CN108520348A - A kind of ecological index prediction technique based on mangrove forest ecological big data - Google Patents
A kind of ecological index prediction technique based on mangrove forest ecological big data Download PDFInfo
- Publication number
- CN108520348A CN108520348A CN201810281241.6A CN201810281241A CN108520348A CN 108520348 A CN108520348 A CN 108520348A CN 201810281241 A CN201810281241 A CN 201810281241A CN 108520348 A CN108520348 A CN 108520348A
- Authority
- CN
- China
- Prior art keywords
- layer
- mangrove forest
- forest ecological
- data
- formula
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 240000002044 Rhizophora apiculata Species 0.000 title claims abstract 34
- 238000012549 training Methods 0.000 claims abstract description 42
- 230000036541 health Effects 0.000 claims abstract description 12
- 230000008569 process Effects 0.000 claims abstract description 11
- 238000013480 data collection Methods 0.000 claims abstract description 5
- 230000005183 environmental health Effects 0.000 claims abstract description 5
- 230000006870 function Effects 0.000 claims description 30
- 239000003550 marker Substances 0.000 claims description 19
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 16
- 241000607479 Yersinia pestis Species 0.000 claims description 8
- 230000007704 transition Effects 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 7
- 230000004048 modification Effects 0.000 claims description 6
- 238000012986 modification Methods 0.000 claims description 6
- 239000002689 soil Substances 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 5
- 238000004321 preservation Methods 0.000 claims description 5
- 241000251468 Actinopterygii Species 0.000 claims description 4
- 241000254173 Coleoptera Species 0.000 claims description 4
- 241000195493 Cryptophyta Species 0.000 claims description 4
- 241000238631 Hexapoda Species 0.000 claims description 4
- 241000255777 Lepidoptera Species 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 4
- 241000143060 Americamysis bahia Species 0.000 claims description 3
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 claims description 3
- 229910002651 NO3 Inorganic materials 0.000 claims description 3
- NHNBFGGVMKEFGY-UHFFFAOYSA-N Nitrate Chemical compound [O-][N+]([O-])=O NHNBFGGVMKEFGY-UHFFFAOYSA-N 0.000 claims description 3
- IOVCWXUNBOPUCH-UHFFFAOYSA-M Nitrite anion Chemical compound [O-]N=O IOVCWXUNBOPUCH-UHFFFAOYSA-M 0.000 claims description 3
- UCKMPCXJQFINFW-UHFFFAOYSA-N Sulphide Chemical compound [S-2] UCKMPCXJQFINFW-UHFFFAOYSA-N 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- XKMRRTOUMJRJIA-UHFFFAOYSA-N ammonia nh3 Chemical compound N.N XKMRRTOUMJRJIA-UHFFFAOYSA-N 0.000 claims description 3
- 229910052799 carbon Inorganic materials 0.000 claims description 3
- 229930002875 chlorophyll Natural products 0.000 claims description 3
- 235000019804 chlorophyll Nutrition 0.000 claims description 3
- ATNHDLDRLWWWCB-AENOIHSZSA-M chlorophyll a Chemical compound C1([C@@H](C(=O)OC)C(=O)C2=C3C)=C2N2C3=CC(C(CC)=C3C)=[N+]4C3=CC3=C(C=C)C(C)=C5N3[Mg-2]42[N+]2=C1[C@@H](CCC(=O)OC\C=C(/C)CCC[C@H](C)CCC[C@H](C)CCCC(C)C)[C@H](C)C2=C5 ATNHDLDRLWWWCB-AENOIHSZSA-M 0.000 claims description 3
- 230000000813 microbial effect Effects 0.000 claims description 3
- 230000003862 health status Effects 0.000 abstract description 5
- 240000003793 Rhizophora mangle Species 0.000 description 68
- 241001269238 Data Species 0.000 description 4
- 241000196324 Embryophyta Species 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000017105 transposition Effects 0.000 description 3
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 229910052760 oxygen Inorganic materials 0.000 description 2
- 239000001301 oxygen Substances 0.000 description 2
- 230000009182 swimming Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 208000028399 Critical Illness Diseases 0.000 description 1
- 241000238557 Decapoda Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 239000013535 sea water Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 239000004577 thatch Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/14—Fourier, Walsh or analogous domain transformations, e.g. Laplace, Hilbert, Karhunen-Loeve, transforms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Mathematical Analysis (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Pure & Applied Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Computational Mathematics (AREA)
- Educational Administration (AREA)
- Mathematical Optimization (AREA)
- Algebra (AREA)
- Life Sciences & Earth Sciences (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Agronomy & Crop Science (AREA)
- Animal Husbandry (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of ecological index prediction techniques based on mangrove forest ecological big data, mainly include the following steps that：1) mangrove forest ecological protection zone is determined.2) mangrove forest ecological achievement data collection is established.3) data set T and data set Y are pre-processed.4) linearization process is carried out to data set T, obtains linear training set S.5) mangrove forest ecological index prediction model is established.6) using the mangrove forest ecological index prediction model, the linear training set S and mangrove forest ecological health assessment grade decision table, mangrove future environmental health situation in the mangrove forest ecological protection zone is predicted.The important ecological index that the present invention is had been predicted that using these, in conjunction with mangrove forest ecological health assessment grade decision table, final realize carries out Accurate Prediction to mangrove forest ecological health status.
Description
Technical field
It is specifically a kind of based on mangrove forest ecological big data the present invention relates to data mining technology and deep learning method
Ecological index prediction technique.
Background technology
Mangrove is the woody plant community, xylium for being grown on tropical and subtropical zone seashore intertidal zone, periodically being submerged by seawater,
It is the Complex Ecological Systems for having both land and ocean characteristic, the important ecologically criticalarea of seashore, is improving bay, wave resistance bank protection, net
Change pollution and swamp protection diversity etc. and plays irreplaceable function.ALONG GUANGXI COAST is the important mangrove in China's Mainland
Woods distributed area, the long 1490km of continent water front, the mangrove area whole nation second are that unit water front length mangrove distribution area is maximum
Provinces and regions.Guangxi is equipped with Mangrove Nature Reserves 3 (the mountain pass Nature Reserve of Hepu of Guangxi, Guangxi Fangcheng Port
Beilun Estuary Nature Reserve and Guangxi Qinzhou thatch tail sea Provincial Nature Reserve).According to 8 years north of past
The report of logical sequence estuary ecological protection zone Environmental Status, will appreciate that mangrove forest ecological index include water environment, depositional environment, life of swimming
Object, phytoplankton, intertidal organism, Mangrove Communities and birds.Wherein planktonic organism, phytoplankton, intertidal organism, mangrove
Mangrove forest ecological health most can directly be reflected with birds index in woods group.The analysis found that having very between these indexs
It closely contacts, for example water environment has prodigious influence to the existence of zooplankter, phytoplankton and intertidal organism, deposits
Environment has prodigious influence to the existence of zooplankter, phytoplankton.But sensing can only collect part with image technique
Mangrove forest ecological index, such as water environment, depositional environment pest, pest.Therefore, sensing with image technique be difficult to accurate evaluation and
Predict mangrove forest ecological health status.
Invention content
Present invention aim to address problems of the prior art.
To realize the present invention purpose and the technical solution adopted is that a kind of such, life based on mangrove forest ecological big data
State index prediction technique, mainly includes the following steps that：
1) mangrove forest ecological protection zone is determined.
2) mangrove forest ecological achievement data collection is established.
The mangrove forest ecological protection zone in α is utilized into sensing and the collected mangrove forest ecological index number of image technique
It is believed that breath I is as data set T.The mangrove forest ecological marker data information I includes mainly water quality information, deposit pH, soil
PH and soil grades index.
The water quality information mainly include water Inversion phenomenon, pH, chlorophyll, ammonia nitrogen, nitrate, nitrite, Phos,
Petroleum-type and COD.
The deposit mainly includes machine carbon and sulfide.
In formula,For mangrove forest ecological marker data information I.
The mangrove forest ecological marker data information II that the ecological preservation area in α is manually acquired is as data set Y.Institute
It includes fish information, shrimps information, microbial information, algae information, chemistry to state mangrove forest ecological marker data information II mainly
Oxygen demand COD information, bod BOD information, mangrove insect pest information, coenotype, zoobenthos information, dynamic plant of swimming
Object information, intertidal organism information, Lepidoptera category information and coleoptera category information.
In formula,For mangrove forest ecological marker data information II.
3) data set T and data set Y are pre-processed.The pretreatment includes mainly denoising and normalization.Count evidence
Integrate after T is normalized as data set R.
4) linearization process is carried out to data set T, obtains linear training set S.
Using geo-nuclear tracin4 algorithm, the key step that linearization process is carried out to data set T is as follows：
4.1) the note input space is X, and Hilbert space H, the new input space is Z.
Definition is from input space X to the mapping function of Hilbert space H, i.e.,：
φ(x):X→H。 (3)
In formula, x is the data in input space X.X is the input space.H is Hilbert space.
Kernel function K (x, z) is defined, i.e.,：
K (x, z)=φ (x) φ (z), x ∈ X, z ∈ Z. (4)
In formula, x is the data in input space X.Z is the data in new input space Z.X is the input space.Z is new defeated
Enter space.
4.2) by input space X described in the data inputting in data set T, as the data x in the input space X.
4.3) mapping function φ (x) is utilized, data set T is mapped in Hilbert space H, to linearly be trained
Collect S.
Linear training set S is as follows：
In formula,For the mangrove forest ecological marker data information I after linearisation.
5) mangrove forest ecological index prediction model is established.
The key step for establishing mangrove forest ecological index prediction model is as follows：
5.1) the reversed decision making algorithm that the design number of plies is B.The reversed decision making algorithm mainly include input layer, transition zone and
Output layer.The number of plies of the input layer is 1.The number of plies of the transition zone is B-1.The number of plies of the output layer is 1.
Using linear training set S and data set Y as the training set D of the mangrove forest ecological index prediction model.Training set D
As follows：
D={ (S1, Y1), (S2, Y2)…(Si, Yi)…(Sn, Yn)}。 (6)
In formula, SiFor the i-th row of linear training set S.YiFor the i-th row of data set Y.
5.2) it is inputted linear training set S as feature.
5.3) remember that i-th layer of weight between jth layer is ξij.Jth layer is biased to δj。
Wherein, ξij∈[-1,0]。δj∈[-1,0]。
5.4) the input I of jth layerjAs follows：
In formula, ξijFor the weight between i-th layer and jth layer.δjFor the biasing of jth layer.ζiFor i-th layer of output.I is layer
Sequence number.J is level serial number.
The output ζ of jth layerjAs follows：
In formula, θ is activation primitive.IjFor the input of jth layer.J is level serial number.
Pass through the input I of jth layerjWith the output ζ of jth layerjForward direction transmits, and the training result for obtaining first layer is R1。
5.5) by R1As next layer of input layer.Define error function Ej.Error function EjAs follows：
In formula, ξjkFor the weight between jth layer and kth layer.ζjFor the output of jth layer.J is level serial number.K is layer order
Number.
Weight changes amount Δ ξijAs follows：
Δξij=(λ) Ejζi。 (10)
In formula, ζiFor i-th layer of output.I is level serial number.J is level serial number.(λ) is learning coefficient.
Bias knots modification Δ δjAs follows：
Δδj=(λ) Ej。 (11)
In formula, j is level serial number.(λ) is learning coefficient.EjFor error function.
Step 3 and step 4 are repeated, the 2nd layer of training result R is obtained2。
5.6) step 3 is repeated to step 5B-1 times, finally obtains the output result Y of pth layer1。
5.7) for output layer, error function A is definedj。
Aj=ζj(I-ζj)(1-ζj)。 (12)
In formula, ζjFor the output of jth layer.J is level serial number.I is the input of output layer.
In negative gradient direction, with minimum AjWith minimum δiBased on be worth, utilize formula 13 and formula 14 to adjust weight ξijWith
Bias δj。
ξ′ij=ξij+Δξij。 (13)
In formula, ξijFor the weight between i-th layer and jth layer.ΔξijFor weight changes amount.ξ′ijFor i-th layer after adjustment and
Weight between jth layer.
δ′j=δj+Δδj。 (14)
In formula, δjFor the biasing of jth layer.ΔδjTo bias knots modification.δ′jIt is inclined between i-th layer and jth layer after adjustment
It sets.
5.8) judge the weight ξ ' between i-th layer and jth layer after adjustingijWhether threshold epsilon is less than1.I-th layer is judged after adjusting
Biasing δ ' between jth layerjWhether threshold epsilon is less than2。
If 5.9) ξ 'ij≥ε1, then by ξ 'ijValue as ξijValue, recurring formula 13 retrieve after adjustment i-th layer and the
Weight ξ ' between j layersij, and repeat step 5.8.
If δ 'j≥ε2, then by δ 'jValue as δjValue, recurring formula 14, retrieve after adjustment i-th layer and jth layer it
Between biasing δ 'j, and repeat step 5.8.
If ξ 'ij＜ ε1With δ 'j＜ ε2It sets up simultaneously, then training terminates, and obtains mangrove forest ecological index prediction model.
6) the mangrove forest ecological index prediction model, the linear training set S and mangrove forest ecological health assessment are utilized
Grade decision table predicts mangrove future environmental health situation in the mangrove forest ecological protection zone.
The effect of the present invention is unquestionable.The present invention by geo-nuclear tracin4 algorithm by being used for mangrove forest ecological big data
Pretreatment, is handled in a manner of unsupervised, realizes the linearisation of data so that the data characterization of composition is easier to prediction model
Deep learning algorithm understand, significantly improve the forecasting accuracy of reversed decision making algorithm.Meanwhile the present invention establishes one instead
To decision model, water environment, depositional environment, the pest data measured by sensing of existing period, image, Accurate Prediction is to next
Planktonic organism in period, phytoplankton, intertidal organism, Mangrove Communities data information, to be sensed, image skill
Being associated between the index that art can acquire and the important indicator that cannot be acquired.The important ecology that the present invention is had been predicted that using these
Index, in conjunction with mangrove forest ecological health assessment grade decision table, final realization carries out mangrove forest ecological health status accurate pre-
It surveys.
Description of the drawings
Fig. 1 is mangrove forest ecological health status Predicting Technique overview flow chart；
Fig. 2 is reversed decision making algorithm structure chart；
Fig. 3 is that geo-nuclear tracin4 linearizes example；
Fig. 4 is that prediction data is compared with truthful data；
Fig. 5 is the ecological index prediction result comparison diagram using geo-nuclear tracin4 algorithm.
Specific implementation mode
With reference to embodiment, the invention will be further described, but should not be construed the above-mentioned subject area of the present invention only
It is limited to following embodiments.Without departing from the idea case in the present invention described above, according to ordinary skill knowledge and used
With means, various replacements and change are made, should all include within the scope of the present invention.
Embodiment 1：
Referring to Fig. 1 to Fig. 5, a kind of ecological index prediction technique based on mangrove forest ecological big data includes mainly following
Step：
1) mangrove forest ecological protection zone is determined.The present embodiment selects Beilun Estuary ecological preservation area.
2) mangrove forest ecological achievement data collection is established.
The mangrove forest ecological protection zone in first 7 years is utilized into sensing and the collected mangrove forest ecological index of image technique
Data information I is as data set T.The mangrove forest ecological marker data information I includes mainly water quality information, deposit pH, soil
Earth pH and soil grades index.
The water quality information mainly include water Inversion phenomenon, pH, chlorophyll, ammonia nitrogen, nitrate, nitrite, Phos,
Petroleum-type and COD.
It most can directly reflect some indexs such as fish, shrimp, microorganism and the algae of mangrove forest ecological health status, COD, BOD
And other trace meter atoms, mangrove insect pest, coenotype, zoobenthos, plankton and intertidal organism, Lepidoptera
Class, coleoptera category information are difficult to acquire by sensing and image technique, it is necessary to it is analyzed by manually acquiring, it is time-consuming and laborious.
Therefore establishing being associated between the index and the important indicator that cannot be acquired that a model is sensed, image technique can acquire is
It is necessary.
The deposit mainly includes machine carbon and sulfide.
In formula,For mangrove forest ecological marker data information I.Provide that the data with same alike result are in training set
One row are, it is specified that different samples is a line in training set.
The mangrove forest ecological marker data information II that the ecological preservation area in first 7 years is manually acquired is as data set Y.
The mangrove forest ecological marker data information II includes mainly fish information, shrimps information, microbial information, algae information, changes
Learn oxygen demand COD information, bod BOD information, mangrove insect pest information, coenotype, zoobenthos information, swim it is dynamic
Plant information, intertidal organism information, Lepidoptera category information and coleoptera category information.
In formula,For mangrove forest ecological marker data information II.Provide that the data with same alike result are in training set
One row are, it is specified that different samples is a line in training set.Yi∈RL。
3) data set T and data set Y are pre-processed.The pretreatment includes mainly denoising and normalization.To having
The unitized processing of the unit of the data correlation column information that unit requires, m indicate unitized treated the dimension of unit.Count evidence
Integrate after T is normalized as data set R.
Min-max is standardized:At data set C1Wherein one
Row, max and min are respectively as maximum value and minimum value in the row, TiFor the data column information after normalization, to data set X
In respectively arrange and be normalized respectively, so that the data of each row is mapped within [0,1], normalization terminates to obtain data set R.
4) it is nonlinear due to having some values inside the data of acquisition, there are certain couplings, cannot be trained, therefore
It needs to carry out data prediction work before training pattern, to obtain training sample.Data set T is carried out at linearisation
Reason, obtains linear training set S, to realize the decoupling of nonlinear data collection.
The key step that linearization process is carried out to data set T is as follows：
4.1) the note input space is X, and Hilbert space H, the new input space is Z.
If the data in input space X and new input space Z are 2-D data.
The input spaceData x=(x in input space X1,x2)T.Wherein, x1For the first columns of input space X
According to.x2For the second column datas of input space X.T is transposition.
The new input spaceData z=(z in new input space Z1,z2)T.Wherein, z1For new input space Z
One column data.z2For new the second column datas of input space Z.T is transposition.
Definition is from input space X to the mapping function of Hilbert space H, i.e.,：
φ(x):X→H。 (4)
In formula, x is the data in input space X.X is the input space.H is Hilbert space.
Kernel function K (x, z) is defined, i.e.,：
In formula, x is the data in input space X.Z is the data in new input space Z.X is the input space.Z is new defeated
Enter space.x1For the first column datas of input space X.x2For the second column datas of input space X.T is transposition.
The inner product operation that m is tieed up higher dimensional space by kernel function is converted into the kernel function calculating that n ties up the low-dimensional input space, to skilful
Solves the problems such as " dimension disaster " that is calculated in high-dimensional feature space wonderfully
The introducing of kernel function avoids " dimension disaster ", substantially reduces calculation amount.And the dimension n of the input space is to core letter
Matrix number is without influence, and therefore, higher-dimension input can be effectively treated in Kernel-Based Methods.
After introducing kernel function, when calculating without knowing the form and parameter of non-linear transform function Φ.
The variation of the form and parameter of kernel function can implicitly change the mapping from the input space to feature space, and then right
The property of feature space has an impact, and finally changes the performance of various Kernel-Based Methods.
Kernel-Based Methods can be combined with different algorithms, form a variety of different methods based on kernel function technology,
And this two-part design can be carried out individually, and can be that different applications selects different kernel function and algorithm
As shown in figure 3, by transformation z=φ (x), former spaceIt is transformed to new spaceIn former space
Point is transformed to the point in new space accordingly.Oval w in former space1(x(1))2+w2(x(2))2+ b=0 is transformed in new space
Straight line w1z(1)+w2z(2)+ b=0.
In space after the conversion, straight line w1z(1)+w2z(2)+ b=0 can correctly divide the positive and negative example points after transformation
It opens, the nonlinear problem in former space has reformed into the linear problem in new space.
4.2) by input space X described in the data inputting in data set T, as the data x in the input space X.
4.3) mapping function φ (x) is utilized, data set T is mapped in Hilbert space H, to linearly be trained
Collect S.
Linear training set S is as follows：
In formula,For the mangrove forest ecological marker data information I after linearisation.Si∈Rd。
5) mangrove forest ecological index prediction model is established.
The key step for establishing mangrove forest ecological index prediction model is as follows：
5.1) the reversed decision making algorithm that the design number of plies is B.The number of plies of reversed decision making algorithm is the transition zone number of plies plus output
It counts layer by layer.The main thought of reversed decision making algorithm is that learning process is divided into two stages：
First stage is the forward-propagating process of information flow.When input information is at " input layer → transition zone → output layer "
When propagating and handle in path, every layer of real output value is calculated.
Second stage is the back-propagation process of error.When output layer fails to obtain desired output, reality output
Difference (i.e. error) between desired output is then propagated in the path of " output layer → hidden layer → input layer ".Specifically,
It is exactly the level error distribution to each layer, to each error signal secondary layer by layer of acquisition and using these error signals as amendment
The foundation of each connection weight.The application repeatedly of the two processes, finally so that error is minimum.
The reversed decision making algorithm includes mainly input layer, transition zone and output layer.The number of plies of the input layer is 1.Institute
The number of plies for stating transition zone is B-1.The number of plies of the output layer is 1.
Using linear training set S and data set Y as the training set D of the mangrove forest ecological index prediction model.Training set D
As follows：
D={ (S1, Y1), (S2, Y2)…(Si, Yi)…(Sn, Yn)}。 (7)
In formula, SiFor the i-th row of linear training set S.YiFor the i-th row of data set Y.
5.2) it is inputted linear training set S as feature.
5.3) remember that i-th layer of weight between jth layer is ξij.Jth layer is biased to δj。
Wherein, ξij∈[-1,0]。δj∈[-1,0]。
5.4) the input I of jth layerjAs follows：
In formula, ξijFor the weight between i-th layer and jth layer.δjFor the biasing of jth layer.ζiFor i-th layer of output.I is layer
Sequence number.J is level serial number.
The output ζ of jth layerjAs follows：
In formula, θ is activation primitive.IjFor the input of jth layer.J is level serial number.
Pass through the input I of jth layerjWith the output ζ of jth layerjForward direction transmits, and the training result for obtaining first layer is R1。
5.5) by R1As next layer of input layer.Define error function Ej.Error function EjAs follows：
In formula, ξjkFor the weight between jth layer and kth layer.ζjFor the output of jth layer.J is level serial number.K is layer order
Number.
Weight changes amount Δ ξijAs follows：
Δξij=(λ) Ejζi。 (11)
In formula, ζiFor i-th layer of output.I is level serial number.J is level serial number.(λ) is learning coefficient, is used for algorithm tune
Ginseng.
Bias knots modification Δ δjAs follows：
Δδj=(λ) Ej。 (12)
In formula, j is level serial number.(λ) is learning coefficient.EjFor error function.
Step 3 and step 4 are repeated, the 2nd layer of training result R is obtained2。
5.6) step 3 is repeated to step 5B-1 times, finally obtains the output result Y of pth layer1。
5.7) for output layer, error function A is definedj。
Aj=ζj(I-ζj)(1-ζj)。 (13)
In formula, ζjFor the output of jth layer.J is level serial number.I is the input of output layer.
In negative gradient direction, with minimum AjWith minimum δiBased on be worth, utilize formula 13 and formula 14 to adjust weight ξijWith
Bias δj.Pass through minimum AjWeight W is calculatedij, then carry out weight update.
ξ′ij=ξij+Δξij。 (14)
In formula, ξijFor the weight between i-th layer and jth layer.ΔξijFor weight changes amount.ξ′ijFor i-th layer after adjustment and
Weight between jth layer.
δ′j=δj+Δδj。 (15)
In formula, δjFor the biasing of jth layer.ΔδjTo bias knots modification.δ′jIt is inclined between i-th layer and jth layer after adjustment
It sets.
5.8) judge the weight ξ ' between i-th layer and jth layer after adjustingijWhether threshold epsilon is less than1.I-th layer is judged after adjusting
Biasing δ ' between jth layerjWhether threshold epsilon is less than2。
If 5.9) ξ 'ij≥ε1, then by ξ 'ijValue as ξijValue, recurring formula 13 retrieve after adjustment i-th layer and the
Weight ξ ' between j layersij, and repeat step 5.8.
If δ 'j≥ε2, then by δ 'jValue as δjValue, recurring formula 14, retrieve after adjustment i-th layer and jth layer it
Between biasing δ 'j, and repeat step 5.8.
If ξ 'ij＜ ε1With δ 'j＜ ε2It sets up simultaneously, then training terminates, and obtains mangrove forest ecological index prediction model.
By the Beilun Estuary of in July, 2016 ecological preservation area sensing and the collected part mangrove forest ecological of image technique
The output of model is compared, from prediction by marker data information as test set T with the relevant parameter that the August manually acquires
As a result it can be seen that it is higher using the accuracy of reversed decision making algorithm prediction, to examine the model.As shown in attached drawing 4,5.
Meanwhile Fig. 5 also illustrates to carry out the forecasting accuracy that linearization process significantly improves reversed decision making algorithm using geo-nuclear tracin4 algorithm.
6) the mangrove forest ecological index prediction model, the linear training set S and mangrove forest ecological health assessment are utilized
Grade decision table predicts mangrove future environmental health situation in the mangrove forest ecological protection zone.Mangrove forest ecological
Health assessment grade decision table is classified mangrove forest ecological health, for example health is 1 grade, and inferior health is 2 grades, unhealthy to be
3 grades, critically ill is 4 grades.
The present invention constructs an accurate mangrove forest ecological index prediction model by reversed decision making algorithm, in prediction mould
A geo-nuclear tracin4 algorithm is added before type, handles mangrove forest ecological big data so that the data characterization finally constituted is easier to give birth to
The deep learning algorithm of state prediction model understands, significantly improves the prediction accuracy of reversed decision prediction model.
Meanwhile the mangrove forest ecological index prediction model can be by the existing period senses, image measure water environment, heavy
Product environment, pest data, planktonic organism, phytoplankton, intertidal organism, Mangrove Communities in Accurate Prediction to subsequent time period
Data information, being associated with to be sensed, between the index that image technique can acquire and the important indicator that cannot be acquired.Profit
The important ecological index being had been predicted that with these, it is final to realize to mangrove in conjunction with mangrove forest ecological health assessment grade decision table
The target of woods environmental health condition predicting.
Claims (4)
1. a kind of ecological index prediction technique based on mangrove forest ecological big data, which is characterized in that mainly include the following steps that：
1) mangrove forest ecological protection zone is determined；
1) the mangrove forest ecological achievement data collection is established；
The mangrove forest ecological protection zone in α is believed using sensing and the collected mangrove forest ecological achievement data of image technique
I is ceased as data set T；
In formula,For mangrove forest ecological marker data information I；
The mangrove forest ecological marker data information II that the ecological preservation area in α is manually acquired is as data set Y；
In formula,For mangrove forest ecological marker data information II；
3) data set T and data set Y are pre-processed；The pretreatment includes mainly denoising and normalization；Note data set T returns
It is data set R after one change；
4) linearization process is carried out to data set T, obtains linear training set S；
5) mangrove forest ecological index prediction model is established；
6) the mangrove forest ecological index prediction model, the linear training set S and mangrove forest ecological health assessment grade are utilized
Decision table predicts mangrove future environmental health situation in the mangrove forest ecological protection zone.
2. a kind of ecological index prediction technique based on mangrove forest ecological big data according to claim 1, feature exist
In：The mangrove forest ecological marker data information I includes mainly water quality information, deposit pH, soil pH and soil grades index；
The water quality information includes mainly water Inversion phenomenon, pH, chlorophyll, ammonia nitrogen, nitrate, nitrite, Phos, oil
Class and COD；
The deposit mainly includes machine carbon and sulfide；
The mangrove forest ecological marker data information II includes mainly fish information, shrimps information, microbial information, algae letter
Breath, COD COD information, bod BOD information, mangrove insect pest information, coenotype, zoobenthos information,
Plankton information, intertidal organism information, Lepidoptera category information and coleoptera category information.
3. a kind of ecological index prediction technique based on mangrove forest ecological big data according to claim 1, feature exist
In the key step for carrying out linearization process to data set T is as follows：
1) the note input space is X, and Hilbert space H, the new input space is Z；
Definition is from input space X to the mapping function of Hilbert space H, i.e.,：
φ(x):X→H； (3)
In formula, x is the data in input space X；X is the input space；H is Hilbert space；
Kernel function K (x, z) is defined, i.e.,：
K (x, z)=φ (x) φ (z), x ∈ X, z ∈ Z； (4)
In formula, x is the data in input space X；Z is the data in new input space Z；X is the input space；Z is that new input is empty
Between；
2) by input space X described in the data inputting in data set T, as the data x in the input space X；
3) mapping function φ (x) is utilized, data set T is mapped in Hilbert space H, to obtain linear training set S；
Linear training set S is as follows：
In formula,For the mangrove forest ecological marker data information I after linearisation.
4. a kind of ecological index prediction technique based on mangrove forest ecological big data according to claim 1, feature exist
In the key step for establishing mangrove forest ecological index prediction model is as follows：
1) the reversed decision making algorithm that the design number of plies is B；The reversed decision making algorithm includes mainly input layer, transition zone and output
Layer；The number of plies of the input layer is 1；The number of plies of the transition zone is B-1；The number of plies of the output layer is 1；
Using linear training set S and data set Y as the training set D of the mangrove forest ecological index prediction model；Training set D is as follows
It is shown：
D={ (S1, Y1), (S2, Y2)…(Si, Yi)…(Sn, Yn)}； (6)
In formula, SiFor the i-th row of linear training set S；YiFor the i-th row of data set Y；
2) it is inputted linear training set S as feature；
3) remember i-th layer to the weight between jth layer be ξij；Jth layer is biased to δj；
Wherein, ξij∈[-1,0]；δj∈[-1,0]；
4) the input I of jth layerjAs follows：
In formula, ξijFor the weight between i-th layer and jth layer；δjFor the biasing of jth layer；ζiFor i-th layer of output；I is layer order
Number；J is level serial number；
The output ζ of jth layerjAs follows：
In formula, θ is activation primitive；IjFor the input of jth layer；J is level serial number；
Pass through the input I of jth layerjWith the output ζ of jth layerjForward direction transmits, and the training result for obtaining first layer is R1；
5) by R1As next layer of input layer；Define error function Ej；Error function EjAs follows：
In formula, ξjkFor the weight between jth layer and kth layer；ζjFor the output of jth layer；J is level serial number；K is level serial number；
Weight changes amount Δ ξijAs follows：
Δξij=(λ) Ejζi； (10)
In formula, ζiFor i-th layer of output；I is level serial number；J is level serial number；(λ) is learning coefficient；
Bias knots modification Δ δjAs follows：
Δδj=(λ) Ej； (11)
In formula, j is level serial number；(λ) is learning coefficient；EjFor error function；
Step 3 and step 4 are repeated, the 2nd layer of training result R is obtained2；
6) step 3 is repeated to step 5B-1 times, obtains B layers of output result Y1；
7) for output layer, error function A is definedj；
Aj=ζj(I-ζj)(1-ζj)； (12)
In formula, ζjFor the output of jth layer；J is level serial number；I is the input of output layer；
In negative gradient direction, with minimum AjWith minimum δiBased on be worth, utilize formula 13 and formula 14 to adjust weight ξijAnd biasing
δj；
ξ′ij=ξij+Δξij； (13)
In formula, ξijFor the weight between i-th layer and jth layer；ΔξijFor weight changes amount；ξ′ijFor i-th layer after adjustment and jth layer
Between weight；
δ′j=δj+Δδj； (14)
In formula, δjFor the biasing of jth layer；ΔδjTo bias knots modification；δ′jFor the biasing after adjustment between i-th layer and jth layer；
8) judge the weight ξ ' between i-th layer and jth layer after adjustingijWhether threshold epsilon is less than1；Judge i-th layer and jth after adjusting
Biasing δ ' between layerjWhether threshold epsilon is less than2；
If 9) ξ 'ij≥ε1, then by ξ 'ijValue as ξijValue, recurring formula 13 retrieve after adjusting between i-th layer and jth layer
Weight ξ 'ij, and repeat step 8；
If δ 'j≥ε2, then by δ 'jValue as δjValue, recurring formula 14 retrieve inclined between i-th layer and jth layer after adjusting
Set δ 'j, and repeat step 8；
If ξ 'ij＜ ε1With δ 'j＜ ε2It sets up simultaneously, then training terminates, and obtains mangrove forest ecological index prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810281241.6A CN108520348B (en) | 2018-04-02 | 2018-04-02 | Ecological index prediction method based on mangrove forest ecological big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810281241.6A CN108520348B (en) | 2018-04-02 | 2018-04-02 | Ecological index prediction method based on mangrove forest ecological big data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108520348A true CN108520348A (en) | 2018-09-11 |
| CN108520348B CN108520348B (en) | 2021-10-26 |
Family
ID=63431014
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810281241.6A Active CN108520348B (en) | 2018-04-02 | 2018-04-02 | Ecological index prediction method based on mangrove forest ecological big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108520348B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738385A (en) * | 2019-07-31 | 2020-01-31 | 暨南大学 | island ecosystem current situation assessment and development prediction method based on Ecospace model |
| CN112861837A (en) * | 2020-12-30 | 2021-05-28 | 北京大学深圳研究生院 | Unmanned aerial vehicle-based mangrove forest ecological information intelligent extraction method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008206421A (en) * | 2007-02-23 | 2008-09-11 | Kansai Electric Power Co Inc:The | Mangrove growth forecasting system, mangrove afforestation right land judging system, mangrove growth forecasting method, and mangrove afforestation right land judging method |
| CN104268181A (en) * | 2014-09-16 | 2015-01-07 | 国家海洋信息中心 | Marine biological survey data rapid check method and device. |
| CN106777879A (en) * | 2016-11-16 | 2017-05-31 | 北京大学深圳研究生院 | Mangrove water environment health Evaluation method based on Biological indicators |
-
2018
- 2018-04-02 CN CN201810281241.6A patent/CN108520348B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008206421A (en) * | 2007-02-23 | 2008-09-11 | Kansai Electric Power Co Inc:The | Mangrove growth forecasting system, mangrove afforestation right land judging system, mangrove growth forecasting method, and mangrove afforestation right land judging method |
| CN104268181A (en) * | 2014-09-16 | 2015-01-07 | 国家海洋信息中心 | Marine biological survey data rapid check method and device. |
| CN106777879A (en) * | 2016-11-16 | 2017-05-31 | 北京大学深圳研究生院 | Mangrove water environment health Evaluation method based on Biological indicators |
Non-Patent Citations (1)
| Title |
|---|
| MINGHUI OU 等: "Evaluation of water quality in mangrove ecosystem for the", 《8TH INTERNATIONAL CONFERENCE ON ENVIRONMENT SCIENCE AND ENGINEERING》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738385A (en) * | 2019-07-31 | 2020-01-31 | 暨南大学 | island ecosystem current situation assessment and development prediction method based on Ecospace model |
| CN110738385B (en) * | 2019-07-31 | 2023-05-02 | 暨南大学 | Island ecological system current situation assessment and development prediction method based on Ecospace model |
| CN112861837A (en) * | 2020-12-30 | 2021-05-28 | 北京大学深圳研究生院 | Unmanned aerial vehicle-based mangrove forest ecological information intelligent extraction method |
| CN112861837B (en) * | 2020-12-30 | 2022-09-06 | 北京大学深圳研究生院 | Unmanned aerial vehicle-based mangrove forest ecological information intelligent extraction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108520348B (en) | 2021-10-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Horrigan et al. | Response of stream macroinvertebrates to changes in salinity and the development of a salinity index | |
| Burd et al. | The development and application of analytical methods in benthic marine infaunal studies | |
| Li et al. | Predicting the effect of land use and climate change on stream macroinvertebrates based on the linkage between structural equation modeling and bayesian network | |
| Schöniger et al. | A statistical concept to assess the uncertainty in b ayesian model weights and its impact on model ranking | |
| Liu et al. | Prediction of dissolved oxygen content in aquaculture of Hyriopsis cumingii using Elman neural network | |
| CN105447248B (en) | The acute reference prediction method of seawater based on metal quantitative structure activity relationship | |
| Gutiérrez-Estrada et al. | A heuristic approach to predicting water beetle diversity in temporary and fluctuating waters | |
| CN108520348A (en) | A kind of ecological index prediction technique based on mangrove forest ecological big data | |
| CN114399073A (en) | Ocean surface temperature field prediction method based on deep learning | |
| CN114242156A (en) | Real-time prediction method and system for relative abundance of pathogenic vibrios on marine micro-plastic | |
| CN115718746A (en) | Rice field methane emission prediction method based on machine learning | |
| CN108510191A (en) | A kind of mangrove forest ecological health assessment method based on stacking noise reduction autocoding algorithm | |
| Wang et al. | Functional diversity of macrofaunal assemblages as indicators to assess heavy metal pollution in the Bohai Sea, China | |
| CN111401444A (en) | Method and device for predicting origin of red wine, computer equipment and storage medium | |
| Gowen et al. | The development of UK pelagic (Plankton) indicators and targets for the MSFD | |
| Seyed Jalali et al. | Prediction of soil cation exchange capacity using different soil parameters by intelligent models | |
| CN114757433B (en) | Method for rapidly identifying relative risk of drinking water source antibiotic resistance | |
| Zhou et al. | Predicting and explaining karst spring dissolved oxygen using interpretable deep learning approach | |
| Hayati et al. | Predicting habitat suitability for the endangered species Hedysarum criniferum Boiss. using a fuzzy model in the semi-arid zone in Iran | |
| CN113221651A (en) | Seafloor sediment classification method using acoustic propagation data and unsupervised machine learning | |
| CN113657021A (en) | Marine measurement period evaluation method based on BP neural network | |
| Jia et al. | A cluster-stacking-based approach to forecasting seasonal chlorophyll-a concentration in coastal waters | |
| Zhao et al. | Quantifying the response of aquatic biodiversity to variations in river hydrology and water quality in a healthy water ecology pilot city, China | |
| CN112446168A (en) | Effluent BOD concentration soft measurement method based on MIC and RBFNN | |
| Zheng et al. | Simulation of phytoplankton biomass in Quanzhou Bay using a back propagation network model and sensitivity analysis for environmental variables |
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 |