CN109325470A - Working face in the pit homework type intelligent identification Method based on gas density parameter - Google Patents
Working face in the pit homework type intelligent identification Method based on gas density parameter Download PDFInfo
- Publication number
- CN109325470A CN109325470A CN201811243460.1A CN201811243460A CN109325470A CN 109325470 A CN109325470 A CN 109325470A CN 201811243460 A CN201811243460 A CN 201811243460A CN 109325470 A CN109325470 A CN 109325470A
- Authority
- CN
- China
- Prior art keywords
- gas density
- signal sequence
- homework type
- working face
- local
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Sampling And Sample Adjustment (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of working face in the pit homework type intelligent identification Method based on gas density parameter.Method obtains face gas concentration data first from mine safety monitoring system, and it is stored in Gas explosion burns database, adaptive noise reduction filtering processing is carried out to the gas density data in database, then history gas density set of data samples is established on the basis of homework type is divided with sensitive features parameter extraction, adaptation module Rough-neural network prediction model is constructed, its corresponding homework type is recognized using the prediction model of building.The method of the present invention noise reduction effect is significant, the effective information obtained in gas density time series can be retained while effectively removing noise, it is able to achieve the analysis and identification of working face homework type, can preferably realize the depth excavation and secondary use of gas density data.
Description
Technical field
The present invention relates to mine gas disaster monitoring prevention and control fields more particularly to a kind of based on gas density data characteristics
Coal mine excavation working face homework type intellectual analysis and recognition methods.
Background technique
China is the first in the world coal big country, according to China " energy Long-and Medium-term Development planning outline (2004-2020) "
The energy based on coal will be adhered to by explicitly pointing out China within following one period.Currently, coal in China exploitation already into
Enter deep mining period, and mining depth just increases year by year.In deep mining period, mine gas pressure increases, ground temperature increases,
Crustal stress increases, and mine disaster is got worse.Coal mining accident disaster seriously harms coal in China situation of production, hinders
The healthy, lasting of coal industry, stable development.Perfect coal mining accident diaster prevention and control Security Strategies, deepens coal mining accident calamity
Theory is realized in evil prevention and control, promotes coal mining accident disaster and supervises Control Technology, for ensuring Safety of Coal Mine Production, promoting national economy
Comprehensively, it develops in a healthy way and is of great significance.
Gas accident, protrusion, bump, coal dust accident, floods, fire, roof accident in coal production etc. are main
The disaster overwhelming majority occurs in getting working face, close with digging, the job relations such as drill, and is occupied with accident during excavation operation
It is more.Mining safety monitoring is the important means of mine accident damage control, obtains and efficiently use having for working face in the pit in time
Information is imitated, the work quality for promoting accident prevention and Disaster control is facilitated.Currently, working face operating environment is monitored both at home and abroad
The monitoring for primarily focusing on the operating environments parameter such as gas density, wind speed, temperature of monitoring technology research.High methane coal roadway tunneling
Working face, the gas situation of working face are the environmental parameter closely related with operating condition, the longtime running of monitoring and controlling system
So that colliery scene has accumulated a large amount of gas monitor data.Researchers at home and abroad have to gas monitor data using early
It is of interest, and a large amount of research has been carried out with the advanced technologies such as data mining, artificial intelligence means, such as utilize Gas explosion burns sequence
Carry out gas density prediction, using gas emission come interpretation outburst hazard etc., these work to mining safety monitoring
Huge facilitation is arrived.
However, forefathers are less to the directly concern of working face forward operation situation, the job execution situation of underground is mainly come
Derived from scheduling scheme of arrangement or artificial statistic record, inevitably there is mistake and careless mistake, and in timeliness by
Great limitation.Currently, the normalization means as Accidents Disasters prevention and control, mining safety monitoring is at this stage via simple
The functions diversification such as safety monitoring diagnoses towards disaster, job scheduling and production management, synthesization, intelligent direction development and
Show non-contact, mobilism, multi-parameter integratedization, modular integrated feature.The influence for adopting movement industry is often many things
Therefore immediate cause, by the relevant information of underground actively, precisely, the flow chart of quick identification working face, and be included in and be
System database all has great practical meaning for work such as Accidents assessment analysis, disaster diagnosis, emergency disposal and accident investigations
Justice is supported basic information is provided for the development of coal mine safety monitoring early warning technology.
Summary of the invention
For the missing of prior art application, the present invention provides a kind of underground works based on Mine Methane parameter
Face homework type intelligent identification Method.The special concentration data for gas monitor of this method is pre-processed, and original is effectively reduced
The noise signal of beginning time series retains and highlights effective gas density sequence signature, passes through building gas ginseng on the basis of secondary
Number sensitive parameter system and homework type sample training library, using the multilayer feedforward nerve net of error backpropagation algorithm training
Network carries out INTELLIGENT IDENTIFICATION to the flow chart type of working face in the pit.
Technical solution of the present invention specifically includes following key step:
Step 1, the original signal sequence x (t) that gas density is obtained by fire damp concentration sensor, as history number
According to deposit gas density database；
Step 2, the actual conditions according to shaft production, by each original signal sequence x (t) according to the class of jobs of working face
Type is divided；
Step 3 carries out based on the special adaptive of local mean value decomposition the gas density data in gas density database
Noise reduction process should be filtered, the gas density sequence after obtaining noise reduction；
Step 4, using the gas density sequence after noise reduction as chaos time sequence, according to the division of homework type to all kinds of
The corresponding data of homework type extract sensitive features parameter respectively, and building homework type divides sensitive parameter system；
Step 5, the homework type neural network intelligent recognition model for constructing working face in the pit, it is dense according to established gas
Degree database and sensitive features parameter are input in model and are trained；
Step 6 is known using the intelligence that the model after training carries out homework type to the gas density data to be measured newly obtained
Not, the corresponding homework type result of gas density data to be measured is obtained.
The present invention creatively judges working face in the pit homework type using gas density parameter, has played the work of data
With wherein adaptive-filtering noise reduction process has made special noise reduction process and to seem dense with the incoherent gas of homework type
Degree parameter can embody characterization working face in the pit homework type.
In the step 2, the homework type of working face is divided into as blown out, coal cutting, coal breakage, relieves, operation of drilling, faces
When supporting, installation anchor pole, unrelated with other operation of maintenance.Other operation forms or regular combination can also be added wherein
Several homework types, it is main to be determined according to the feasibility recognized according to safety monitoring early warning actual demand and homework type
's.
In the step 4, the sensitive features parameter includes the mean value, mean square deviation, variance, pole of gas density data
Difference, Differential Characteristics value etc..Specifically specific product component can also be selected as sensitive features parameter.In specific implementation, or
Improvement parameter or other mining again parameters on such parameter basis；Also it can be used and produced in Noise reducing of data preprocessing process
Raw other sensitive features parameters including product vector sequence etc..
In the step 3, adaptive-filtering noise reduction process specifically:
3.1, obtain whole extreme point n of original signal sequence x (t)i, i=1,2,3 ..., I, I is extreme value points, then
According to extreme point, the average value of all neighborhood local-area extreme points is acquired as local mean value mi=(ni+ni+1)/2 and local envelope
Estimated value is as local envelope value ai=(ni-ni+1)/2；
3.2, to all local mean value miWith all local envelope value aiMultiple smoothing processing point is carried out using moving average method
Local mean function m ' is not obtained1(t)、m′2... and local envelope function a ' (t),1(t)、a′2(t) ..., subscript 1,2 indicates the
Once, the function result that second of smoothing processing is obtained；
3.3, iteration is handled after the smoothing processing each time of step 3.2 until obtaining pure FM signal sk(t):
After the first time smoothing processing of step 3.2, first time smoothing processing is isolated from original signal sequence x (t)
The local mean function m ' of acquisition1(t) retain and obtain separating signal sequence h among first1(t)=x (t)-m '1(t), then
The local envelope function a ' obtained further according to first time smoothing processing1(t) it carries out demodulating to obtain the first intermediate demodulation signal sequence s1
(t)=h1(t)/a′1(t)；
Then by the first intermediate demodulation signal sequence s1(t) it is considered as original signal sequence x (t) and repeats step, in conjunction with second
The local mean function m ' that secondary smoothing processing obtains1(t) and local envelope function a '1(t) it calculates to obtain and separates signal among second
Sequence and the second intermediate demodulation signal sequence；
With this iterative processing until the intermediate demodulation signal sequence of kth meets -1≤sKAnd its office next time (t)≤1
Domain envelope function a 'K+1(t)=1, K is the number being iterated, then by the intermediate demodulation signal sequence s of kthK(t) as pure
FM signal stops iterative processing；Specific iterative process indicates are as follows:
In formula, K is the total degree being iterated；
3.4, it will the above pure FM signal sK(t) all local envelope function a ' generated in iterative processk(t) it is multiplied and obtains
Obtain the envelope signal A of the first product component (Production function, PF)1(t)=a '1(t)a′2(t)…a′K(t), k
=1,2,3 ..., K, then by the envelope signal A of the first product component1(t) pure FM signal s corresponding with itsK(t) it is multiplied and obtains
First product component PF of original signal sequence1(t)=A1(t)sk(t)；
3.5, by the first product component PF1(t) it is separated from original signal sequence x (t), reservation obtains the first remnants
Signal u1(t), and with the first residue signal u1(t) it repeats the above steps 3.1-3.4 iterative processing for original signal sequence x (t),
Until going out all product components by J multi-cycle separation, and meet J residue signal uJIt (t) is a monotonic function, thus structure
Build the decomposition expression formula of following original signal sequence x (t):
Wherein, j indicates that the secondary ordinal number of step 3.1-3.4 iterative processing, j=1,2,3 ..., J, J indicate that step 3.1-3.4 changes
The total degree of generation processing；
Original gas density signal is transformed to the sum of one group of product component and a residual components.
3.6, correlation analysis is carried out to each product component and original signal sequence x (t) of acquisition together, takes all multiply
The energy value of integration amount constitutes frequency modulation separation of the global minimum of sequence as low-and high-frequency ingredient, before frequency modulation separation
The high fdrequency component of product component be all used as noise product component, place will be reconstructed to signal after cancelling noise product component
Reason, the gas density sequence dx (t) after obtaining noise reduction.
Processing is reconstructed to signal to refer to each product component progress adduction.
The homework type neural network intelligent recognition model is combined using rough set theory and artificial neural network
Rough-neural network (rough membership neural network, RMNN), with realize to down-hole mining working face work
The intelligent recognition of type, gas density data are quick when calculating separately each homework type using the thick neuron of neural network input layer
Feel characteristic parameter, and neural unit is replaced using fuzzy neuron, using the sensitive features of partial history gas density data
Parameter is completed to carry out intelligent recognition to the corresponding homework type of gas data recently as training sample.
The present invention influences prediction model precision of prediction for much noise is usually contained in current mine gas density data
The problem of, it is decomposed by local mean value and history gas density sequence data is resolved into multiple pure FM signal components, then will divided
FM signal function after solution can while effectively removing noise by low-pass filter adaptive threshold noise reduction filtering
Retain the effective information in gas density time series to greatest extent, denoising effect is ideal.
The beneficial effects of the present invention are:
The defects of present invention is for the timeliness retardance and inaccuracy of manual record flow chart and type constructs one
The Rough-neural network prediction model of adaptation module, model can statistical parameters or its structure based on a variety of gas density parameters
At comprehensive parameters according to gas density sequence, can be preferably real for realizing the analysis and identification of working face homework type
The depth of existing gas density data is excavated and secondary use.
The method of the present invention noise reduction effect is significant, can retain while effectively removing noise and obtain gas density time series
In effective information.
The method of the present invention, which can be realized, precisely sentences knowledge to underground real work type, meets current mine safety monitoring system
Demand for development.
Detailed description of the invention
Fig. 1 is flow chart of the embodiment of the present invention；
Fig. 2 is original gas density data time series figure provided in an embodiment of the present invention；
Fig. 3 is the gas density data time series figure after noise reduction provided in an embodiment of the present invention；
Fig. 4 is the Rough-neural network model homework type classification chart provided in an embodiment of the present invention based on comprehensive parameters.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Implement below
Example is not intended to limit the scope of the invention for illustrating the present invention.
The embodiment of the present invention is specific as follows:
The gas density data that certain channel of coal mining working surface is acquired by gas concentration sensor, as embodiment research pair
As, while the artificial homework type tracking carried out in constant time range and record, as sample database, process is embodied as shown in Figure 1,
The specific method is as follows.
Step 1 automatically continuously obtains development end Gas Dynamic Time Series data by mine safety monitoring system,
The gas data that mine uses KJ90 (substation-F16) safety monitoring system to obtain in the present embodiment, gas concentration sensor are
KG9701A methane transducer, gas density data acquiring frequency are every 5 minutes data.Remember fire damp concentration sensor
Original gas density original series x (t) is obtained, database is stored in as historical data, arbitrarily intercepts the gas density of certain period
Signal is as shown in Figure 2；
Step 2, the actual conditions according to shaft production carry out classifying rationally to the homework type of working face.The present embodiment
Middle mine operation mode blows out, coal cutting, coal breakage, relieves, operation of drilling, interim branch using " four or six system " entire duty cycle
The several work forms such as shield, installation anchor pole, unrelated with other operation of maintenance, for convenient for illustrating, according only to adopting in the present embodiment
Homework type is divided into three kinds of " no operatton ", " operation of drilling " and " excavation operation " by the intensity of operation.
Step 3, basis screen n group gas density sequence sample to the manual record of homework type respectively within the corresponding period
This, and data are carried out with the adaptive-filtering noise reduction process decomposed based on local mean value, specific implementation process is as follows；
3.1, obtain whole extreme point n of original signal sequence x (t)i, i=1,2,3 ..., I, I is extreme value points, then
According to extreme point, the average value of all neighborhood local-area extreme points is acquired as local mean value mi=(ni+ni+1)/2 and local envelope
Estimated value is as local envelope value ai=(ni-ni+1)/2；
3.2, to all local mean value miWith all local envelope value aiMultiple smoothing processing point is carried out using moving average method
Local mean function m ' is not obtained1(t)、m′2... and local envelope function a ' (t),1(t)、a′2(t),…；
3.3, iteration is handled after the smoothing processing each time of step 3.2 until obtaining pure FM signal sk(t):
After the first time smoothing processing of step 3.2, first time smoothing processing is isolated from original signal sequence x (t)
The local mean function m ' of acquisition1(t) retain and obtain separating signal sequence h among first1(t)=x (t)-m '1(t), then
The local envelope function a ' obtained further according to first time smoothing processing1(t) it carries out demodulating to obtain the first intermediate demodulation signal sequence s1
(t)=h1(t)/a′1(t)；
Then by the first intermediate demodulation signal sequence s1(t) it is considered as original signal sequence x (t) and repeats step, in conjunction with second
The local mean function m ' that secondary smoothing processing obtains1(t) and local envelope function a '1(t) it calculates to obtain and separates signal among second
Sequence and the second intermediate demodulation signal sequence；
With this iterative processing until the intermediate demodulation signal sequence of kth meets -1≤sKAnd its office next time (t)≤1
Domain envelope function a 'K+1(t)=1, K is the number being iterated, then by the intermediate demodulation signal sequence s of kthK(t) as pure
FM signal stops iterative processing；Specific iterative process indicates are as follows:
In formula, K is the total degree being iterated；
3.4, it will the above pure FM signal sK(t) all local envelope function a ' generated in iterative processk(t) it is multiplied and obtains
Obtain the envelope signal A of the first product component1(t)=a '1(t)a′2(t)…a′K(t), k=1,2,3 ..., K then multiply first
The envelope signal A of integration amount1(t) pure FM signal s corresponding with itsK(t) it is multiplied and obtains the first product point of original signal sequence
Measure PF1(t)=A1(t)sk(t)；
3.5, by the first product component PF1(t) it is separated from original signal sequence x (t), reservation obtains the first remnants
Signal u1(t), and with the first residue signal u1(t) it repeats the above steps 3.1-3.4 iterative processing for original signal sequence x (t),
Until going out all product components by J multi-cycle separation, and meet J residue signal uJIt (t) is a monotonic function, thus will
Original gas density signal is transformed to the sum of one group of product component and a residual components.
3.6, correlation analysis is carried out to each product component and original signal sequence x (t) of acquisition together, takes product point
Frequency modulation separation of the global minimum of amount as low-and high-frequency ingredient, by the high fdrequency component of the product component before frequency modulation separation
All as noise product component, processing will be reconstructed to signal after cancelling noise product component, by each product component into
Row adduction, the gas density sequence dx (t) after obtaining noise reduction, as shown in Figure 3.
Step 4 takes gas density data of the time step for T after noise reduction, building homework type division sensitive parameter system.
The sensitive parameter that the present embodiment is selected mainly includes the system such as gas density mean value, mean square deviation, variance, very poor, Differential Characteristics value
Parameter is counted, respectively represents the Gas feature under different work type condition, each statistical parameter is as shown in table 1.
1 gas density sequence sensitive features parameter set of table
Wherein, T indicates the time step to calculate each sensitive parameter gas data, and Xmax indicates gas in the taken period
The maximum value and Xmin of concentration respectively indicate the minimum value of gas density in the taken period.
Step 5, the homework type neural network intelligent recognition model for constructing working face in the pit, it is dense according to established gas
Degree database and sensitive features parameter are input in model and are trained.
The present embodiment has selected three homework types, i.e. three kinds of excavation operation, bore operation and no operatton situations, remembers respectively
For w1、w2And w3, respective statistical parameter feature is calculated separately according to the sample of different work type, is summarized as shown in table 2.It adopts
Combined with rough set theory and artificial neural network Rough-neural network (rough membership neural network,
RMNN), the characteristic quantity as classification such as gas density mean value, mean square deviation, variance, very poor, Differential Characteristics value is extracted, according to each
The characteristics of characteristic quantity, carries out homework type classification.In embodiment, sample training, and root are carried out according to different Parameters respectively
Weight i is assigned according to the precision and contribution of each parameter, then determines comprehensive parameters Y=d1i1+d2i2+…+d8i8.Wherein i1~i8 points
The contribution weight of aforementioned 8 sensitive indicators is not indicated.
Using the neural network model of comprehensive Y value and aforementioned building, sample training is re-started, until can be preferably real
The accurate identification of existing homework type, the embodiment are as shown in Figure 4 according to the recognition result of comprehensive Y value.
2 different work type sample statistics parameter attribute value of table summarizes
Step 6 is known using the intelligence that the model after training carries out homework type to the gas density data to be measured newly obtained
Not, and adaptively it improves.
The flow chart intelligence of the present invention based on gas density feature it can be seen from the algorithm arrangement of above-mentioned offer
Energy recognition methods, mainly using gas density parameter as research object, for the purpose of realizing working face homework type intelligent recognition, into
The special noise reduction process of row, using Rough-neural network model realization to the intelligent recognition of down-hole mining working face job category, algorithm
Realization facilitate the handling situations for quickly and effectively obtaining underground, can widen mine safety monitoring system disaster diagnose,
The functional diversities of job scheduling and production management etc. and intelligence, meanwhile, identification result is analyzed Accidents assessment, is answered
The work such as anxious disposition and accident investigation also have reference value.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify to technical solution documented by previous embodiment, or some or all of the technical features are equal
Replacement；And these are modified or replaceed, model defined by the claims in the present invention that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (5)
1. a kind of working face in the pit homework type intelligent identification Method based on gas density parameter, it is characterised in that method is main
Including following key step:
Step 1, the original signal sequence x (t) that gas density is obtained by fire damp concentration sensor, are stored in gas density number
According to library；
Step 2 divides each original signal sequence x (t) according to the homework type of working face；
Step 3 carries out adaptive-filtering noise reduction process to the gas density data in gas density database, after obtaining noise reduction
Gas density sequence；
Step 4, using the gas density sequence after noise reduction as chaos time sequence, according to the division of homework type to all kinds of operations
The corresponding data of type extract sensitive features parameter respectively；
Step 5, the homework type neural network intelligent recognition model for constructing working face in the pit, according to established gas density number
It is input in model and is trained according to library and sensitive features parameter；
Step 6, the intelligent recognition that using the model after training the gas density data to be measured newly obtained are carried out with homework type.
2. a kind of working face in the pit homework type intelligent recognition side based on gas density parameter according to claim 1
Method, it is characterised in that: in the step 2, the homework type of working face is divided into as blown out, coal cutting, coal breakage, relieves, drills
Operation, gib, installation anchor pole, maintenance operation unrelated with other.
3. a kind of working face in the pit homework type intelligent recognition side based on gas density parameter according to claim 1
Method, it is characterised in that: in the step 4, the sensitive features parameter includes the mean value, mean square deviation, side of gas density data
Poor, very poor, Differential Characteristics value etc..
4. a kind of working face in the pit homework type intelligent recognition side based on gas density parameter according to claim 1
Method, it is characterised in that: in the step 3, adaptive-filtering noise reduction process specifically:
3.1, obtain whole extreme point n of original signal sequence x (t)i, i=1,2,3 ..., I, I is extreme value points, then according to pole
It is worth point, acquires the average value of all neighborhood local-area extreme points as local mean value mi=(ni+ni+1)/2 and local envelope estimated value
As local envelope value ai=(ni-ni+1)/2；
3.2, to all local mean value miWith all local envelope value aiMultiple smoothing processing is carried out using moving average method to obtain respectively
To local mean function m '1(t)、m′2... and local envelope function a ' (t),1(t)、a′2(t),…；
3.3, iteration is handled after the smoothing processing each time of step 3.2 until obtaining pure FM signal sk(t):
After the first time smoothing processing of step 3.2, the acquisition of first time smoothing processing is isolated from original signal sequence x (t)
Local mean function m '1(t) retain and obtain separating signal sequence h among first1(t)=x (t)-m '1(t), then root again
The local envelope function a ' obtained according to first time smoothing processing1(t) it carries out demodulating to obtain the first intermediate demodulation signal sequence s1(t)=
h1(t)/a′1(t)；
Then by the first intermediate demodulation signal sequence s1(t) it is considered as original signal sequence x (t) and repeats step, it is smooth in conjunction with second
Handle the local mean function m ' obtained1(t) and local envelope function a '1(t) calculate obtain second among separation signal sequence and
Second intermediate demodulation signal sequence；
With this iterative processing until the intermediate demodulation signal sequence of kth meets -1≤sKAnd its local packet next time (t)≤1
Network function a 'K+1(t)=1, K is the number being iterated, then by the intermediate demodulation signal sequence s of kthK(t) it is used as pure frequency modulation
Signal stops iterative processing；Specific iterative process indicates are as follows:
In formula, K is the total degree being iterated；
3.4, it will the above pure FM signal sK(t) all local envelope function a ' generated in iterative processk(t) it is multiplied and obtains the
The envelope signal A of one product component1(t)=a '1(t)a′2(t)…a′K(t), k=1,2,3 ..., K, then by the first product point
The envelope signal A of amount1(t) pure FM signal s corresponding with itsK(t) it is multiplied and obtains the first product component of original signal sequence
PF1(t)=A1(t)sk(t)；
3.5, by the first product component PF1(t) it is separated from original signal sequence x (t), reservation obtains the first residue signal
u1(t), and with the first residue signal u1(t) it repeats the above steps 3.1-3.4 iterative processing for original signal sequence x (t), until
Go out all product components by J multi-cycle separation, and meets J residue signal uJ(t) be a monotonic function, thus construct with
The decomposition expression formula of lower original signal sequence x (t):
Wherein, j indicates that the secondary ordinal number of step 3.1-3.4 iterative processing, j=1,2,3 ..., J, J indicate at step 3.1-3.4 iteration
The total degree of reason；
3.6, correlation analysis is carried out to each product component and original signal sequence x (t) of acquisition together, takes all products point
The energy value of amount constitutes frequency modulation separation of the global minimum of sequence as low-and high-frequency ingredient, by multiplying before frequency modulation separation
The high fdrequency component of integration amount is all used as noise product component, processing will be reconstructed to signal after cancelling noise product component,
Gas density sequence dx (t) after obtaining noise reduction, wherein the energy value of product component according to component all sequences data square
Be calculated.
5. a kind of working face in the pit homework type intelligent recognition side based on gas density parameter according to claim 1
Method, it is characterised in that:
The homework type neural network intelligent recognition model is combined thick using rough set theory and artificial neural network
Neural network (rough membership neural network, RMNN) utilizes the thick neuron point of neural network input layer
Gas density data sensitive characteristic parameter when not calculating each homework type, and neural unit is replaced using fuzzy neuron.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811243460.1A CN109325470B (en) | 2018-10-24 | 2018-10-24 | Intelligent underground working face operation type identification method based on gas concentration parameters |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811243460.1A CN109325470B (en) | 2018-10-24 | 2018-10-24 | Intelligent underground working face operation type identification method based on gas concentration parameters |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109325470A true CN109325470A (en) | 2019-02-12 |
| CN109325470B CN109325470B (en) | 2021-09-03 |
Family
ID=65263225
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811243460.1A Active CN109325470B (en) | 2018-10-24 | 2018-10-24 | Intelligent underground working face operation type identification method based on gas concentration parameters |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109325470B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112308306A (en) * | 2020-10-27 | 2021-02-02 | 贵州工程应用技术学院 | Multi-mode input coal and gas outburst risk prediction method |
| CN112686513A (en) * | 2020-12-23 | 2021-04-20 | 精英数智科技股份有限公司 | Method and device for identifying operation state of underground working face and production decision system |
| CN114046179A (en) * | 2021-09-15 | 2022-02-15 | 山东省计算中心（国家超级计算济南中心） | Method for intelligently identifying and predicting underground safety accident based on CO monitoring data |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6289328B2 (en) * | 1998-04-17 | 2001-09-11 | The United States Of America As Represented By The Secretary Of The Navy | Chemical sensor pattern recognition system and method using a self-training neural network classifier with automated outlier detection |
| US20040133531A1 (en) * | 2003-01-06 | 2004-07-08 | Dingding Chen | Neural network training data selection using memory reduced cluster analysis for field model development |
| CN104537199A (en) * | 2014-10-20 | 2015-04-22 | 天地（常州）自动化股份有限公司 | Method for recognizing production processes of mine working surface by gas density time sequence |
| CN106296435A (en) * | 2016-08-18 | 2017-01-04 | 西安科技大学 | A kind of mine gas monitoring disorder data recognition method |
| EP3126787A2 (en) * | 2014-04-01 | 2017-02-08 | Saudi Arabian Oil Company | Tomographic imaging of multiphase flows |
| CN106503798A (en) * | 2015-09-08 | 2017-03-15 | 中国科学院沈阳自动化研究所 | Based on rough set and the method for diagnosing faults of the pump of BP neural network |
| US9652684B2 (en) * | 2012-11-08 | 2017-05-16 | Total Sa | Image processing for classification and segmentation of rock samples |
-
2018
- 2018-10-24 CN CN201811243460.1A patent/CN109325470B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6289328B2 (en) * | 1998-04-17 | 2001-09-11 | The United States Of America As Represented By The Secretary Of The Navy | Chemical sensor pattern recognition system and method using a self-training neural network classifier with automated outlier detection |
| US20040133531A1 (en) * | 2003-01-06 | 2004-07-08 | Dingding Chen | Neural network training data selection using memory reduced cluster analysis for field model development |
| US9652684B2 (en) * | 2012-11-08 | 2017-05-16 | Total Sa | Image processing for classification and segmentation of rock samples |
| EP3126787A2 (en) * | 2014-04-01 | 2017-02-08 | Saudi Arabian Oil Company | Tomographic imaging of multiphase flows |
| CN104537199A (en) * | 2014-10-20 | 2015-04-22 | 天地（常州）自动化股份有限公司 | Method for recognizing production processes of mine working surface by gas density time sequence |
| CN106503798A (en) * | 2015-09-08 | 2017-03-15 | 中国科学院沈阳自动化研究所 | Based on rough set and the method for diagnosing faults of the pump of BP neural network |
| CN106296435A (en) * | 2016-08-18 | 2017-01-04 | 西安科技大学 | A kind of mine gas monitoring disorder data recognition method |
Non-Patent Citations (3)
| Title |
|---|
| 李刚: "瓦斯浓度的分形分析与混沌预测模型研究", 《中国博士学位论文全文数据库工程科技Ⅰ辑》 * |
| 毕慧杰等: "基于多因素模式识别的煤与瓦斯突出预测研究", 《中国安全生产科学技术》 * |
| 王川等: "基于瓦斯涌出特征的机掘面作业类型判识技术", 《煤矿安全》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112308306A (en) * | 2020-10-27 | 2021-02-02 | 贵州工程应用技术学院 | Multi-mode input coal and gas outburst risk prediction method |
| CN112686513A (en) * | 2020-12-23 | 2021-04-20 | 精英数智科技股份有限公司 | Method and device for identifying operation state of underground working face and production decision system |
| CN112686513B (en) * | 2020-12-23 | 2024-07-09 | 精英数智科技股份有限公司 | Method and device for identifying working state of underground working face and production decision-making system |
| CN114046179A (en) * | 2021-09-15 | 2022-02-15 | 山东省计算中心（国家超级计算济南中心） | Method for intelligently identifying and predicting underground safety accident based on CO monitoring data |
| CN114046179B (en) * | 2021-09-15 | 2023-09-22 | 山东省计算中心（国家超级计算济南中心） | Method for intelligently identifying and predicting underground safety accidents based on CO monitoring data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109325470B (en) | 2021-09-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109325470A (en) | Working face in the pit homework type intelligent identification Method based on gas density parameter | |
| CN103996077B (en) | Electrical equipment fault prediction method based on multi-dimensional time sequence | |
| CN110619353B (en) | Multi-scale logging curve automatic identification method based on deep learning | |
| CN102930253B (en) | Coal and rock identification method based on image discrete multi-wavelet transform | |
| CN109918822A (en) | The shield driving parameter error calculation method and system returned based on random forest | |
| CN105069689B (en) | Based on the coal seam floor water-inrush risk evaluation method that grey correlation is combined with FDAHP | |
| CN112396320A (en) | Tunnel collapse risk assessment method based on ISM and fuzzy Bayesian network | |
| CN113673119A (en) | Dynamic and static coupling evaluation method for coal mine rock burst danger based on Bayes method | |
| Kabiesz et al. | Application of rule-based models for seismic hazard prediction in coal mines. | |
| CN110348510B (en) | Data preprocessing method based on staged characteristics of deepwater oil and gas drilling process | |
| CN116384554A (en) | Method and device for predicting mechanical drilling speed, electronic equipment and computer storage medium | |
| CN110837701A (en) | Mining area bottom plate fault water inrush quantitative evaluation method based on full-coupling analysis | |
| CN106382102A (en) | Overflow early warning method based on clustering algorithm | |
| CN115329853A (en) | Equipment parameter prediction and knowledge transfer method based on multi-source domain migration | |
| CN115438918A (en) | Method for evaluating ecological restoration effect of surface mine | |
| CN115438090A (en) | Borehole wall instability type analysis and prediction method and system based on lost block image characteristics | |
| CN112434896A (en) | Prediction method for subway shield construction risk and development trend thereof | |
| CN106530109A (en) | Oilfield development appraisal well decision method based on information value | |
| CN110991363A (en) | Method for extracting CO emission characteristics of coal mine safety monitoring system under different coal mining processes | |
| Kravis et al. | A case based system for oil and gas well design with risk assessment | |
| Wang et al. | Use of retrospective optimization for placement of oil wells under uncertainty | |
| CN113431557B (en) | Underground borehole track tracking method based on artificial intelligence | |
| CN113393334B (en) | Drilling parameter optimization recommendation method and system | |
| CN115640329A (en) | Intelligent diagnosis method for oil well fault based on multi-source data analysis | |
| Guo et al. | Prediction of gas emission by bp neural network based on wavelet analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |