namely as best input layer weights and hidden layer side-play amount, then training obtains output layer weights, Fig. 5 shows the schematic diagram of fender graded extreme learning machine model of the present invention.
The mathematical model of extreme learning machine is as follows:
For N number of different learning sample (x
i, y
i) ∈ R
n× R
m(i=1,2 ..., N), there is L hidden layer node, single hidden layer feedforward neural network that hidden layer activation function is g (x), i-th sample output valve can adopt following formula to represent:
In formula (1), o
ibe the output valve of i-th sample, α
j=[α
j1, α
j2..., α
jn]
t, represent the connection weights of input layer to hidden layer; b
j=[b
j1, b
j2..., b
jm]
trepresent the side-play amount of hidden layer node, β
j=[β
j1, β
j2..., β
jm]
trepresent the connection weights of hidden layer i-th node to output layer, g (x) is activation function;
If described network approaches training sample with zero error, then:
Formula (2) can be referred to as:
Hβ＝Y(3)
Wherein,
Here a
j, b
j, β
jimplication identical with the implication in formula (1), H is neural network hidden layer output matrix, H (x
i) be i-th row vector of H, the jth of H is classified as input variable x
1, x
2..., x
ntime output corresponding to a jth hidden layer.
The learning algorithm of extreme learning machine generally comprises following three steps:
1. hidden layer node (neuron) number is determined, the connection weights a between setting input layer and hidden layer and the side-play amount b of hidden layer node;
2. the function that selection one infinitely can be micro-as the activation function of hidden layer node, and then calculates hidden layer output matrix H;
3. calculate output layer weights β, said process extreme learning machine, can to output layer weights generation unique solution by arranging hidden layer at random to the weights of input layer and side-play amount, if hidden layer node is abundant, and any continuous function of programmable single-chip system in theory.
Principle and the step of improvement of differential evolution algorithm (DE) are as follows:
G is made to be N for the number of vector in population
p, in G generation, vector can be expressed as x
i,G, i=1,2 ..., N
p, each vectorial individuality comprises D component, and DE algorithmic procedure is as follows:
1) initial population is produced: random generation meets the N of independent variable bound constraint in D dimension space
pindividual chromosome, formula is as follows:
i＝1,2,…,N
P；j＝1,2,…,D.
X in formula
ij u,
be respectively the upper bound and the lower bound of a jth component, rand
ij(0,1) is the random number between [0,1].
2) mutation operation: in DE algorithm, the difference in convergent-divergent population between any two object vector individualities on the 3rd the vectorial individuality be added in population, form new variable, this process is called variation.For G for each object vector, its vectorial jth component that makes a variation is:
v
i,j(G+1)＝x
r1j(G)+F(x
r2j(G)-x
r3j(G))(5)
In formula, subscript r1, r2, r3 are random integers in [1, NP] and different, and F is zoom factor, are used for regulating the step-length amplitude of vectorial difference, value in 0 ~ 2.Formula (5) is basic variation mode, is referred to as DE/rand/1 pattern; Along with the change of this formula, still other pattern can be formed, as DE/best/1, DE/best/2, DE/rand/2 etc.
3) interlace operation: by object vector x
i,Gwith variation vector v
i, G+1according to following rule hybridization, generate new sample vector u
i, G+1:
R in formula
j∈ [0,1] is the random number corresponding with a vectorial jth component; CR ∈ [0,1] is probability of crossover constant; Rn
ifor 1,2 ..., random choose integer in D, to guarantee the vectorial V that makes a variation
i(G+1), in, one-component is had at least by sample vector u
i(G+1) adopt.
4) operation is selected: adopt greedy search method to carry out selection operation.By sample vector u
i(G+1) with object vector x
i(G) compare, if u
i(G+1) corresponding less target function value, then select vectorial u
i(G+1); If instead, x
i(G) corresponding less target function value, then retain vector x
i(G).
Wherein, fender graded evolution extreme learning machine model can adopt P=DEELC (X) to represent, wherein P represents surrounding rock category, DEELC represents evolution modelling classification variation limits learning machine model, X represents in tunnel 1 digging process the wall rock geology parameter disclosing or observe formation, the convenience considering wall rock geology parameter acquiring during practical application and the reveal information principle made full use of in work progress, simultaneously with reference to the country rock gross index (BQ stage method) of " JTGD70-2004 vcehicular tunnel design specifications ", can by Rock Mass Integrality index K
v, underground water situation, ORIENTATION OF DISCONTINUITY IN ROCK MASS, the displacement of country rock Vault settlement and country rock convergence displacement these wall rock geology parameters are as the input of fender graded evolution extreme learning machine model.
The above; be only the present invention's preferably embodiment; but protection scope of the present invention is not limited thereto; anyly be familiar with those skilled in the art in the technical scope that the present invention discloses; be equal to according to technical scheme of the present invention and inventive concept thereof and replace or change, all should be encompassed within protection scope of the present invention.
Claims (10)
1. a tunnel surrounding dynamic classification method, is characterized in that described method comprises the steps:
Step 1: lay wall rock geology parameter acquisition devices in tunnel tunnel face place; Described wall rock geology parameter acquisition devices is connected with the first wireless communication module;
Step 2: lay firing box in tunnel portal and ensure that the riding position of described firing box is within the coverage of mobile phone signal outward; The GPRS communication module that described firing box comprises the second wireless communication module and is connected with described second wireless communication module; Described GPRS communication module is connected with server via GPRS network;
Step 3: lay relay station between described wall rock geology parameter acquisition devices and described firing box; Described relay station connects described first wireless communication module and described second wireless communication module;
Step 4: wall rock geology parameter acquisition devices obtains the wall rock geology parameter at tunnel tunnel face place, and this wall rock geology parameter transfers to described server through the first wireless communication module, relay station and firing box;
Step 5: described server using described wall rock geology parameter as input, fender graded result is exported by fender graded evolution extreme learning machine model, and corresponding for described fender graded result tunnel face surface information is stored in database, be distributed on network simultaneously.
2. tunnel surrounding dynamic classification method according to claim 1, is characterized in that described wall rock geology parameter at least comprises: Rock Mass Integrality, groundwater parameter, rock mass discontinuity, the displacement of country rock Vault settlement and country rock convergence displacement; Described wall rock geology parameter acquisition devices comprises:
For obtaining geology detecting instrument and the ground penetrating radar of Rock Mass Integrality and rock mass discontinuity;
For obtaining the pore pressure gauge of the groundwater parameter of country rock;
For obtaining the displacement meter of country rock Vault settlement displacement;
And for obtaining the convergence gauge of country rock convergence displacement.
3. tunnel surrounding dynamic classification method according to claim 1, is characterized in that described fender graded evolution extreme learning machine model draws as follows:
Steps A: the sample set building fender graded;
Step B: described sample set is divided into training sample set and test sample book collection two parts;
Step C: using the input layer weights of extreme learning machine and the hidden layer side-play amount individuality as improvement of differential evolution algorithm, by described improvement of differential evolution algorithm stochastic generation initial population;
Step D: the adaptive value calculating each individuality of current population;
Step e: judge whether the individuality of current population meets evolution termination condition, is perform step G, otherwise perform step F;
Step F: usage variance evolution algorithm carries out mutation operation and interlace operation successively to the individuality in parent population, obtain new progeny population, for obtained new progeny population and its parent population, perform and select operation, select individuality that in two generation populations, adaptive value is outstanding as population of future generation, return step D;
Step G: the individuality exporting adaptive value optimum in current population, and obtain corresponding optimum input layer weights and hidden layer side-play amount, perform step H;
Step H: the input layer weights and the hidden layer side-play amount that utilize described optimum, is trained training sample set respectively by extreme learning machine and learns, and obtains fender graded evolution extreme learning machine model.
4. tunnel surrounding dynamic classification method according to claim 3, is characterized in that,
Described steps A is specially: collect the wall rock geology parameter in constructing tunnel process, and set up with wall rock geology parameter for input, Grades of Surrounding Rock is the sample set of the fender graded exported;
Described step D is specially:
For the Different Individual of current population, by extreme learning machine described training sample set trained and learn, obtaining fender graded extreme learning machine model;
Adopt described fender graded extreme learning machine model to predict described test sample book collection, obtain predicated error using this predicated error as adaptive value;
Evolution termination condition in described step e is that the adaptive value of a certain individuality in current population is less than preset value.
5. tunnel surrounding dynamic classification method according to claim 1, is characterized in that described relay station is laid in knee, tunnel and/or turnoff, tunnel.
6. a tunnel surrounding dynamic classification device, is characterized in that described device comprises:
Be laid in tunnel tunnel face place, for obtaining the wall rock geology parameter acquisition devices of wall rock geology parameter;
The first wireless communication module be connected with described wall rock geology parameter acquisition devices;
Be laid in the firing box outside tunnel portal, the riding position of this firing box is within the coverage of mobile phone signal; The GPRS communication module that described firing box comprises the second wireless communication module and is connected with described second wireless communication module;
Be laid in the relay station between described wall rock geology parameter acquisition devices and described firing box; Described relay station connects described first wireless communication module and described second wireless communication module;
The server of described GPRS communication module is connected by GPRS network; The wall rock geology parameter that described wall rock geology parameter acquisition devices obtains transfers to described server through the first wireless communication module, relay station and firing box, this server using described wall rock geology parameter as input, fender graded result is exported by fender graded evolution extreme learning machine model, and corresponding for described fender graded result tunnel face surface information is stored in database, be distributed on network simultaneously.
7. tunnel surrounding dynamic classification device according to claim 6, is characterized in that described wall rock geology parameter at least comprises: Rock Mass Integrality, groundwater parameter, rock mass discontinuity, the displacement of country rock Vault settlement and country rock convergence displacement; Described wall rock geology parameter acquisition devices comprises:
For obtaining geology detecting instrument and the ground penetrating radar of Rock Mass Integrality and rock mass discontinuity;
For obtaining the pore pressure gauge of the groundwater parameter of country rock;
For obtaining the displacement meter of country rock Vault settlement displacement;
And for obtaining the convergence gauge of country rock convergence displacement.
8. tunnel surrounding dynamic classification device according to claim 6, is characterized in that described device also comprises:
Sample builds module, for building the sample set of fender graded;
Sample divides module, for described sample set being divided into training sample set and test sample book collection two parts;
Improvement of differential evolution algorithm module, for using the input layer weights of extreme learning machine and the hidden layer side-play amount individuality as improvement of differential evolution algorithm, stochastic generation initial population, and when the individuality of current population does not meet evolution termination condition, successively mutation operation and interlace operation are carried out to the individuality in parent population, obtains new progeny population, for obtained new progeny population and its parent population, perform and select operation, select individuality that in two generation populations, adaptive value is outstanding as population of future generation;
Computing module, for calculating the adaptive value of each individuality of current population;
Judge module, for judging whether the individuality of current population meets evolution termination condition;
Output module, for when the individuality of current population meets evolution termination condition, exports the individuality of adaptive value optimum in current population, and obtains corresponding optimum input layer weights and hidden layer side-play amount;
Evolution extreme learning machine study module, for utilizing input layer weights and the hidden layer side-play amount of described optimum, to be trained training sample set by extreme learning machine and learning, obtaining fender graded evolution extreme learning machine model.
9. tunnel surrounding dynamic classification device according to claim 8, is characterized in that described evolution termination condition is that the adaptive value of a certain individuality in current population is less than preset value; Described computing module comprises:
Extreme learning machine study module, for the Different Individual for current population, to be trained described training sample set by extreme learning machine and learns, and obtains fender graded extreme learning machine model;
With adaptive value acquisition module, for adopting described fender graded extreme learning machine model to predict described test sample book collection, obtain predicated error using this predicated error as adaptive value.
10. tunnel surrounding dynamic classification device according to claim 6, is characterized in that described relay station is laid in knee, tunnel and/or turnoff, tunnel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510925540.5A CN105353427A (en) | 2015-12-14 | 2015-12-14 | Tunnel surrounding rock dynamic grading method and device thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510925540.5A CN105353427A (en) | 2015-12-14 | 2015-12-14 | Tunnel surrounding rock dynamic grading method and device thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105353427A true CN105353427A (en) | 2016-02-24 |
Family
ID=55329428
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510925540.5A Pending CN105353427A (en) | 2015-12-14 | 2015-12-14 | Tunnel surrounding rock dynamic grading method and device thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105353427A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107193001A (en) * | 2017-06-05 | 2017-09-22 | 广西大学 | Karst Tunnel unfavorable geology prediction expert system and implementation method based on GPR |
| CN107526117A (en) * | 2017-07-06 | 2017-12-29 | 天津科技大学 | SVEL Forecasting Methodology based on autocoding and the study joint network that transfinites |
| CN107575234A (en) * | 2017-09-07 | 2018-01-12 | 长沙有色冶金设计研究院有限公司 | A kind of deep-well support design method |
| CN109163997A (en) * | 2018-09-18 | 2019-01-08 | 天津大学 | A kind of rock surface strength detection method based on sonograph deep learning |
| CN109165406A (en) * | 2018-07-20 | 2019-01-08 | 大连海事大学 | The quick dynamic classification prediction technique of country rock during a kind of constructing tunnel |
| CN109241627A (en) * | 2018-09-07 | 2019-01-18 | 大连海事大学 | The dynamic shoring method of probability hierarchical and the device of Automated Design supporting scheme |
| CN109374456A (en) * | 2018-09-03 | 2019-02-22 | 山东大学 | Suitable for tunnel surrounding quality grading and rate of decay discriminating gear and method |
| CN109557284A (en) * | 2019-01-31 | 2019-04-02 | 四川省交通运输厅交通勘察设计研究院 | A kind of tunnel surrounding rank intelligent quickly decision-making system and method |
| CN109783913A (en) * | 2018-12-30 | 2019-05-21 | 中铁十四局集团有限公司 | The supporting construction selection method and device in tunnel |
| CN110109895A (en) * | 2019-03-29 | 2019-08-09 | 山东大学 | Fender graded unified prediction and application suitable for TBM driving tunnel |
| CN110472363A (en) * | 2019-08-22 | 2019-11-19 | 山东大学 | Surrouding rock deformation grade prediction technique and system suitable for Railway Tunnel |
| CN110671150A (en) * | 2019-10-14 | 2020-01-10 | 中铁第四勘察设计院集团有限公司 | Surrounding rock grading correction method and device based on tunnel buried depth and weathering layer thickness |
| CN110780343A (en) * | 2020-01-02 | 2020-02-11 | 四川大学 | Automatic microseismic signal identification method based on waveform frequency band characteristics |
| WO2020125668A1 (en) * | 2018-12-18 | 2020-06-25 | 中国铁建重工集团股份有限公司 | Method and system for automatically identifying surrounding rock level by applying while-drilling parameters |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2265743A (en) * | 1992-04-01 | 1993-10-06 | Inst Francais Du Petrole | Geological data collection |
| CN202140118U (en) * | 2011-07-25 | 2012-02-08 | 中铁二十一局集团有限公司 | Tunnel construction safety early warning system based on three-dimensional digital tunnel platform |
| 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 |
| CN204436476U (en) * | 2015-02-16 | 2015-07-01 | 大连海事大学 | Based on the Tunnel Landslide automated watch-keeping facility of structure reaction |
-
2015
- 2015-12-14 CN CN201510925540.5A patent/CN105353427A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2265743A (en) * | 1992-04-01 | 1993-10-06 | Inst Francais Du Petrole | Geological data collection |
| CN202140118U (en) * | 2011-07-25 | 2012-02-08 | 中铁二十一局集团有限公司 | Tunnel construction safety early warning system based on three-dimensional digital tunnel platform |
| 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 |
| CN204436476U (en) * | 2015-02-16 | 2015-07-01 | 大连海事大学 | Based on the Tunnel Landslide automated watch-keeping facility of structure reaction |
Non-Patent Citations (3)
| Title |
|---|
| 姜谙男 等: ""桥基隧道明挖施工多元信息集成智能方法与应用"", 《岩土力学》 * |
| 陈吉华 等: ""公路隧道围岩动态分级研究"", 《西部探矿工程》 * |
| 黄生文 等: ""隧道监控量测在围岩动态分级中的应用"", 《长沙理工大学学报（自然科学版）》 * |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107193001B (en) * | 2017-06-05 | 2020-04-14 | 广西大学 | Karst tunnel unfavorable geology forecasting expert system based on ground penetrating radar and implementation method |
| CN107193001A (en) * | 2017-06-05 | 2017-09-22 | 广西大学 | Karst Tunnel unfavorable geology prediction expert system and implementation method based on GPR |
| CN107526117B (en) * | 2017-07-06 | 2019-08-13 | 天津科技大学 | Based on autocoding and transfinites and learn the acoustic speed prediction technique of joint network |
| CN107526117A (en) * | 2017-07-06 | 2017-12-29 | 天津科技大学 | SVEL Forecasting Methodology based on autocoding and the study joint network that transfinites |
| CN107575234A (en) * | 2017-09-07 | 2018-01-12 | 长沙有色冶金设计研究院有限公司 | A kind of deep-well support design method |
| CN109165406A (en) * | 2018-07-20 | 2019-01-08 | 大连海事大学 | The quick dynamic classification prediction technique of country rock during a kind of constructing tunnel |
| CN109374456A (en) * | 2018-09-03 | 2019-02-22 | 山东大学 | Suitable for tunnel surrounding quality grading and rate of decay discriminating gear and method |
| CN109241627A (en) * | 2018-09-07 | 2019-01-18 | 大连海事大学 | The dynamic shoring method of probability hierarchical and the device of Automated Design supporting scheme |
| CN109241627B (en) * | 2018-09-07 | 2020-08-18 | 大连海事大学 | Probability grading dynamic support method and device for automatically designing support scheme |
| CN109163997A (en) * | 2018-09-18 | 2019-01-08 | 天津大学 | A kind of rock surface strength detection method based on sonograph deep learning |
| WO2020125668A1 (en) * | 2018-12-18 | 2020-06-25 | 中国铁建重工集团股份有限公司 | Method and system for automatically identifying surrounding rock level by applying while-drilling parameters |
| CN109783913A (en) * | 2018-12-30 | 2019-05-21 | 中铁十四局集团有限公司 | The supporting construction selection method and device in tunnel |
| CN109557284A (en) * | 2019-01-31 | 2019-04-02 | 四川省交通运输厅交通勘察设计研究院 | A kind of tunnel surrounding rank intelligent quickly decision-making system and method |
| CN109557284B (en) * | 2019-01-31 | 2021-07-20 | 四川省交通运输厅交通勘察设计研究院 | Tunnel surrounding rock level intelligent rapid determination system and method |
| CN110109895A (en) * | 2019-03-29 | 2019-08-09 | 山东大学 | Fender graded unified prediction and application suitable for TBM driving tunnel |
| CN110109895B (en) * | 2019-03-29 | 2021-05-28 | 山东大学 | Surrounding rock grading combined prediction method suitable for TBM tunneling tunnel and application |
| CN110472363A (en) * | 2019-08-22 | 2019-11-19 | 山东大学 | Surrouding rock deformation grade prediction technique and system suitable for Railway Tunnel |
| CN110671150A (en) * | 2019-10-14 | 2020-01-10 | 中铁第四勘察设计院集团有限公司 | Surrounding rock grading correction method and device based on tunnel buried depth and weathering layer thickness |
| CN110671150B (en) * | 2019-10-14 | 2021-04-27 | 中铁第四勘察设计院集团有限公司 | Surrounding rock grading correction method and device based on tunnel buried depth and weathering layer thickness |
| CN110780343A (en) * | 2020-01-02 | 2020-02-11 | 四川大学 | Automatic microseismic signal identification method based on waveform frequency band characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105353427A (en) | Tunnel surrounding rock dynamic grading method and device thereof | |
| CN104182622B (en) | Based on feedback analysis method and its device in the constructing tunnel of extreme learning machine | |
| Lai et al. | Prediction of soil deformation in tunnelling using artificial neural networks | |
| CN109061731B (en) | Surface wave dispersion and bulk wave compose the global optimization method than joint inversion shallow-layer speed | |
| CN113239439B (en) | Shield construction ground surface settlement prediction system and method | |
| Hajihassani et al. | 3D prediction of tunneling-induced ground movements based on a hybrid ANN and empirical methods | |
| CN105526925A (en) | Method for optimized designing of regional land subsidence level monitoring network | |
| CN104965969A (en) | Inversion method for surrounding rock mechanical parameters of large cavern group | |
| CN105320817B (en) | A kind of Shallow-buried Large-span city tunnel flat ratio determines method | |
| CN205210324U (en) | A country rock geology parameter acquisition device that is used for tunnel country rock developments hierarchical | |
| Jiang et al. | Landslide risk prediction by using GBRT algorithm: Application of artificial intelligence in disaster prevention of energy mining | |
| CN109948198A (en) | A kind of fender graded method for evaluating reliability based on nonlinear function | |
| Song et al. | Back analysis of geomechanical parameters using hybrid algorithm based on difference evolution and extreme learning machine | |
| Li et al. | Stochastic, goal-oriented rapid impact modeling of uncertainty and environmental impacts in poorly-sampled sites using ex-situ priors | |
| Wang et al. | Direct interval prediction of landslide displacements using least squares support vector machines | |
| CN106055865A (en) | Method and system for evaluating influences of tunnel construction on water environment | |
| Nguyen et al. | Prediction of ground subsidence due to underground mining through time using multilayer feed-forward artificial neural networks and back-propagation algorithm–case study at Mong Duong underground coal mine (Vietnam) | |
| CN113989431A (en) | Construction method of three-dimensional visual dynamic monitoring structure model of underground water resource | |
| CN109711063A (en) | A kind of station tunneling drilling depth method of adjustment and device suitable for upper-soft lower-hard ground | |
| Shaorui et al. | Predictions of overbreak blocks in tunnels based on the wavelet neural network method and the geological statistics theory | |
| Yu et al. | A back-propagation neural-network-based displacement back analysis for the identification of the geomechanical parameters of the Yonglang landslide in China | |
| Arshid | Knowledge based prediction of standard penetration resistance of soil using geotechnical database | |
| CN104636576A (en) | Rapid subway construction risk evaluation method based on weighted average method | |
| Chehreghani et al. | Rock mass excavatability estimation using artificial neural network | |
| Kim et al. | A fundamental study on the automation of tunnel blasting design using a machine learning model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160224 |