CN108104807A - A kind of ocean oil and gas resource exploration system and its application method - Google Patents
A kind of ocean oil and gas resource exploration system and its application method Download PDFInfo
- Publication number
- CN108104807A CN108104807A CN201711348264.6A CN201711348264A CN108104807A CN 108104807 A CN108104807 A CN 108104807A CN 201711348264 A CN201711348264 A CN 201711348264A CN 108104807 A CN108104807 A CN 108104807A
- Authority
- CN
- China
- Prior art keywords
- oil
- gas
- mrow
- sample
- mover
- 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 71
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000005553 drilling Methods 0.000 claims abstract description 23
- 238000013528 artificial neural network Methods 0.000 claims abstract description 22
- 238000004891 communication Methods 0.000 claims abstract description 14
- 238000003860 storage Methods 0.000 claims abstract description 13
- 239000003208 petroleum Substances 0.000 claims abstract description 10
- 229930195733 hydrocarbon Natural products 0.000 claims description 53
- 150000002430 hydrocarbons Chemical class 0.000 claims description 53
- 239000004215 Carbon black (E152) Substances 0.000 claims description 52
- 238000012549 training Methods 0.000 claims description 42
- 239000002253 acid Substances 0.000 claims description 29
- 230000003301 hydrolyzing effect Effects 0.000 claims description 28
- 230000008569 process Effects 0.000 claims description 27
- 238000004422 calculation algorithm Methods 0.000 claims description 26
- 230000002159 abnormal effect Effects 0.000 claims description 23
- ATUOYWHBWRKTHZ-UHFFFAOYSA-N Propane Chemical compound CCC ATUOYWHBWRKTHZ-UHFFFAOYSA-N 0.000 claims description 20
- 230000001360 synchronised effect Effects 0.000 claims description 17
- 238000005516 engineering process Methods 0.000 claims description 14
- 238000011160 research Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 13
- OTMSDBZUPAUEDD-UHFFFAOYSA-N Ethane Chemical compound CC OTMSDBZUPAUEDD-UHFFFAOYSA-N 0.000 claims description 12
- VGGSQFUCUMXWEO-UHFFFAOYSA-N Ethene Chemical compound C=C VGGSQFUCUMXWEO-UHFFFAOYSA-N 0.000 claims description 12
- 239000005977 Ethylene Substances 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 11
- 239000013049 sediment Substances 0.000 claims description 11
- 239000001294 propane Substances 0.000 claims description 10
- 230000005484 gravity Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004557 single molecule detection Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000005183 dynamical system Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 241001074085 Scophthalmus aquosus Species 0.000 claims 1
- 238000009826 distribution Methods 0.000 abstract description 7
- 239000007789 gas Substances 0.000 description 84
- 238000011156 evaluation Methods 0.000 description 10
- UHOVQNZJYSORNB-UHFFFAOYSA-N Benzene Chemical compound C1=CC=CC=C1 UHOVQNZJYSORNB-UHFFFAOYSA-N 0.000 description 9
- YXFVVABEGXRONW-UHFFFAOYSA-N Toluene Chemical compound CC1=CC=CC=C1 YXFVVABEGXRONW-UHFFFAOYSA-N 0.000 description 9
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 6
- 150000001335 aliphatic alkanes Chemical class 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 238000013508 migration Methods 0.000 description 5
- 230000005012 migration Effects 0.000 description 5
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- 230000007935 neutral effect Effects 0.000 description 4
- 230000015572 biosynthetic process Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 239000013535 sea water Substances 0.000 description 3
- 230000004936 stimulating effect Effects 0.000 description 3
- 241000894006 Bacteria Species 0.000 description 2
- 102000004190 Enzymes Human genes 0.000 description 2
- 108090000790 Enzymes Proteins 0.000 description 2
- YNQLUTRBYVCPMQ-UHFFFAOYSA-N Ethylbenzene Chemical compound CCC1=CC=CC=C1 YNQLUTRBYVCPMQ-UHFFFAOYSA-N 0.000 description 2
- 240000004808 Saccharomyces cerevisiae Species 0.000 description 2
- 238000007621 cluster analysis Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 229910052742 iron Inorganic materials 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000003345 natural gas Substances 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 239000011435 rock Substances 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- -1 Arene compound Chemical class 0.000 description 1
- WYLIRYQDDKDHLT-UHFFFAOYSA-N CC1=CC=CC=C1C.CC1=CC=CC=C1C Chemical compound CC1=CC=CC=C1C.CC1=CC=CC=C1C WYLIRYQDDKDHLT-UHFFFAOYSA-N 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 150000001491 aromatic compounds Chemical class 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 150000001555 benzenes Chemical class 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- XTCQUBCCCSJAKJ-UHFFFAOYSA-N ethylbenzene Chemical compound CCC1=CC=CC=C1.CCC1=CC=CC=C1 XTCQUBCCCSJAKJ-UHFFFAOYSA-N 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 239000003673 groundwater Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000013011 mating Effects 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000003209 petroleum derivative Substances 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Classifications
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B49/00—Testing the nature of borehole walls; Formation testing; Methods or apparatus for obtaining samples of soil or well fluids, specially adapted to earth drilling or wells
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/62—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light
- G01N21/63—Systems in which the material investigated is excited whereby it emits light or causes a change in wavelength of the incident light optically excited
- G01N21/64—Fluorescence; Phosphorescence
- G01N21/6402—Atomic fluorescence; Laser induced fluorescence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Mining & Mineral Resources (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Geology (AREA)
- Immunology (AREA)
- General Engineering & Computer Science (AREA)
- Biochemistry (AREA)
- Environmental & Geological Engineering (AREA)
- Fluid Mechanics (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Pathology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Optics & Photonics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
Abstract
The present invention relates to a kind of ocean oil and gas resource exploration system and its application method, including drilling equipment, communication apparatus, storage device, central processing unit；The drilling equipment is connected by communication apparatus with central processing unit, and storage device is connected with central processing unit；The drilling equipment includes multiple data collectors, and each data collector preserves the data feedback of acquisition to central processing unit on a storage device；The central processing unit further includes petroleum resources judging unit, and the petroleum resources judging unit carries out Exploration of Oil And Gas using advanced exploitation method.The present invention can use for reference the learning ability of BP neural network, improve the accuracy of Exploration of Oil And Gas, it solves artificial subjective random strong in existing method, the different skimble-scamble problems of scout's deposit index, so as to improve the precision and accuracy of the calculating of exploration targets distribution, prospect is had a vast market in exploration of ocean resources field.
Description
Technical field
The present invention relates to exploration of ocean resources fields, and in particular to a kind of ocean oil and gas resource exploration system and its user
Method.
Background technology
In recent years, with progressively exhausted and science and technology the high speed development of land resource, marine resources because of it economically
Great potential and strategic critical role be increasingly valued by the people.Marine oil and gas resource in marine resources is even more each
First of the strategic reserves resource of state.
Marine oil and gas resource is different from conventional oil gas reservoir, and exploration and development is in the exploratory stage, is concentrated mainly on into and hides mould
Formula, geologic feature etc. research, effect of the geophysical techniques in ocean oil and gas resource exploration exploitation are also to be developed.Rock
Stone geophysics, geophysical log, seismic exploration technique play an important role in oil-gas exploration.
Geophysical prospecting for oil be according to the difference of subterranean strata physical property, by physical quantity, texture over the ground
It makes or nature of ground is studied, with search for oil and the geophysical exploration of natural gas.In oil exploration, for by table
Tu ﹑ deserts and seawater cover the area of no direct exposure in rock stratum, rely primarily on geophysical prospecting for oil method and understand indirectly
Geological structure and nature of ground find oil-gas reservoir.At present, geophysical prospecting for oil has become one kind that the area of coverage explores for oil
Indispensable means.Liu Shuanlian and Lu Huangsheng exist《Shale gas logging evaluation technical characterstic and evaluation method are inquired into》(well logging skill
Art, 2011,35 (2)：112-116) in a text, start with from investigation North America shale gas successful exploration developing example, in reservoir geology
On the basis of background research, the Main Differences of shale gas and conventional oil gas-bearing formation Logging Evaluation Method are analyzed.Chinese patent application
CN201410697345.7 discloses a kind of method and system of definite ocean controllable electric magnetic source exploration stimulating frequency, which leads to
Quick forward modeling algorithm is crossed, obtains normalizing the flat distribution map that abnormal amplitude changes with offset distance and stimulating frequency point, so as to
To be accurately determined optimal emission source stimulating frequency point.Chinese patent application CN201410195161.0 discloses a kind of ocean
Drilling platforms and its equipment Information Management System improve information retrieval, improve information management situation, so as to improve drilling well
System configuration type selecting efficiency, it is mating to offshore rig that there is important directive significance.However, current technology does not exist
The exploration of hydrocarbon storage amount, judge on do more researchs, this aspect Technical comparing falls behind.
As a kind of direct detection of oil and gas method, hydrocarbon geochemical exploration is for quickly drawing a circle to approve prospecting divisions, non-tectonic stress is explored
It is unique there is its with the evaluation of oil and gas bearing property of structural trap etc..The success or not of this method depends on geochemical anomaly
The correct degree of decision process, being appropriately determined prospective area needs many experience processing.Conventional hydrocarbon geochemical exploration is substantially using alkane
Basic index of the hydro carbons as oil-gas exploration.With exploration analysis means and horizontal raising and the introducing of high-tech, directly
Benzene compounds in Biological Mark Compounds such as non-alkanes substance become the index of oil and gas geochemical exploration technique predicting hydrocarbon reservoirs.Benzene
(Benzene), the oil such as toluene (Toluene), ethylbenzene (Ethyl benzene) and dimethylbenzene (Xylene) (being abbreviated as BTEX)
Arene compound is the main matter ingredient for being present in petroleum and natural gas mineral reserve, the seawater related with oil-gas reservoir, underground water
There is the exception of this 4 kinds of substances in contained dissolvable hydrocarbon.BTEX has following important feature：(1) specificity, ground
Table biological chemistry action is difficult to generate BTEX, this is conducive to exclude humanity activities influence, and BTEX is made to be kept extremely with Deep Oil-gas
Good correspondence；(2) stability is good, and BTEX is very slow by the accretion rate of bacterium, yeast, enzyme.More than 100 kinds of energy having found
It degrades only a small number of degradable aromatic compounds in the bacterium of alkane, yeast, enzyme；(3) it is easy to migrate, BTEX is readily volatilized,
It is present in the form of a vapor in oil gas cap rock, molecular volume is smaller；(4) it is abnormal easily to form top, the methane in underground water
Solubility is 35000 × 10-6, and benzene, toluene, ethylbenzene are respectively 1800 × 10-6, 550 × 10-6, 110 × 10-6, it is lower than methane
Very much, therefore the horizontal proliferation effects that are influenced by ground water movement of BTEX are not notable.Since BTEX indexs and oil gas are in the origin cause of formation
It is in close relations, it is considered as the direct indicator of hydrocarbon geochemical exploration under identical hydrogeologic condition.External research is also show that BTEX
Content (concentration) with increasing close to oil-gas reservoir, exist linearly with oil sources (distance) in oil-gas migration lithic drainage environment
Relation, it is possible thereby to preferably deduce the distribution trend of oil-gas reservoir.In gas and oil in sea, if it is shallow to analyze seabed
The content distribution of the BTEX indexs obtained in surface sediments can provide important evidence for the exploration of oil gas, so as to as marine oil
Gasify one of effective ways visited.But the content of BTEX is very low in actual acquisition sample, and volatile in earth's surface hydrological environment,
The sample of acquisition, which is difficult to ensure, to be deposited, and is difficult to obtain reliable exception with conventional analytical techniques.Tsinghua University uses RESEARCH ON LASER INDUCED SINGLE MOLECULE DETECTION
Technology realizes the high-acruracy survey to BTEX in seawater and bottom sediment so that the technology application is possibly realized.
Artificial neural network is one of research hotspot of current nonlinear method technology, because of its exclusive learning and memory and non-
Linear approximation ability has been successfully applied to many research fields, in geophysical information processing with having in petroleum exploration domain
Many successful applications.The performance of neutral net depends on network structure, neuron property and training method, neural network model
Performance Evaluation be mainly that established model is weighed for the predictive ability newly inputted, i.e., extensive energy by inspection data collection
Power.The adaptivity and fault-tolerant ability of neutral net are strong, and prediction or recognition speed are fast, but need progress attribute preferred before predicting,
And when being trained to network, it is necessary to enough representative sample datas.Attribute is preferably basis, network instruction
White silk is crucial.
Problem one：Falling behind relatively in Exploration of Oil And Gas field technology, neural network algorithm can not directly be applied wherein,
Whether can be by the optimization on algorithm and flow, so as to propose a kind of advanced technology for being suitable for Exploration of Oil And Gas.
Problem two：How the correctness of preceding method is judgedIt was trained for the neural network learning that error is met the requirements
How journey to judge its generalization ability, i.e. the precision size issue of neural network algorithm, before this to neural network algorithm
Research is not directed to.In the case where network structure, training parameter are not quite similar, how to judge same group of learning training sample with
And the precision height of training process, it is a problem for hindering Application of Neural Network.
Problem three：A variety of detection methods exist in the prior art, for the validity of its result of detection, shortage judge according to
According to.
The content of the invention
In order to overcome drawbacks described above, the present invention adopts the following technical scheme that：
A kind of ocean oil and gas resource exploration system, it is characterised in that：Including drilling equipment, communication apparatus, storage device, in
Central processor；The drilling equipment is connected by communication apparatus with central processing unit, and storage device is connected with central processing unit
It connects.
As a kind of selection, the communication apparatus includes Wireless Communication Equipment.
The drilling equipment is drill ship, bottom-supported platform, semisubmersible platform, tension leg type platform, the tower platform of lasso trick
And one kind in jack-up unit.
The drilling equipment, including hoisting system equipment, rotary system equipment, circulatory system equipment, dynamical system equipment,
Tubing processing system equipment, ancillary equipment and rig substructure equipment.The hoisting system equipment includes derrick, winch, overhead traveling crane, trip
The equipment such as vehicle, hook, brake；The rotary system equipment includes the equipment such as turntable, top drive, tap；The circulatory system is set
It is standby to include the equipment such as drilling pump, vibrating screen, desilter, blender, desander, centrifuge, deaerator, air blower；The power
System equipment includes the equipment such as power generation, distribution；The tubing processing system includes dynamic catwalk, iron driller, grab pipe machine, tubing row
The equipment such as put；The ancillary equipment includes the equipment such as preventer and its board migration device, production tree and its board migration device.
The drilling equipment includes multiple data collectors, and each data collector is by the data feedback of acquisition to centre
Device is managed, and is preserved on a storage device.
Exploration of Oil And Gas is carried out on ocean by above-mentioned ocean oil and gas resource exploration system, gathers bottom sediment sample
Product data, and lab analysis is carried out to the sediment sample of acquisition using RESEARCH ON LASER INDUCED SINGLE MOLECULE DETECTION technology.Lab analysis is except acquisition
Outside the observation data of BTEX indexs, also obtain synchronous fluorescence series, Acid Hydrolytic Hydrocarbon series, heat and release 23 kinds of routines such as hydrocarbon system row
The observation data of Geochemical Indices.Synchronous fluorescence 330nm (TB330), acid are therefrom chosen according to the cluster analysis result of 23 kinds of indexs
Solution hydrocarbon propane (SC), Acid Hydrolytic Hydrocarbon ethylene (SC2H4), heat release hydrocarbon ethane (RC2) when 4 kinds of indexs are as BTEX exception overall merits
Geochemical Characteristics parameter, and combine oil-gas geology, the data such as geophysics, carry out the integrated judgment of BTEX exceptions.The number of acquisition
According to including at least：Synchronous fluorescence 330nm (TB330), Acid Hydrolytic Hydrocarbon propane (SC3), Acid Hydrolytic Hydrocarbon ethylene (SC2H4), heat release hydrocarbon ethane
(RC2), oil-gas geology, Gravity vertical second dervative, the vertical first derivative of magnetic force.
The geophysical informations such as the Gravity vertical second dervative, the vertical first derivative of magnetic force obtain by searching for the past data
It obtains or increases related exploring equipment in ocean oil and gas resource exploration system, such as gyroscope, gravity sensor and inertial navigation
System.
Further, a kind of ocean oil and gas resource exploration system, the central processing unit include petroleum resources judging unit,
The petroleum resources judging unit carries out Exploration of Oil And Gas using following method：
Step 1：To the Oil/gas Geochemical Anomalies of trial zone, basic data is gathered.
Exploration of Oil And Gas is carried out on ocean by ocean oil and gas resource exploration system, gathers bottom sediment sample number
According to, and the sediment sample of acquisition is analyzed using RESEARCH ON LASER INDUCED SINGLE MOLECULE DETECTION technology, the observation data of BTEX indexs are obtained,
And synchronous fluorescence series, Acid Hydrolytic Hydrocarbon series, heat release the observation data of the conventional Geochemical Indices of hydrocarbon system row, therefrom choose synchronous glimmering
Light 330nm, Acid Hydrolytic Hydrocarbon propane, Acid Hydrolytic Hydrocarbon ethylene, heat release hydrocarbon ethane as Geochemical Characteristics parameter, and combine oil-gas geology, the earth
Physical data carries out the integrated judgment of BTEX exceptions；
Step 2：Establish BP network models.
Using three layers of BP network topology structures, input layer number is m=10, using single hidden layer, number of nodes s=7, output
Node layer is n, and as n=6, " desired output " is 6 kinds：[1,0,0,0,0,0]、[0,1,0,0,0,0]、[0,0,1,0,0,0]、
[0,0,0,1,0,0], [0,0,0,0,1,0], [0,0,0,0,0,1] represent oil-containing gas reservoir, oil reservoir, containing a small amount of oil reservoir, gas respectively
Hide, containing a small amount of gas reservoir, without 6 kinds of classifications of oil gas；Each neuron using the continuous function output between (0,1), selects output valve
Neuron output where the maximum is result classification；
Step 3：After establishing BP network models, learning sample is trained.
Attribute sample is not extracted to known oily sample in trial zone and at oily sample, carry out Neural Network Science
Practise training；The known exception of the trial zone 10 is chosen as mode of learning for e-learning, maximum study number is 50000 times,
System worst error is 0.00005, is learnt using serial mode during study, and using learning rate changing and during weighed value adjusting
Increase the method for momentum term to improve network convergence rate slowly and the shortcomings that be easily trapped into local minimum.
Input parameter includes F1-F7, and totally 7 parameters, meaning are followed successively by：Synchronous fluorescence 330nm (TB330), Acid Hydrolytic Hydrocarbon third
Alkane (SC3), Acid Hydrolytic Hydrocarbon ethylene (SC2H4), that heat releases hydrocarbon ethane (RC2), oil-gas geology, Gravity vertical second dervative, magnetic force is vertical
First derivative.
It is restrained after network training 30000 times, learning training process terminates, and exports result；
Result is exported as " study stage each abnormal reality output "；
Step 4：" study stage each abnormal reality output " is corresponded into " desired output ", passes through " desired output " and " class
Other result " the table of comparisons exports the category result of each sample；
Step 5：Residual error v can obtain by the difference of step 4 Learning Samples desired output and reality output, be using following formula
Variance of unit weight valuation can be acquired：
In formula, P is unit battle array I, redundant observation r=q-1=9；
Following equation calculating is recycled to acquire Naa、
Naa=AP-1AT
In formula, Q be P association's factor battle array, Q=P-1, A is 7*6 rank matrixes, and all elements are 1 in A；
Then the middle error of learning sample real output value is：
It willWith threshold value comparison, whenDuring less than threshold value, continue in next step；Otherwise, deleting causes error larger
Sample continues to learn；
Step 6：The unknown BTEX in the trial zone 10 is calculated extremely using trained BP neural network model,
Export category result.
Further, the above method is further comprising the steps of：
Step 7：Calculation procedure six exports the precision of category result；
Step 8：Actual well drilled is explored, and verifies the correctness of above-mentioned exploitation method, the main region for verifying that precision is relatively low.
Further, as a kind of selection, using a variety of learning training data in step 2 to five, error in therefrom selecting
Minimum one group is used as optimum network structure.
In the step 7, using following method computational accuracy：
(1) small increment e is applied per node layer to learning sampleij=0.001；
(2) learning sample reality output (" study stage each abnormal reality output ") y is calculated using BP algorithmij', by yij′
With applying increment eijPreceding output yijSubtracting each other can obtain corresponding to eijVariation pij, then total differential
(3) by Qyy=gTQxxG, the middle error that can obtain forecast sample are Using learning training process
Result of calculation regards forecast sample input vector as independent of observation, Qxx=P-1=I.
Compared with prior art, the present invention has the following advantages：
1. providing determination methods for Exploration of Oil And Gas, solving artificial subjective randomness in existing method, by force, difference is surveyed
The skimble-scamble problem of spy personnel's deposit index, so as to improve the precision and accuracy of the calculating of exploration targets distribution.
2. the learning ability of BP neural network can be used for reference, can obtain advantageously accounting for optimizing from known information asking
The conclusion of topic, and then improve the accuracy of Exploration of Oil And Gas.
It is nerve net 3. being estimated using the basic knowledge that measurement error is handled the precision of neutral net exploitation method
The structure design of network model provides effectively reference.
Description of the drawings
It in order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention, for those of ordinary skill in the art, without creative efforts, can be with
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is the Exploration of Oil And Gas method flow diagram of the present invention.
Specific embodiment
Hereinafter reference will be made to the drawings, and system and its method of work provided by the invention are described in detail by way of example.
It should be noted that the descriptions of the manners of these embodiments are used to facilitate the understanding of the present invention, but do not form to this hair
Bright restriction.
The terms "and/or", is only a kind of incidence relation for describing affiliated partner, and expression may have three kinds of passes
System, for example, A and/or B, can represent：Individualism A, individualism B exist simultaneously tri- kinds of situations of A and B, the terms
"/and " it is another affiliated partner relation of description, expression may have two kinds of relations, for example, A/ and B, can represent：Individually deposit
In A, two kinds of situations of individualism A and B, in addition, character "/" herein, it is that a kind of "or" is closed to typically represent forward-backward correlation object
System.
Embodiment one
The present embodiment mainly introduces the composition of the ocean oil and gas resource exploration system of the present invention.
A kind of ocean oil and gas resource exploration system, it is characterised in that：Including drilling equipment, communication apparatus, storage device, in
Central processor；The drilling equipment is connected by communication apparatus with central processing unit, and storage device is connected with central processing unit
It connects.
As a kind of selection, the communication apparatus can include Wireless Communication Equipment.
The drilling equipment is drill ship, bottom-supported platform, semisubmersible platform, tension leg type platform, the tower platform of lasso trick
And one kind in jack-up unit.
The drilling equipment, including hoisting system equipment, rotary system equipment, circulatory system equipment, dynamical system equipment,
Tubing processing system equipment, ancillary equipment and rig substructure equipment.The hoisting system equipment includes derrick, winch, overhead traveling crane, trip
The equipment such as vehicle, hook, brake；The rotary system equipment includes the equipment such as turntable, top drive, tap；The circulatory system is set
It is standby to include the equipment such as drilling pump, vibrating screen, desilter, blender, desander, centrifuge, deaerator, air blower；The power
System equipment includes the equipment such as power generation, distribution；The tubing processing system includes dynamic catwalk, iron driller, grab pipe machine, tubing row
The equipment such as put；The ancillary equipment includes the equipment such as preventer and its board migration device, production tree and its board migration device.
The drilling equipment includes multiple data collectors, and each data collector is by the data feedback of acquisition to centre
Device is managed, and is preserved on a storage device.
Exploration of Oil And Gas is carried out on ocean by above-mentioned ocean oil and gas resource exploration system, gathers bottom sediment sample
Product data, and the sediment sample of acquisition is analyzed and (lab analysis can be selected) using RESEARCH ON LASER INDUCED SINGLE MOLECULE DETECTION technology.
Analysis also obtains synchronous fluorescence series, Acid Hydrolytic Hydrocarbon series, heat and releases hydrocarbon system row in addition to the observation data for obtaining BTEX indexs
Deng the observation data of 23 kinds of conventional Geochemical Indices.Synchronous fluorescence 330nm is therefrom chosen according to the cluster analysis result of 23 kinds of indexs
(TB330), it is abnormal as BTEX to release 4 kinds of indexs such as hydrocarbon ethane (RC2) for Acid Hydrolytic Hydrocarbon propane (SC), Acid Hydrolytic Hydrocarbon ethylene (SC2H4), heat
Geochemical Characteristics parameter during overall merit, and the data such as oil-gas geology, geophysics are combined, the synthesis for carrying out BTEX exceptions is commented
Valency.The data of acquisition include at least：Synchronous fluorescence 330nm (TB330), Acid Hydrolytic Hydrocarbon propane (SC3), Acid Hydrolytic Hydrocarbon ethylene (SC2H4),
Heat releases hydrocarbon ethane (RC2), oil-gas geology, Gravity vertical second dervative, the vertical first derivative of magnetic force.
Embodiment two
The present embodiment mainly introduces the basic principle of petroleum resources judging unit of the present invention, and sea is being applied to comprising BP algorithm
Special agreement and design during foreign Exploration of Oil And Gas.
The central processing unit includes petroleum resources judging unit.
1 BP algorithm principle
Counterpropagation network (Back-Propagation Network, abbreviation BP network) has excellent None-linear approximation
Ability is most common a kind of algorithm in neutral net research.It is theoretically verified：It is hidden with deviation and at least one S types
Containing layer plus the network of a linear convergent rate layer, arbitrarily complicated nonlinear function can be approached with arbitrary precision.Therefore exist
When ocean oil and gas resource exploration studies BP algorithm, for simplicity, if hidden layer is only one layer, then in one three layers of BP networks,
First layer is input layer, and the second layer is hidden layer, and third layer is output layer.It is connected between each layer by weight coefficient, each layer of output
For next layer of input.Under normal circumstances, different layers have different neurons (node) number, and each neuron carries one
A input is 1 deviation (threshold value) b, now sets input layer number and (is determined as m by input variable dimension), node in hidden layer
For s (optional), output layer number of nodes is n (being determined by output variable dimension).It is (- 1,1) to initialize the weights of each layer and deviation
Interior random number, and set appropriate target error ε and maximum frequency of training T.Activation primitive f generally use S type functions：
F (x)=1/ (1+e-x) (1)
If given q group learning samples, input, output vector are respectively xp=[xp1,xp2,…,xpm]T, dp=[dp1,
dp2,…,dpn]T, (p=1,2 ..., q).Network is trained using learning sample group, that is, adjusts the connection weight w of network,
Make the output y of networkkAs close possible to true output dp, realize given input, the mapping relations of output.Net so after training
Network inputs for non-sample collection, can also provide suitable output.
In order to eliminate influence of the input sample order to training result, batch processing mode can be taken to change weights.Work as network
After training by global learning pattern, the gradient that each mode of learning is generated is average together, can so obtain more
Accurate gradient estimation.If mode of learning is complete, that is, includes all input pattern and target pattern, then can obtain
Estimate to accurate gradient.Then by the average value of accumulative square error compared with target error, then it is each in corrective networks
Weights and deviation.
Wherein mark p refers in particular to pth group learning sample above and below.For whole q groups learning samples, learning error function is：
The classical BP algorithm of network training is to use gradient descent method corrective networks connection weight w, i.e.,：
Wherein η is learning rate.By continuous loop iteration until error function E is sufficiently small, network training terminates.By
Learning training, the weights in BP networks between each node layer determine, it is possible to be predicted using the network, output knot at this time
Fruit is the prediction result corresponding to input.
2 BP neural network arithmetic accuracies
It is well known that the first purpose of Measurement and Data Processing is the precision of Measurement results to be calculated, Measurement results precision bag
Include two aspects：First, the available accuracy of observation；Second is that the precision of the observation value function obtained by observation through adjustment.For
For BP algorithm, it is the available accuracy of learning process Learning Samples that the first aspect is corresponding；The second aspect is corresponding to be
The precision of prediction result during prediction.
2.1 BP algorithm learning training Process Precisions
Learning process is constantly to adjust weights according to the difference between reality output and target output, due to target output
Know, be regarded as observation L, reality output is considered as adjusted valueIf residual error is v, then haveI.e.It is residual sum of squares (RSS) V by the valuation of variance of unit weightTThe PV divided by degree of freedom r of the problem (redundant observation number),
I.e.：
Since each input vector value individually to obtain, can be considered independent of observation, then observation power battle array P is in learning sample
Unit matrix I.If being solved with adjustment of condition equation method, normal equation coefficient is：
Naa=AP-1AT (6)
A is m*n rank matrixes, and all elements are 1 in A.Then real output valueAssociation's factor battle array be：
Q be P association's factor battle array, Q=P-1.Then the middle error of learning sample real output value is：
Available for the precision height for calculating BP algorithm learning training process, the value is smaller, shows learning training process essence
Degree is higher, illustrates that learning sample selection is more reasonable.
2.2 BP algorithms predict Process Precision
BP algorithm after the completion of learning training process, between each node in the form of connection weight consolidated by the contact between each layer
It decides.BP networks can be calculated pair according to the input vector of prediction process, using fixed weights and bias vector
The prediction result answered, calculating process are as follows：
Input layer：
Hidden layer：
Output layer：
If by xip, i=1,2 ..., n solves ykp, the process of k=1,2 ..., m determines then have by function g：
ykp=g (xip), i=1,2 ..., n；K=1,2 ..., m
Rule, Q are then propagated by nonlinear function association factoryy=gTQxxG, g are the coefficient to each variable after function g total differentials
Battle array, the middle error of forecast sample are：
Using the result of calculation of learning training process, if regarding forecast sample input vector as independent of observation, Qxx=
P-1=I.Thereby realize the accuracy computation to BP algorithm prediction process.
By the Specialty Design and verification of more than theoretical property, the present invention adopts the following technical scheme that：
A kind of ocean oil and gas resource exploration system, the central processing unit include petroleum resources judging unit, the oil gas
Resource judgment unit carries out Exploration of Oil And Gas using following method：
Step 1：To the Oil/gas Geochemical Anomalies of trial zone, basic data is gathered.
Exploration of Oil And Gas is carried out on ocean by ocean oil and gas resource exploration system, gathers bottom sediment sample number
According to, and the sediment sample of acquisition is analyzed using RESEARCH ON LASER INDUCED SINGLE MOLECULE DETECTION technology, the observation data of BTEX indexs are obtained,
And synchronous fluorescence series, Acid Hydrolytic Hydrocarbon series, heat release the observation data of the conventional Geochemical Indices of hydrocarbon system row, therefrom choose synchronous glimmering
Light 330nm, Acid Hydrolytic Hydrocarbon propane, Acid Hydrolytic Hydrocarbon ethylene, heat release hydrocarbon ethane as Geochemical Characteristics parameter, and combine oil-gas geology, the earth
Physical data carries out the integrated judgment of BTEX exceptions.Step 2：Establish BP network models；Using three layers of BP network topology structures,
Input layer number is m=10, and using single hidden layer, number of nodes s=7, output node layer is n, and as n=6, " desired output " is
6 kinds：[1,0,0,0,0,0]、[0,1,0,0,0,0]、[0,0,1,0,0,0]、[0,0,0,1,0,0]、[0,0,0,0,1,0]、[0,
0,0,0,0,1], represent respectively oil-containing gas reservoir, oil reservoir, containing a small amount of oil reservoir, gas reservoir, containing a small amount of gas reservoir, without 6 kinds of classifications of oil gas；Often
A neuron using the continuous function output between (0,1), when evaluation select output valve the maximum where neuron output be
As a result classification；
Step 3：After establishing BP network models, learning sample is trained.
Attribute sample is not extracted to known oily sample in trial zone and at oily sample, carry out Neural Network Science
Practise training；The known exception of the trial zone 10 is chosen as mode of learning for e-learning, maximum study number is 50000 times,
System worst error is 0.00005, is learnt using serial mode during study, and using learning rate changing and during weighed value adjusting
Increase the method for momentum term to improve network convergence rate slowly and the shortcomings that be easily trapped into local minimum.
Input parameter includes F1-F7, and totally 7 parameters, meaning are followed successively by：Synchronous fluorescence 330nm (TB330), Acid Hydrolytic Hydrocarbon third
Alkane (SC3), Acid Hydrolytic Hydrocarbon ethylene (SC2H4), that heat releases hydrocarbon ethane (RC2), oil-gas geology, Gravity vertical second dervative, magnetic force is vertical
First derivative.
It is restrained after network training 30000 times, learning training process terminates, and exports result；
Result is exported as " study stage each abnormal reality output "；
Step 4：" study stage each abnormal reality output " is corresponded into " desired output ", passes through " desired output " and " class
Other result " the table of comparisons exports the category result of each sample；
Step 5：Residual error v can obtain by the difference of step 4 Learning Samples desired output and reality output, be using following formula
Variance of unit weight valuation can be acquired：
In formula, P is unit battle array I, and redundant observation r=q-1=9, A are 7*6 rank matrixes, and all elements are 1 in A；
Following equation calculating is recycled to acquire Naa、
Naa=AP-1AT
In formula, Q be P association's factor battle array, Q=P-1；
Then the middle error of learning sample real output value is：
It willWith threshold value comparison, whenDuring less than threshold value, continue in next step；Otherwise, deleting causes error larger
Sample continues to learn；
Step 6：The unknown BTEX in the trial zone 10 is calculated extremely using trained BP neural network model,
Export category result.
Further, the above method is further comprising the steps of：
Step 7：Calculation procedure six exports the precision of category result；
Step 8：Actual well drilled is explored, and verifies the correctness of above-mentioned exploitation method, the main region for verifying that precision is relatively low.
Further, as a kind of selection, using a variety of learning training data in step 2 to five, error in therefrom selecting
Minimum one group is used as optimum network structure.
In the step 7, using following method computational accuracy：
(1) small increment e is applied per node layer to learning sampleij=0.001；
(2) learning sample reality output y is calculated using BP algorithmij', by yij' with applying increment eijPreceding output yijSubtract each other
It can obtain corresponding to eijVariation pij, then total differential
(3) by Qyy=gTQxxG, the middle error that can obtain forecast sample are Using learning training process
Result of calculation regards forecast sample input vector as independent of observation, Qxx=P-1=I.
Embodiment three
The present embodiment carries out on the basis of previous embodiment 1-2, in order to illustrate the present invention's with reference to real case
Technical solution.
First to the Oil/gas Geochemical Anomalies of Bohai Sea Gulf trial zone delineation, the change spy in binding tests area, oil-gas geology, the earth
The data such as physics carry out abnormal judgement using BP neural network algorithm.It should be noted that using neural network algorithm into
During row oil-gas exploration, attribute sample is extracted to known oily sample in trial zone and not at oily sample, carries out god
Through network learning and training, oil-gas exploration is carried out to entire trial zone with trained network.
Using three layers of BP network topology structures, input layer number is 7, and using single hidden layer, number of nodes 10 exports node layer
For 6, represent respectively oil-containing gas reservoir, oil reservoir, containing a small amount of oil reservoir, gas reservoir, containing a small amount of gas reservoir, without 6 kinds of classifications such as oil gas as a result, it is expected
Output is as shown in table 1.For each neuron using the continuous function output between (0,1), when evaluation, selects output valve the maximum institute
Neuron output for result classification.
1 BP neural network of table exploration model output meaning
Table 1 The output signification of the BP neural network evaluation
model output
| Sequence number | Desired output | Meaning |
| 1 | [1,0,0,0,0,0] | Oil-containing gas reservoir |
| 2 | [0,1,0,0,0,0] | Oil reservoir |
| 3 | [0,0,1,0,0,0] | Containing a small amount of oil reservoir |
| 4 | [0,0,0,1,0,0] | Gas reservoir |
| 5 | [0,0,0,0,1,0] | Containing a small amount of gas reservoir |
| 6 | [0,0,0,0,0,1] | Without oil gas |
After establishing BP network models, you can learning sample is trained, the known exception of the trial zone 10 is chosen and makees
Be mode of learning for e-learning, maximum study number is 50000 times, and system worst error is 0.00005, using string during study
Line mode learns, and increases the method for momentum term using learning rate changing and during weighed value adjusting to improve network convergence rate
Slowly the shortcomings that and being easily trapped into local minimum.It is restrained after network training 30000 times, learning training process terminates, and model is actually defeated
Go out to be shown in Table 2.The 1st is classified as each abnormal number for participating in study in table 2, and number EA, EB is East exception, and number WA, WB is
West is abnormal, and 2-8 is classified as score of each exception in 7 evaluating indexs, the wherein meaning of the F1-F7 such as bet of table 2 institute
Show, the 9th is classified as study stage each abnormal desired output, and 10-11 is classified as study stage each abnormal reality output and corresponding exploration
As a result.
The unknown BTEX in the trial zone 10 is explored extremely using trained BP neural network model, category result
As shown in table 3.The 1st is classified as each abnormal number for participating in prospecting in table 3, and 2-8 is classified as each exception obtaining on 7 parameter indexes
Point, the meaning of F1-F7 is classified as each abnormal reality output and corresponding category result with table 2, last 2.As can be seen that trial zone contains
There are four types of oil gas category result type is common：HYDROCARBON-BEARING REGION, You Qu, gas area, a small amount of gas area.Through being compared with actual exploration data, with reality
Border drilling well result is coincide preferably.
2 BP neural network of table prospects model learning training sample and output
Table 2 The BP neural network study and training model and output
Note：EA is abnormal for East；WA is abnormal for West；10 be no oil and gas anomaly.F1~F7 is each evaluation index, successively
For：Synchronous fluorescence 330nm (TB330), Acid Hydrolytic Hydrocarbon propane (SC3), Acid Hydrolytic Hydrocarbon ethylene (SC2H4), heat release hydrocarbon ethane (RC2), oil
Gas geology, Gravity vertical second dervative, the vertical first derivative of magnetic force.
3 BP neural network of table prospects model prediction result
Table 3 The BP neural network evaluation result of BTEX anomalies
(note：EA, EB are abnormal for East；WA, WB are abnormal for West；F1-F7 meanings are the same as table 2.)
Hereinafter, BP algorithm accuracy computation is carried out.
(1) precision of learning training process
Residual error v can obtain by the difference of 2 Learning Samples desired output of table and reality output, list can be acquired using formula (5)
Position power variance valuationP is unit battle array I, redundant observation r=q-1=9.A is 7*6 rank matrixes, and all elements are 1 in A.Again
It is calculated using formula (6), (7) and acquires Naa、It can obtain the middle error of learning sample real output value
Variance of unit weight valuation is：
The coefficient matrix of normal equation is：
Association's factor battle array of real output value is：
The middle error of learning sample real output value is：
The middle error of the actual each output valve of learning sample is respectively：
It can be seen that the precision of BP algorithm learning training process is higher, it is only 0.0014, Ke Yiman that error is maximum in each output valve
The requirement that foot is explored extremely shows that the design of BP network structures is reasonable, and learning sample selection is reasonable.
(2) precision of process is predicted
Since the function g of input x to output y during prediction do not determine its concrete form, then factor arrays g can not be direct
It acquires.For simplicity, defined from total differential, if applying small increment e per node layer to learning sampleij=0.001
(2-8 is arranged in table 4) calculates learning sample reality output y using BP algorithmij' (as shown in the row of table 4 the 9th), by yij' increase with applying
Measure eijPreceding output yijSubtracting each other can obtain corresponding to eijVariation pij, then total differentialBy Qyy=
gTQxxG, the middle error that can obtain forecast sample are Using the result of calculation of learning training process, depending on pre- test sample
This input vector be independent of observation, Qxx=P-1=I.
Table 4 applies little increment eijBP neural network exploration model prediction result afterwards
Table 4 The BP neural network comprehensive evaluation result of BTEX
anomalies(eijbe exerted)
Note：EA, EB are abnormal for East；WA, WB are abnormal for West；F1-F7 meanings are the same as table 2.
Association's factor battle array of forecast sample output valve is：
The middle error of forecast sample is：
The middle error of each output valve of forecast sample is respectively：
Error in each output valve of learning training process is compared, it can be found that the middle error of each output valve of BP algorithm forecast period
Difference is larger, and is generally more than the learning training stage, but still has reached preferable prediction effect, and wherein error is up to
0.0043, minimum 0.0010；Error is up to 0.0014 in learning training the output of process value, and minimum is only 0.0002.Into
When row BP network structures design, the middle error of learning training and forecast period under heterogeneous networks structure can be calculated respectively,
The precision height of each algorithm is calculated, so as to provide theoretical foundation for the selection of optimum network structure.
As described above, it can preferably realize the present invention.For a person skilled in the art, do not departing from the present invention's
Principle and spirit in the case of these embodiments are changed, change, replace, integrating and modification still fall within the present invention protection
In the range of.The part of specified otherwise or restriction is not carried out in the present invention, is implemented using the prior art.
Claims (9)
1. a kind of ocean oil and gas resource exploration system, it is characterised in that：Including drilling equipment, communication apparatus, storage device, center
Processor；The drilling equipment is connected by communication apparatus with central processing unit, and storage device is connected with central processing unit.
2. a kind of ocean oil and gas resource exploration system according to claim 1, it is characterised in that：The communication apparatus includes
Wireless Communication Equipment.
3. a kind of ocean oil and gas resource exploration system according to claim 2, it is characterised in that：The drilling equipment is brill
One kind in well ship, bottom-supported platform, semisubmersible platform, tension leg type platform, the tower platform of lasso trick and jack-up unit.
4. a kind of ocean oil and gas resource exploration system according to claim 3, it is characterised in that：The drilling equipment includes
Hoisting system equipment, rotary system equipment, circulatory system equipment, dynamical system equipment, tubing processing system equipment, ancillary equipment
With rig substructure equipment.
5. a kind of ocean oil and gas resource exploration system according to claim 3 or 4, it is characterised in that：The drilling equipment
Including multiple data collectors, the data feedback of acquisition to central processing unit, and is stored in storage and set by each data collector
It is standby upper.
6. a kind of ocean oil and gas resource exploration system according to claim 5, it is characterised in that：The data collector is adopted
The data of collection include synchronous fluorescence 330nm, Acid Hydrolytic Hydrocarbon propane, Acid Hydrolytic Hydrocarbon ethylene, heat and release hydrocarbon ethane, oil-gas geology, Gravity vertical
The vertical first derivative of second dervative, magnetic force.
A kind of 7. method of usage right requirement 1-6 any one of them ocean oil and gas resource exploration systems, it is characterised in that：Institute
Stating central processing unit includes petroleum resources judging unit, and the petroleum resources judging unit carries out oil gas money using following method
It explores in source：
Step 1：To the Oil/gas Geochemical Anomalies of trial zone, basic data is gathered；By ocean oil and gas resource exploration system in ocean
Upper carry out Exploration of Oil And Gas is gathered bottom sediment sample data, and is sunk using RESEARCH ON LASER INDUCED SINGLE MOLECULE DETECTION technology to acquisition
Product object sample is analyzed, and the observation data and synchronous fluorescence series, Acid Hydrolytic Hydrocarbon series, heat for obtaining BTEX indexs release hydrocarbon system
The observation data of the conventional Geochemical Indices of row therefrom choose synchronous fluorescence 330nm, Acid Hydrolytic Hydrocarbon propane, Acid Hydrolytic Hydrocarbon ethylene, heat and release hydrocarbon
Ethane combines oil-gas geology, geophysical information as Geochemical Characteristics parameter, carries out the integrated judgment of BTEX exceptions；
Step 2：Establish BP network models；Using three layers of BP network topology structures, input layer number is q=10, using single hidden
Layer, number of nodes s=7, output node layer is n, and as n=6, " desired output " is 6 kinds：[1,0,0,0,0,0]、[0,1,0,0,
0,0], [0,0,1,0,0,0], [0,0,0,1,0,0], [0,0,0,0,1,0], [0,0,0,0,0,1], represent oily respectively
Tibetan, oil reservoir, containing a small amount of oil reservoir, gas reservoir, containing a small amount of gas reservoir, without 6 kinds of classifications of oil gas；Each neuron is using between (0,1)
Continuous function exports, and the neuron output where selecting output valve the maximum is result classification；
Step 3：After establishing BP network models, learning sample is trained；To known oily sample in trial zone and
Attribute sample is not extracted at oily sample, carries out neural network learning training；Choose the known exception conduct of the trial zone 10
Mode of learning supplies e-learning, and maximum study number is 50000 times, and system worst error is 0.00005, using serial during study
Mode learns, and the method for increase momentum term is slow to improve network convergence rate using learning rate changing and during weighed value adjusting
The shortcomings that with local minimum is easily trapped into；Input parameter includes F1-F7, and totally 7 parameters, meaning are followed successively by：Synchronous fluorescence
330nm, Acid Hydrolytic Hydrocarbon propane, Acid Hydrolytic Hydrocarbon ethylene, heat release hydrocarbon ethane, oil-gas geology, Gravity vertical second dervative, the vertical single order of magnetic force
Derivative；It is restrained after network training 30000 times, learning training process terminates, and exports result；Result is exported as " the study stage is different
Normal reality output "；
Step 4：" study stage each abnormal reality output " is corresponded into " desired output ", passes through " desired output " and " classification knot
Fruit " the table of comparisons exports the category result of each sample；
Step 5：Residual error v can obtain by the difference of step 4 Learning Samples desired output and reality output, can be asked using following formula
Obtain variance of unit weight valuation：
<mrow>
<msubsup>
<mover>
<mi>&delta;</mi>
<mo>^</mo>
</mover>
<mn>0</mn>
<mn>2</mn>
</msubsup>
<mo>=</mo>
<mfrac>
<mrow>
<msup>
<mi>V</mi>
<mi>T</mi>
</msup>
<mi>P</mi>
<mi>V</mi>
</mrow>
<mi>r</mi>
</mfrac>
</mrow>
In formula, P is unit battle array I, redundant observation r=q-1=9；
Following equation calculating is recycled to acquire Naa、
Naa=AP-1AT
<mrow>
<msub>
<mi>Q</mi>
<mrow>
<mover>
<mi>L</mi>
<mo>^</mo>
</mover>
<mover>
<mi>L</mi>
<mo>^</mo>
</mover>
</mrow>
</msub>
<mo>=</mo>
<msub>
<mi>Q</mi>
<mrow>
<mi>L</mi>
<mi>L</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>Q</mi>
<mrow>
<mi>V</mi>
<mi>V</mi>
</mrow>
</msub>
<mo>=</mo>
<mi>Q</mi>
<mo>-</mo>
<msup>
<mi>QA</mi>
<mi>T</mi>
</msup>
<msubsup>
<mi>N</mi>
<mrow>
<mi>a</mi>
<mi>a</mi>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mi>A</mi>
<mi>Q</mi>
</mrow>
In formula, Q be P association's factor battle array, Q=P-1, A is 7*6 rank matrixes, and all elements are 1 in A；
Then the middle error of learning sample real output value is：
<mrow>
<msub>
<mover>
<mi>&delta;</mi>
<mo>^</mo>
</mover>
<mover>
<mi>L</mi>
<mo>^</mo>
</mover>
</msub>
<mo>=</mo>
<msub>
<mover>
<mi>&delta;</mi>
<mo>^</mo>
</mover>
<mn>0</mn>
</msub>
<msqrt>
<msub>
<mi>Q</mi>
<mrow>
<mover>
<mi>L</mi>
<mo>^</mo>
</mover>
<mover>
<mi>L</mi>
<mo>^</mo>
</mover>
</mrow>
</msub>
</msqrt>
</mrow>
It willWith threshold value comparison, whenDuring less than threshold value, continue in next step；Otherwise, the sample for causing error larger is deleted,
Continue to learn；
Step 6：The unknown BTEX in the trial zone 10 is calculated extremely using trained BP neural network model, is exported
Category result.
8. the method according to claim 7 using ocean oil and gas resource exploration system, it is characterised in that：The method is also
Comprise the following steps：
Step 7：Calculation procedure six exports the precision of category result；
Step 8：Actual well drilled is explored, and verifies the correctness of above-mentioned exploitation method, the relatively low region of emphasis verification precision.
9. the method according to claim 8 using ocean oil and gas resource exploration system, it is characterised in that：The step 7
In, using following method computational accuracy：
(1) small increment e is applied per node layer to learning sampleij=0.001；
(2) learning sample " study stage each abnormal reality output " y ' is calculated using BP algorithmij, by y 'ijWith applying increment eijBefore
Output yijSubtracting each other can obtain corresponding to eijVariation pij, then total differential
(3) by Qyy=gTQxxG, the middle error that can obtain forecast sample areUsing the calculating of learning training process
As a result, forecast sample input vector is regarded as independent of observation, Qxx=P-1=I.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711348264.6A CN108104807A (en) | 2017-12-16 | 2017-12-16 | A kind of ocean oil and gas resource exploration system and its application method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711348264.6A CN108104807A (en) | 2017-12-16 | 2017-12-16 | A kind of ocean oil and gas resource exploration system and its application method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108104807A true CN108104807A (en) | 2018-06-01 |
Family
ID=62216168
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711348264.6A Pending CN108104807A (en) | 2017-12-16 | 2017-12-16 | A kind of ocean oil and gas resource exploration system and its application method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108104807A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109298410A (en) * | 2018-11-02 | 2019-02-01 | 北京遥测技术研究所 | A kind of marine oil spill detecting laser radar |
| CN109857782A (en) * | 2019-01-28 | 2019-06-07 | 中国石油大学胜利学院 | A kind of Monitor of Logging Data Processing System |
| CN110107231A (en) * | 2019-06-06 | 2019-08-09 | 吉林大学 | A kind of automatic cat road of adaptive neural network and its control method |
| CN110796155A (en) * | 2019-07-12 | 2020-02-14 | 大港油田集团有限责任公司 | Crude oil water content data analysis method based on clustering algorithm |
| CN117514086A (en) * | 2023-12-29 | 2024-02-06 | 山东成林石油工程技术有限公司 | Semi-submersible centrifugal jet device and use method thereof |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20040099092A (en) * | 2002-03-26 | 2004-11-26 | 카운슬 오브 사이언티픽 앤드 인더스트리얼 리서치 | Improved performance of artificial neural network models in the presence of instrumental noise and measurement errors |
| CN101514624A (en) * | 2008-03-05 | 2009-08-26 | 中国科学院自动化研究所 | Embedded oilwell parameter real-time acquisition system |
| CN103970101A (en) * | 2014-05-09 | 2014-08-06 | 中国石油大学(华东) | Marine drilling platform and equipment information management system thereof |
| US9734220B2 (en) * | 2012-12-04 | 2017-08-15 | Planet Os Inc. | Spatio-temporal data processing systems and methods |
-
2017
- 2017-12-16 CN CN201711348264.6A patent/CN108104807A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20040099092A (en) * | 2002-03-26 | 2004-11-26 | 카운슬 오브 사이언티픽 앤드 인더스트리얼 리서치 | Improved performance of artificial neural network models in the presence of instrumental noise and measurement errors |
| CN101514624A (en) * | 2008-03-05 | 2009-08-26 | 中国科学院自动化研究所 | Embedded oilwell parameter real-time acquisition system |
| US9734220B2 (en) * | 2012-12-04 | 2017-08-15 | Planet Os Inc. | Spatio-temporal data processing systems and methods |
| CN103970101A (en) * | 2014-05-09 | 2014-08-06 | 中国石油大学(华东) | Marine drilling platform and equipment information management system thereof |
Non-Patent Citations (5)
| Title |
|---|
| YU, L (YU, LEAN); LAI, KK (LAI, KIN KEUNG); WANG, S (WANG, SHOUY: "《Multistage RBF neural network ensemble learning》", 《NEUROCOMPUTING》 * |
| 武汉大学测绘学院测量平差学科组: "《误差理论与测量平差基础》", 31 January 2003 * |
| 白洪伟，王华，李进，穆星，吴满意: "《大型建筑物沉降变形的动态预测方法研究》", 《河北北方学院学报(自然科学版)》 * |
| 赵健，刘展，张勇: "《激光单分子BTEX 异常的人工神经网络评价技术》", 《海洋科学》 * |
| 赵健，刘展: "《基于灵敏度分析的海洋油气资源BP神经网络预测模型的优化》", 《海洋科学》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109298410A (en) * | 2018-11-02 | 2019-02-01 | 北京遥测技术研究所 | A kind of marine oil spill detecting laser radar |
| CN109857782A (en) * | 2019-01-28 | 2019-06-07 | 中国石油大学胜利学院 | A kind of Monitor of Logging Data Processing System |
| CN110107231A (en) * | 2019-06-06 | 2019-08-09 | 吉林大学 | A kind of automatic cat road of adaptive neural network and its control method |
| CN110107231B (en) * | 2019-06-06 | 2024-03-29 | 吉林大学 | Self-adaptive neural network automatic catwalk and control method thereof |
| CN110796155A (en) * | 2019-07-12 | 2020-02-14 | 大港油田集团有限责任公司 | Crude oil water content data analysis method based on clustering algorithm |
| CN117514086A (en) * | 2023-12-29 | 2024-02-06 | 山东成林石油工程技术有限公司 | Semi-submersible centrifugal jet device and use method thereof |
| CN117514086B (en) * | 2023-12-29 | 2024-03-08 | 山东成林石油工程技术有限公司 | Semi-submersible centrifugal jet device and use method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Thanh et al. | Integrated workflow in 3D geological model construction for evaluation of CO2 storage capacity of a fractured basement reservoir in Cuu Long Basin, Vietnam | |
| CN108104807A (en) | A kind of ocean oil and gas resource exploration system and its application method | |
| Wang et al. | An improved neural network for TOC, S1 and S2 estimation based on conventional well logs | |
| Ahmadi et al. | Evolving simple-to-use method to determine water–oil relative permeability in petroleum reservoirs | |
| Elkatatny | A self-adaptive artificial neural network technique to predict total organic carbon (TOC) based on well logs | |
| Krahenbuhl et al. | Time-lapse gravity: A numerical demonstration using robust inversion and joint interpretation of 4D surface and borehole data | |
| CN104632202B (en) | Method and device for determining dry clay three-porosity logging parameter values | |
| Kalogirou et al. | Artificial neural networks for the generation of geothermal maps of ground temperature at various depths by considering land configuration | |
| CN106529667A (en) | Logging facies identification and analysis method based on fuzzy depth learning in big data environment | |
| Shi et al. | A new method for rock brittleness evaluation in tight oil formation from conventional logs and petrophysical data | |
| Ahangari et al. | On the prediction of geochemical parameters (TOC, S1 and S2) by considering well log parameters using ANFIS and LSSVM strategies | |
| Spulber et al. | Methane emission from natural gas seeps and mud volcanoes in Transylvania (Romania) | |
| Al-qaness et al. | An optimized neuro-fuzzy system using advance nature-inspired Aquila and Salp swarm algorithms for smart predictive residual and solubility carbon trapping efficiency in underground storage formations | |
| CN110056346A (en) | A kind of three-dimensional original water saturation analogy method of the oil reservoir based on Long-term change trend function | |
| Alalimi et al. | Developing the efficiency-modeling framework to explore the potential of CO2 storage capacity of S3 reservoir, Tahe oilfield, China | |
| CN106570262A (en) | Reservoir configuration structure description method | |
| Ren et al. | Spatial prediction of oil and gas distribution using tree augmented Bayesian network | |
| Gao et al. | Identification of low-resistivity-low-contrast pay zones in the feature space with a multi-layer perceptron based on conventional well log data | |
| Shope et al. | Geothermal resource assessment: A detailed approach to low-grade resources in the states of New York and Pennsylvania | |
| Alizadeh et al. | Using stacked generalization ensemble method to estimate shear wave velocity based on downhole seismic data: A case study of Sarab-e-Zahab, Iran | |
| Hunt et al. | Geophysical investigations of the Wairakei Field | |
| Wang et al. | Numerical simulation of hydrocarbon migration in tight reservoir based on Artificial Immune Ant Colony Algorithm: A case of the Chang 81 reservoir of the Triassic Yanchang Formation in the Huaqing area, Ordos Basin, China | |
| Susantoro et al. | Applying random forest to oil and gas exploration in Central Sumatra basin Indonesia based on surface and subsurface data | |
| Singh et al. | Application of NeuroFuzzy pattern recognition method in borehole geophysics | |
| Hao et al. | AFSs-RBF neural network for predicting earthquake-induced liquefaction of light loam |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180601 |
|
| RJ01 | Rejection of invention patent application after publication |