CN106529725A - Gas outburst prediction method based on firefly algorithm and SOM network - Google Patents
Gas outburst prediction method based on firefly algorithm and SOM network Download PDFInfo
- Publication number
- CN106529725A CN106529725A CN201611005261.8A CN201611005261A CN106529725A CN 106529725 A CN106529725 A CN 106529725A CN 201611005261 A CN201611005261 A CN 201611005261A CN 106529725 A CN106529725 A CN 106529725A
- Authority
- CN
- China
- Prior art keywords
- som
- value
- gas
- initial
- input
- 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
- 238000000034 method Methods 0.000 title claims abstract description 33
- 241000254158 Lampyridae Species 0.000 title abstract 3
- 238000012549 training Methods 0.000 claims abstract description 17
- 239000007789 gas Substances 0.000 claims description 64
- 210000002569 neuron Anatomy 0.000 claims description 22
- GNBHRKFJIUUOQI-UHFFFAOYSA-N fluorescein Chemical compound O1C(=O)C2=CC=CC=C2C21C1=CC=C(O)C=C1OC1=CC(O)=CC=C21 GNBHRKFJIUUOQI-UHFFFAOYSA-N 0.000 claims description 18
- 241000131894 Lampyris noctiluca Species 0.000 claims description 17
- 239000003245 coal Substances 0.000 claims description 14
- 210000004205 output neuron Anatomy 0.000 claims description 8
- 238000005065 mining Methods 0.000 claims description 5
- 238000010521 absorption reaction Methods 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 claims description 4
- 238000009792 diffusion process Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 210000002364 input neuron Anatomy 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000004069 differentiation Effects 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 230000008447 perception Effects 0.000 claims description 2
- 238000013519 translation Methods 0.000 claims description 2
- 239000000203 mixture Substances 0.000 claims 1
- 238000005457 optimization Methods 0.000 abstract description 2
- 238000007781 pre-processing Methods 0.000 abstract 2
- 238000012216 screening Methods 0.000 abstract 1
- 238000004088 simulation Methods 0.000 abstract 1
- 230000008901 benefit Effects 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 230000007935 neutral effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 235000015170 shellfish Nutrition 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Development Economics (AREA)
- Molecular Biology (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Entrepreneurship & Innovation (AREA)
- Mathematical Physics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a gas outburst prediction method based on a firefly algorithm and an SOM network. The gas outburst prediction method comprises the steps of acquiring initial input variables according to a numerical simulation experiment, performing data preprocessing on the acquired initial input variables to acquire an initial input feature vector, enabling the initial input feature vector to act as input of a training sample so as to build an initial gas outburst SOM network prediction model, screening the initial input variables by adopting an MIV algorithm to acquire a final input variable, performing data preprocessing on the acquired final input variable so as to acquire a final input feature vector, enabling the final input feature vector to act as input of the training sample so as to build a GSO-SOM network, optimizing a weight value and a threshold value of the SOM network by adopting a GSO algorithm, acquiring an optimal weight value and an optimal threshold value, acquiring a gas outburst SOM network prediction model based on firefly algorithm optimization, predicting the gas outburst type, and outputting a prediction result of gas outburst. The gas outburst prediction method provided by the invention has the characteristics of high precision, high reliability, good accuracy and the like, and can be widely applied to the field of prediction.
Description
Technical field
The present invention relates to electric powder prediction, more particularly to a kind of gas based on glowworm swarm algorithm and SOM networks is dashed forward
Go out Forecasting Methodology.
Background technology
With the increasing of China's coal mine excavation depth and intensity, Gas Outburst problem seriously governs coal in China work
The development of industry, brings great threat to the security of the lives and property of Safety of Coal Mine Production and staff.It is pre- for Gas Outburst
Survey problem, at present, Chinese scholars propose various Gas Outburst methods.Rough set theory, support vector machine, shellfish are adopted mainly
The methods such as leaf this classification method, fuzzy logic, neutral net are predicted to Gas Outburst.Rough set theory is processing fuzzy and not
Determine, but its decision rules is very unstable, accuracy is poor, and is based on complete letter
Breath system, during processing data, can often run into loss of data phenomenon.Support vector machine are solving small sample, non-linear and higher-dimension mould
There is in formula identification problem advantage, but identification ability is easily affected by inherent parameters.Bayes Method needs known definite dividing
Other probability, and can not actually provide definite difference probability.Fuzzy logic needs certain priori, and parameter is selected
With stronger dependency.Neutral net has simple structure and very strong problem solving ability, and can preferably process and make an uproar
Sound data, but there is local optimum in algorithm, and convergence is poor, limited reliability.
As can be seen here, in the prior art, there is low precision, poor reliability, predict the outcome and deposit in gas outbursts Prediction method
The problems such as relatively large deviation.
The content of the invention
In view of this, present invention is primarily targeted at providing a kind of high accuracy, good reliability, predicting the outcome accurate watt
This outburst prediction method.
In order to achieve the above object, technical scheme proposed by the present invention is：
A kind of gas outbursts Prediction method based on glowworm swarm algorithm and SOM networks, the gas outbursts Prediction method bag
Include following steps：
The mechanism of step 1, analysis Gas Outburst, is constituted using tunnel, gases at high pressure and pressure transducer by cavity, is simulated
Experimental system carry out Numerical Experiment, obtain coal seam gas-bearing capacity W, coal seam thickness h, mining depth H, gas pressure P, watt
This diffusion initial speed Δ P, the firmness coefficient F of coal and seven initial input variable Xs of geology destructiveness S；
Step 2, the pretreatment that the initial input variable X is carried out data, obtain initial input characteristic vector T；
Step 3, using initial input characteristic vector T as the input of training sample, set up initial Gas Outburst SOM
Network Prediction Model；
Step 4, SOM networks are trained using initial input characteristic vector T, and using MIV algorithms to initial
Input variable X is screened, and finally enters variable X ' after being screened；
Step 5, the variable X ' that finally enters is carried out into the pretreatment of data, obtain finally entering characteristic vector T '；
Step 6, using the characteristic vector T ' that finally enters as the input of training sample, set up GSO-SOM networks；
Step 7, the weights and threshold value of SOM networks are optimized using GSO algorithms；
Step 8, differentiation meet whether end condition is set up；If set up, execution step 9；If be false, perform
Step 7；
Step 9, the weights and threshold value that obtain optimum, obtain the SOM Network Prediction Models optimized based on glowworm swarm algorithm, right
Gas Outburst type is predicted, and output predicts the outcome.
In sum, the gas outbursts Prediction method based on glowworm swarm algorithm and SOM networks of the present invention by obtain
After initial input variable carries out the pretreatment of data, described initial input characteristic vector is obtained, by initial input characteristic vector
As the input of training sample, initial Gas Outburst SOM Network Prediction Models are set up, initial input is become using MIV algorithms
Amount is screened, and obtains finally entering variable, is finally entered what is obtained after the pretreatment that variable carries out data, obtains final defeated
Enter characteristic vector, input of the characteristic vector as training sample will be finally entered, GSO-SOM networks are set up, using GSO algorithms pair
The weights and threshold value of SOM networks are optimized, and obtain optimum weights and threshold value, obtain the gas optimized based on glowworm swarm algorithm
Prominent SOM Network Prediction Models, predict Gas Outburst type, export predicting the outcome for Gas Outburst, dash forward so as to improve gas
Go out precision, accuracy and the reliability of prediction.
Description of the drawings
Fig. 1 is a kind of flow chart of gas outbursts Prediction method based on glowworm swarm algorithm and SOM networks of the present invention.
Fig. 2 is SOM network topology structures schematic diagram of the present invention.
Specific embodiment
It is to make the object, technical solutions and advantages of the present invention clearer, right below in conjunction with the accompanying drawings and the specific embodiments
The present invention is described in further detail.
Fig. 1 is a kind of flow chart of gas outbursts Prediction method based on glowworm swarm algorithm and SOM networks of the present invention.
As shown in figure 1, gas outbursts Prediction method of the present invention, comprises the steps：
The mechanism of step 1, analysis Gas Outburst, is constituted using tunnel, gases at high pressure and pressure transducer by cavity, is simulated
Experimental system carry out Numerical Experiment, obtain coal seam gas-bearing capacity W, coal seam thickness h, mining depth H, gas pressure P, watt
This diffusion initial speed Δ P, the firmness coefficient F of coal and seven initial input variable Xs of geology destructiveness S；
Step 2, the pretreatment that the initial input variable X is carried out data, obtain initial input characteristic vector T；
Step 3, using initial input characteristic vector T as the input of training sample, set up initial Gas Outburst SOM
Network Prediction Model；
Step 4, SOM networks are trained using initial input characteristic vector T, and using MIV algorithms to initial
Input variable X is screened, and finally enters variable X ' after being screened；
Step 5, the variable X ' that finally enters is carried out into the pretreatment of data, obtain finally entering characteristic vector T '；
Step 6, using the characteristic vector T ' that finally enters as the input of training sample, set up GSO-SOM networks；
Step 7, the weights and threshold value of SOM networks are optimized using GSO algorithms；
Step 8, differentiation meet whether end condition is set up；If set up, execution step 9；If be false, perform
Step 7；
Step 9, the weights and threshold value that obtain optimum, obtain the SOM Network Prediction Models optimized based on glowworm swarm algorithm, right
Gas Outburst type is predicted, and output predicts the outcome.
In a word, the gas outbursts Prediction method based on glowworm swarm algorithm and SOM networks of the present invention is initial by what is obtained
After input vector carries out the pretreatment of data, described initial characteristicses vector is obtained, using initial characteristicses vector as training sample
Input, set up initial Gas Outburst SOM Network Prediction Models, initial input variable screened using MIV algorithms, obtained
To variable is finally entered, finally enter what is obtained after the pretreatment that variable carries out data, obtain finally entering characteristic vector, will
Characteristic vector is finally entered as the input of training sample, GSO-SOM networks are set up, using weights of the GSO algorithms to SOM networks
It is optimized with threshold value, obtains optimum weights and threshold value, the Gas Outburst SOM networks obtained based on glowworm swarm algorithm optimization is pre-
Survey model, predict Gas Outburst type, export Gas Outburst and predict the outcome, so as to improve gas outbursts Prediction precision,
Accuracy and reliability.
In step 2 of the present invention, the calculating formula of the data prediction is as follows：
Wherein, t (i) is the sample value of i-th initial input variable, xactI () is the reality of i-th initial input variable
Value, xminI () is the minima of i-th initial input variable, xmaxI () is the maximum of i-th initial input variable.
In the inventive method, the step 4 comprises the steps：
Step 41, calculating initial input characteristic vector T are adopted in the distance of moment t to all output nodes
Eucliden distances, calculating formula are as follows：
Wherein, TiT () is value of the initial input characteristic vector in t, wijFor i-th input neuron node with
Connection weight between j-th output neuron node.
Step 42, selection produce minimum range djNode as the neuron for most matching,Neuron i
X () is triumph neuron.
Step 43, the weights and threshold value that to triumph neuron, update SOM networks respectively, the calculating formula of weights are as follows：
wij(t+1)=wij(t)+η(t)hJ, i(x)[Ti(t)-wij(t)]
Wherein, η (t) is the learning efficiency, 0 ＜ η (t) ＜ 1, and is reduced with time t is dull, hJ, i (t)T () is nerve of winning
Neighborhood function around first, calculating formula are as follows：
Wherein, rj, ri(x)It is SOM network output node j respectively, the position of i (x)；
The calculating formula of SOM network thresholds is as follows：
B=el-log[(1-β)e-log(b)+β×α]
Wherein, learning rates of the β for threshold value, the output of 0 ＜ β ＜ 1, α for output layer neuron, α=[α1, α2...,
α36],
Step 44, enter line translation to initial input characteristic vector T, concrete transform is as follows,
Step 45, using initial input characteristic vector T after conversion as SOM networks input, calculate SOM networks output
Value Yi 1And Yi 2, concrete calculating formula is as follows：
Difference IV of step 46, the output valve of calculating SOM networksi, Mean Impact Value MIV is obtained, concrete calculating formula is as follows：
IVt=Yi 1-Yi 2
Step 47, it is ranked up according to the absolute value of calculated MIV, deleting affects less initial input to output
Variable, obtains finally entering variable X ' for mining depth H, gas pressure P, gas diffusion initial speed Δ p, the firmness coefficient F of coal
With geology destructiveness S.
In step 5 of the present invention, the calculating formula of the data prediction is as follows：
Wherein, the sample value of t ' (i) to be i-th finally enter variables, x 'actI () is i-th finally enters the reality of variable
Actual value, x 'minI () is i-th finally enters the actual value of variable, x 'maxI () is i-th finally enters the maximum of variable.
In the inventive method, the step 6 comprises the steps：
Step 61, using the characteristic vector T ' that finally enters as the neuron of GSO-SOM network input layers, GSO-SOM
For one-dimensional, the output layer of GSO-SOM networks is a two-dimensional network for having 6 × 6 output neurons to the input layer of network, output layer
Neuron line up a neighbour structure, each neuron is laterally attached with other neurons around it, each input nerve
Unit is connected to all output neurons.
Step 62, by the connection weight and threshold value of GSO-SOM networks using real number vector form encode, constitute Lampyridea at the beginning of
Beginning population, initializes number n of Lampyridea population, captivation factor beta0, absorption coefficient of light γ and randomness factor alpha0, wherein, inhale
Gravitational coefficients β0The random number that=1, absorption coefficient of light γ are distributed for [0,1], randomness factor alpha0∈ [0,1].
In the inventive method, the step 7 comprises the steps：
Step 71, calculating Lampyridea individual adaptation degree functional value, concrete calculating formula are as follows：
Wherein, f be Lampyridea ideal adaptation angle value, Z for training sample number, ykFor actual output valve, tkTo expect
Output valve.
Step 72, calculating fluorescein value lk(t), fluorescein value lk(t) and current location xkT () represents each Lampyridea
Individual k, fluorescein value lkT the calculating formula of () is as follows：
lk(t+1)=(1-0.95 δ1)×lk(t)+ξ1×J(xk(t+1))
Wherein, k is that Lampyridea is individual, xkT () is the individual current location of Lampyridea, lkT () is Lampyridea individuality in t
The size of fluorescein value, l during secondary iterationk(t+1) it is the size of the individual fluorescein value in the t+1 time iteration of Lampyridea, J (xk(t
+ 1) it is) target function value, δ1For fluorescein value volatility coefficient, ξ1To strengthen coefficient.
, more than the Lampyridea quantity of itself, calculating formula is as follows for step 73, calculating：
Mk(t)={ q：dkq(t) ＜ rk；lk(t) ＜ lq(t)}
Wherein, MkT () is more than the Lampyridea number of itself, d for all fluorescein values in sensing rangekqFor individual k and q it
Between distance, rkFor the perception radius.
Step 74, acquisition fluorescence are most strong individual, update Lampyridea position, and specific formula for calculation is as follows：
Wherein, PiiIndividual for most hyperfluorescence, P is that all fluorescein values are individual more than the Lampyridea of itself in sensing range,
lpThe size of (t) for P fluorescein values in the t time iteration.
In step 8 of the present invention, end condition is specially training error less than 0.01 or iterationses reach 1000.
Step 9 of the present invention is specially：Optimum weights and threshold value are obtained, obtains dashing forward based on the gas that glowworm swarm algorithm optimizes
Go out neural network prediction model, using the characteristic vector T ' that finally enters as the input of forecast model, to Gas Outburst type
It is predicted, output predicts the outcome.
Embodiment
Characteristic vector will be finally entered as input, the SOM Network Prediction Models part instruction optimized based on glowworm swarm algorithm
Practice data as shown in table 1.Predicting the outcome for Gas Outburst is as shown in table 2.
1 part training sample of table
Table 2 predicts the outcome
From the data in table 2, it can be seen that when train epochs are 10, training data 1,2 is divided into a class, and 3,4,5,6,7,8 are divided
For another kind of, SOM networks have carried out preliminary classification to data, when train epochs are 100,1 and 2,3 and 4,5 and 6,7 and 8 quilts
It is divided into same class, at this moment SOM networks can correctly be classified to the type of Gas Outburst to data Further Division.From SOM nets
From the point of view of the test result of network, the SOM Network Prediction Models optimized using glowworm swarm algorithm can judge Gas Outburst exactly
Type, accuracy are high.
In sum, presently preferred embodiments of the present invention is these are only, is not intended to limit protection scope of the present invention.
All any modification, equivalent substitution and improvements within the spirit and principles in the present invention, made etc., should be included in the present invention's
Within protection domain.
Claims (9)
1. a kind of gas outbursts Prediction method based on glowworm swarm algorithm and SOM networks, it is characterised in that the Gas Outburst is pre-
Survey method comprises the steps：
The mechanism of step 1, analysis Gas Outburst, using by cavity, simulates the reality of tunnel, gases at high pressure and pressure transducer composition
Check system carries out Numerical Experiment, obtains coal seam gas-bearing capacity W, coal seam thickness h, mining depth H, gas pressure P, gas and puts
Scattered initial velocity Δ p, the firmness coefficient F of coal and seven initial input variable Xs of geology destructiveness S；
Step 2, the pretreatment that the initial input variable X is carried out data, obtain initial input characteristic vector T；
Step 3, using initial input characteristic vector T as the input of training sample, set up initial Gas Outburst SOM networks
Forecast model；
Step 4, SOM networks are trained using initial input characteristic vector T, and using MIV algorithms to initial input
Variable X is screened, and finally enters variable X ' after being screened；
Step 5, the variable X ' that finally enters is carried out into the pretreatment of data, obtain finally entering characteristic vector T '；
Step 6, using the characteristic vector T ' that finally enters as the input of training sample, set up GSO-SOM networks；
Step 7, the weights and threshold value of SOM networks are optimized using GSO algorithms；
Step 8, differentiation meet whether end condition is set up；If set up, execution step 9；If be false, execution step
7；
Step 9, the weights and threshold value that obtain optimum, obtain the SOM Network Prediction Models optimized based on glowworm swarm algorithm, to gas
Prominent type is predicted, and output predicts the outcome.
2. gas outbursts Prediction method according to claim 1, it is characterised in that in step 2, the data prediction
Calculating formula is as follows：
Wherein, t (i) is the sample value of i-th initial input variable, xactI () is the actual value of i-th initial input variable,
xminI () is the minima of i-th initial input variable, xmaxI () is the maximum of i-th initial input variable.
3. gas outbursts Prediction method according to claim 1, it is characterised in that the step 3 is specially：Using initial
Neurons of the input feature value T as SOM network input layers, the input layer of SOM networks is one-dimensional；The output layer of SOM networks
For a two-dimensional network for having 6 × 6 output neurons, the neuron of output layer lines up a neighbour structure, and each neuron is same
Other neurons around it are laterally attached, and each input neuron is connected to all output neurons；SOM networks it is initial
Connection weight is IW=[w1, w2..., w8]36×8, the initial threshold of SOM networks is
Wherein, w1, w2... w8For less non-zero random number.
4. gas outbursts Prediction method according to claim 1, it is characterised in that the step 4 includes walking in detail below
Suddenly：
The distance of step 41, calculating initial input characteristic vector T in moment t to all output nodes, using Eucliden
Distance, calculating formula are as follows：
Wherein, TiT () is value of the initial input characteristic vector in t, wijFor i-th input neuron node and jth
Connection weight between individual output neuron node；
Step 42, selection produce minimum range djNode as the neuron for most matching,Neuron i (x) is
Triumph neuron；
Step 43, the weights and threshold value that to triumph neuron, update SOM networks respectively, the calculating formula of weights are as follows：
wij(t+1)=wij(t)+η(t)hJ, i (x)[Ti(t)-wij(t)]
Wherein, η (t) is the learning efficiency, 0 ＜ η (t) ＜ 1, and is reduced with time t is dull, hJ, i (t)T () is triumph neuron week
The neighborhood function for enclosing, calculating formula are as follows：
Wherein, rj, ri(x)It is SOM network output node j respectively, the position of i (x)；
The calculating formula of SOM network thresholds is as follows：
B=e1-log[(1-β)e-log(b)+β×α]
Wherein, learning rates of the β for threshold value, the output of 0 ＜ β ＜ 1, α for output layer neuron, α=[α1, α2..., α36],
Step 44, enter line translation to initial input characteristic vector T, concrete transform is as follows,
Step 45, using initial input characteristic vector T after conversion as SOM networks input, calculate SOM networks output valve
WithConcrete calculating formula is as follows：
Difference IV of step 46, the output valve of calculating SOM networksi, Mean Impact Value MIV is obtained, concrete calculating formula is as follows：
Step 47, it is ranked up according to the absolute value of calculated MIV, deleting affects less initial input to become output
Amount, obtains final input variable X ' for mining depth H, gas pressure P, gas diffusion initial speed Δ p, the firmness coefficient F of coal
With geology destructiveness S.
5. gas outbursts Prediction method according to claim 1, it is characterised in that in step 5, the data prediction
Calculating formula is as follows：
Wherein, the sample value of t ' (i) to be i-th finally enter variables, x 'actI () is i-th finally enters the actual value of variable,
x′minI () is i-th finally enters the actual value of variable, x 'maxI () is i-th finally enters the maximum of variable.
6. gas outbursts Prediction method according to claim 1, it is characterised in that the step 6 includes walking in detail below
Suddenly：
Step 61, using the characteristic vector T ' that finally enters as the neuron of GSO-SOM network input layers, GSO-SOM networks
Input layer for one-dimensional, the output layer of GSO-SOM networks is a two-dimensional network for having 6 × 6 output neurons, the god of output layer
Jing units line up a neighbour structure, and each neuron is laterally attached with other neurons around it, each input neuron
It is connected to all output neurons；
Step 62, by the connection weight and threshold value of GSO-SOM networks using real number vector form encode, constitute Lampyridea initially plant
Group, initializes number n of Lampyridea population, captivation factor beta0, absorption coefficient of light γ and randomness factor alpha0, wherein, captivation
Factor beta0The random number that=1, absorption coefficient of light γ are distributed for [0,1], randomness factor alpha0∈ [0,1].
7. gas outbursts Prediction method according to claim 1, it is characterised in that the step 7 includes walking in detail below
Suddenly：
Step 71, calculating Lampyridea individual adaptation degree functional value, concrete calculating formula are as follows：
Wherein, f be Lampyridea ideal adaptation angle value, Z for training sample number, ykFor actual output valve, tkFor desired defeated
Go out value；
Step 72, calculating fluorescein value lk(t), fluorescein value lk(t) and current location xkT () represents each Lampyridea individual
K, fluorescein value lkT the concrete calculating formula of () is as follows：
lk(t+1)=(1-0.95 δ1)×lk(t)+ξ1×J(xk(t+1))
Wherein, k is that Lampyridea is individual, xkT () is the individual current location of Lampyridea, lkT () changes at the t time for Lampyridea is individual
For when fluorescein value size, lk(t+1) it is the size of the individual fluorescein value in the t+1 time iteration of Lampyridea, J (xk(t+1))
For target function value, δ1For fluorescein value volatility coefficient, ξ1To strengthen coefficient；
, more than the Lampyridea quantity of itself, calculating formula is as follows for step 73, calculating：
Mk(t)={ q：dkq(t) ＜ rk；lk(t) ＜ lq(t)}
Wherein, MkT () is more than the Lampyridea number of itself, d for all fluorescein values in sensing rangekqFor between individual k and q
Distance, rkFor the perception radius；
Step 74, acquisition fluorescence are most strong individual, update Lampyridea position, and specific formula for calculation is as follows：
Wherein, PijIndividual for most hyperfluorescence, P is that all fluorescein values are individual more than the Lampyridea of itself in sensing range, lp(t)
For the size of P fluorescein values in the t time iteration.
8. gas outbursts Prediction method according to claim 1, it is characterised in that in step 8, the end condition are instruction
Practice error less than 0.01 or iterationses reach 1000.
9. gas outbursts Prediction method according to claim 1, it is characterised in that the step 9 is specially：Obtain optimum
Weights and threshold value, obtain based on glowworm swarm algorithm optimize Gas Outburst neural network prediction model, finally enter described
Inputs of the characteristic vector T ' as forecast model, is predicted to Gas Outburst type, and output predicts the outcome.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611005261.8A CN106529725A (en) | 2016-11-10 | 2016-11-10 | Gas outburst prediction method based on firefly algorithm and SOM network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611005261.8A CN106529725A (en) | 2016-11-10 | 2016-11-10 | Gas outburst prediction method based on firefly algorithm and SOM network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106529725A true CN106529725A (en) | 2017-03-22 |
Family
ID=58352045
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611005261.8A Pending CN106529725A (en) | 2016-11-10 | 2016-11-10 | Gas outburst prediction method based on firefly algorithm and SOM network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106529725A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657361A (en) * | 2018-12-21 | 2019-04-19 | 成都四方伟业软件股份有限公司 | Gas content prediction method, terminal device and computer readable storage medium |
| CN109801364A (en) * | 2019-01-23 | 2019-05-24 | 厦门嵘拓物联科技有限公司 | A kind of 3-dimensional digital modeling method and digitlization workshop management system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103034904A (en) * | 2013-01-11 | 2013-04-10 | 上海电机学院 | Firefly grouping method, as well as power dispatching system and power dispatching method based on same |
| CN104453875A (en) * | 2014-10-29 | 2015-03-25 | 中国石油集团川庆钻探工程有限公司 | Shale gas reservoir recognition method based on self-organizing competitive neural network |
| CN105335698A (en) * | 2015-09-15 | 2016-02-17 | 河南理工大学 | Gear failure diagnosis method based on adaptive genetic algorithm and SOM (Self-Organizing Map) network |
-
2016
- 2016-11-10 CN CN201611005261.8A patent/CN106529725A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103034904A (en) * | 2013-01-11 | 2013-04-10 | 上海电机学院 | Firefly grouping method, as well as power dispatching system and power dispatching method based on same |
| CN104453875A (en) * | 2014-10-29 | 2015-03-25 | 中国石油集团川庆钻探工程有限公司 | Shale gas reservoir recognition method based on self-organizing competitive neural network |
| CN105335698A (en) * | 2015-09-15 | 2016-02-17 | 河南理工大学 | Gear failure diagnosis method based on adaptive genetic algorithm and SOM (Self-Organizing Map) network |
Non-Patent Citations (5)
| Title |
|---|
| SAMANTHA JACOBS 等: "Self organizing maps for monitoring parameter deterioration of DC and AC motors", 《2013 PROCEEDINGS OF IEEE SOUTHEASTCON》 * |
| 付华 等: "基于ASGSO-ENN算法的瓦斯涌出量动态预测模型", 《计算机工程》 * |
| 卢永艳: "《上市公司财务困境分析与预测》", 30 June 2012, 西安电子科技大学出版社 * |
| 阎平凡等: "《人工神经网络与模拟进化计算第2版》", 30 September 2005, 清华大学出版社 * |
| 雷艳玲: "煤与瓦斯突出预测SOM-SVM模型", 《中国优秀硕士学位论文全文数据库-工程科技Ⅰ辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109657361A (en) * | 2018-12-21 | 2019-04-19 | 成都四方伟业软件股份有限公司 | Gas content prediction method, terminal device and computer readable storage medium |
| CN109657361B (en) * | 2018-12-21 | 2022-09-13 | 成都四方伟业软件股份有限公司 | Gas content prediction method, terminal device and computer readable storage medium |
| CN109801364A (en) * | 2019-01-23 | 2019-05-24 | 厦门嵘拓物联科技有限公司 | A kind of 3-dimensional digital modeling method and digitlization workshop management system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103226741B (en) | Public supply mains tube explosion prediction method | |
| CN110321603A (en) | A kind of depth calculation model for Fault Diagnosis of Aircraft Engine Gas Path | |
| CN106453293B (en) | A kind of network security situation prediction method based on improved BPNN | |
| CN106372402B (en) | The parallel method of fuzzy region convolutional neural networks under a kind of big data environment | |
| CN112052992B (en) | Deep learning-based construction project progress prediction system and method | |
| Olatunji et al. | Improved sensitivity based linear learning method for permeability prediction of carbonate reservoir using interval type-2 fuzzy logic system | |
| CN109242147A (en) | Signal fused fan condition prediction technique based on Bp neural network | |
| CN114969953B (en) | Optimized shield underpass tunnel design method and equipment based on Catboost-NSGA-III | |
| CN107222333A (en) | A kind of network node safety situation evaluation method based on BP neural network | |
| CN106096723B (en) | It is a kind of that complex industrial properties of product appraisal procedure is used for based on hybrid neural networks algorithm | |
| CN108446771A (en) | A method of preventing Sale Forecasting Model over-fitting | |
| CN106970981A (en) | A kind of method that Relation extraction model is built based on transfer matrix | |
| CN104732067A (en) | Industrial process modeling forecasting method oriented at flow object | |
| Ketova et al. | Development of an effective adaptive forecasting system based on the combination of neural network and genetic algorithm | |
| CN106529725A (en) | Gas outburst prediction method based on firefly algorithm and SOM network | |
| CN113505492A (en) | Scheduling method of cross-basin water transfer project based on digital twin technology | |
| CN115359197A (en) | Geological curved surface reconstruction method based on spatial autocorrelation neural network | |
| CN113642785A (en) | Method, system and equipment for long-term prediction of space debris orbit based on prior information | |
| Sugunnasil et al. | Modelling a neural network using an algebraic method | |
| CN113537540A (en) | Heat supply gas consumption prediction model based on automatic characteristic engineering | |
| Sun et al. | The use of neural network in defense audit nonlinear dynamic processing under the background of big data | |
| Rajbhandari et al. | The use of artificial neural networks (ANN) for preliminary design of high-rise buildings | |
| CN113283174B (en) | Reservoir productivity prediction method, system and terminal based on algorithm integration and self-control | |
| Yang | [Retracted] Evaluation and Analysis of Multimedia Collaborative Building Design Relying on Particle Swarm Optimization Algorithm | |
| CN107886159A (en) | Virtual site |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170322 |