CN109840873A - A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning - Google Patents
A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning Download PDFInfo
- Publication number
- CN109840873A CN109840873A CN201910106649.4A CN201910106649A CN109840873A CN 109840873 A CN109840873 A CN 109840873A CN 201910106649 A CN201910106649 A CN 201910106649A CN 109840873 A CN109840873 A CN 109840873A
- Authority
- CN
- China
- Prior art keywords
- data
- basin
- decision tree
- hydro
- machine learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning, comprising the following steps: step 1, data preparation；Step 2, acquisition have data Watershed Hydrologic Models parameter set；Step 3, drainage characteristics attribute Principle component extraction；Step 4, building machine learning decision Tree algorithms data set；Step 5, building decision tree, generate optimum decision tree；Step 6, acquisition basin optimal classification rule regular according to decision tree classification；Step 7, unga(u)ged basin parameter obtain.The invention proposes a kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning, not only objectivity is strong but also fast and accurately can be that unga(u)ged basin finds Choosing Hydrological Reference Basin, obtains model parameter by machine learning algorithm for the parameter region method, provides support for Cross Some Region Without Data hydrologic forecast.
Description
Technical field
The present invention relates to Cross Some Region Without Data Hydrological Forecasting Technique field more particularly to a kind of non-avaibles based on machine learning
Regional Hydro-Model Parameter Calibration Technology fields method, main application decision tree machine learning algorithm solve non-avaible Hydro-Model Parameter Calibration Technology
Identification problem, for work such as Cross Some Region Without Data hydrological simulation and flood forecastings.
Background technique
Cross Some Region Without Data hydrologic forecast (PUB) is the difficult point and hot issue of international hydrology research.Cross Some Region Without Data is logical
Often lack actual measurement hydrological data (or flow data), Hydro-Model Parameter Calibration Technology calibration can not be carried out, significantly limit hydrological model
Application.Therefore to Cross Some Region Without Data Hydro-Model Parameter Calibration Technology compartmentalization research to solution Cross Some Region Without Data hydrologic forecast, raising water
Literary forecast precision is of great significance.
Common parameter region method includes grafting, the Return Law and interpolation method.Parameter grafting is according to River Basin Hydrology
Similitude handle has data watershed parameters (Choosing Hydrological Reference Basin) to divert from one use to another to unga(u)ged basin (target basin), including closely located method and category
Property analogue method.But both parameter region methods carry out hydrological similarity and judge no quantitative criterion, and subjectivity is larger, and parameter is moved
It is big with not knowing.The Return Law is to establish the statistical relationship of basin attribute factor Yu each parameter, is a kind of lump type empirical method, suddenly
It has omited in basin and has produced mechanism of converging, isolated the globality of runoff, it is poor using result.Interpolation method generally refers to sky
Interpolation method, this method needs biggish Choosing Hydrological Reference Basin sample, and cannot react the globality of runoff, and practical application is difficult
Degree is big, uncertain strong.
Therefore, that there are subjectivities is strong, cannot react the defects of runoff mechanism for all kinds of parameter region methods at present,
It can not effectively be applied in the identification of Cross Some Region Without Data parameter, cannot be provided effectively for the hydrological simulation and forecast of unga(u)ged basin
Help.
Summary of the invention
The invention proposes a kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning, using machine
Device learning algorithm establishes Hydro-Model Parameter Calibration Technology and drainage characteristics relationship, can quick and precisely identify the stream of studying and comparing of unga(u)ged basin
Domain carries out Cross Some Region Without Data parameter region.
In order to solve above-mentioned technical problem, present invention employs following scheme:
A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning, comprising the following steps:
Step 1, data preparation: including the Underlying Surface Data and long Series of Water vigour of style image data in data basin, non-avaible
The Underlying Surface Data in basin；
Step 2, acquisition have a data Watershed Hydrologic Models parameter set: using long Series of Water vigour of style image data to hydrological model into
Row calibration and verifying obtain；
Step 3, drainage characteristics attribute Principle component extraction: to every class drainage characteristics, establishing Factor Analysis Model, with it is main at
Divide analytic approach solving model, obtains the principal component score value of each drainage characteristics type；
Step 4, building machine learning decision Tree algorithms data set: data in data set include respectively have data basin it is main at
Get the deterministic coefficient of the distance between a point difference, basin, watershed parameters cross validation, wherein deterministic coefficient is as decision
The dependent variable of building is set, distance is as independent variable between principal component scores difference and basin；
Step 5, building decision tree, generate optimum decision tree；
Step 6, acquisition basin optimal classification rule regular according to decision tree classification；
Step 7, unga(u)ged basin parameter obtain: calculating unga(u)ged basin and respectively have data basin principal component difference, without money
Stream domain and respectively there is distance between data basin, has data basin as optimal according to step 6 gained optimal classification Rules Filtering
Choosing Hydrological Reference Basin obtains Hydro-Model Parameter Calibration Technology.
Further, in step 1, the Underlying Surface Data includes constructing topography and geomorphology data, the soil that hydrological model needs
Ground utilizes the climate characteristic data and river basin authorities data of data, soil types data and basin.
Further, in step 1, the long Series of Water vigour of style image data for having data basin refers to rainfall in 10 years or more
Diameter flow data, data, which should include that large, medium and small water is various, represents the time；Flood play, humid region are no less than 50, punja
Many 25 of area.
Further, in step 2, with 2/3 session data calibration Hydro-Model Parameter Calibration Technology before long Series of Water vigour of style image data,
1/3 data verification Hydro-Model Parameter Calibration Technology afterwards.
In step 2, Hydro-Model Parameter Calibration Technology calibration and verifying precision select deterministic coefficient to evaluate, it is desirable that calibration and verifying
Certainty Coefficient Mean be all larger than 0.7.
Further, in step 3, the drainage characteristics in drainage characteristics attribute Principle component extraction includes that watershed unit landforms are special
Sign, soil types and soil texture feature, land use and vegetative coverage feature, the climate characteristic in basin and river basin authorities are special
Sign.
Each data type and it includes attribute factor be shown in Table 1:
Table 1
Principal component analysis is carried out respectively to five kinds of drainage characteristics types listed by upper table in step 3, extract it is all types of it is main at
Divide information.
Further, in step 4, machine learning data set includes training set, verifying collection and test set.
Further, in step 5, complete decision tree is constructed by training set, beta pruning is carried out to decision tree by verifying collection
And optimization, test set advanced optimize decision tree, finally obtain optimum decision tree.
Constructing complete decision tree by training set in step 5 is based on dependent variable square error minimum come branch.
The pruning method of decision tree is CCP (Cost Complexity Pruning) the beta pruning method pair afterwards used in step 5
Complete decision tree carries out beta pruning, as index carrys out beta pruning formation optimum decision tree so that error rate is minimum.
Further, in step 6, decision tree classification rule refers to the branching rule of decision tree, the quantity of the rule with certainly
The leaf node number of plan tree is identical；Basin optimal classification rule refers to the maximum decision tree classification rule of dependent variable (deterministic coefficient)
Then.
The present invention is based on the Cross Some Region Without Data parameter region methods of machine learning to have the advantages that
(1) present invention proposes Cross Some Region Without Data Hydro-Model Parameter Calibration Technology compartmentalization calculation on the basis of machine learning algorithm
Method, the algorithm can choose Choosing Hydrological Reference Basin automatically for unga(u)ged basin, obtain according to unga(u)ged basin and having data drainage characteristics
Unga(u)ged basin Hydro-Model Parameter Calibration Technology, is greatly saved cost of labor, improves computational efficiency.
(2) present invention is to be classified automatically with machine learning algorithm to the similar basin of the hydrology, and Choosing Hydrological Reference Basin is chosen
Objectivity is strong, and parameter region precision is high, provides scientific support for the identification of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology.
With reference to the accompanying drawing and specific embodiment is described in further detail invention.
Detailed description of the invention
A kind of Fig. 1: Cross Some Region Without Data region parameter method flow block diagram based on machine learning of the present invention；
Fig. 2: machine learning Principal Component Analysis process blocks schematic diagram of the present invention；
Fig. 3: machine learning decision Tree algorithms process blocks schematic diagram of the present invention.
Specific embodiment
Below with reference to Fig. 1, Fig. 2 and Fig. 3, the present invention will be further described:
According to Fig. 1, the present invention is built upon on the basis of machine learning algorithm, and specific embodiment has:
1, data preparation:
The data of preparation include Underlying Surface Data and Hydro-Model Parameter Calibration Technology rate needed for the building of data Watershed Hydrologic Models
Fixed required long Series of Water vigour of style image data；Underlying Surface Data needed for the building of unga(u)ged basin hydrological model.Wherein underlying surface
Data include watershed unit relief data, soil types and soil texture data, land use and vegetative coverage data, are in addition gone back
Including climate characteristic data and river basin authorities data needed for drainage characteristics analysis.Meet the length of Hydro-Model Parameter Calibration Technology calibration needs
The time limit of Series of Water vigour of style image data is generally not less than 10 years, should include the various data for representing the time of high, medium and low water, wherein
Humid region flood play is no less than 50, and arid area is no less than 25.
2, data Watershed Hydrologic Models parameter determines:
Data Watershed Hydrologic Models parameter is determined according to the basin hydrometeorological data calibration of long series.With hydrology gas
As before data system 2/3 play flood data to hydrological model carry out calibration, later 1/3 play flood data is to hydrological model
It is verified.In calibration and verification process, using deterministic coefficient maximum as objective function, deterministic coefficient (DC) calculation formula
Such as formula (1):
In formula, ycIt (i) is the forecasting runoff of i-th of step-length, m3/s；y0It (i) is the measured discharge of i-th of step-length, m3/s；For measured discharge mean value, m3/s。
3, basin attributive character Principle component extraction:
Basin attributive character is primarily referred to as that the attribute of basin water regime can be reacted, including in step 1 topography and geomorphology,
Soil types and the soil texture, land use and vegetative coverage and the climate characteristic and basin structural character in basin.Basin belongs to
Property feature Principle component extraction include Factor Analysis Model building, Factor Analysis Model solve and etc. (Fig. 2).
Factor Analysis Model is constructed using each basin attributive character factor as test variable.If assuming, test variable matrix is
X=(X1, X2... Xp)T, mean vector E (X)=0, covariance matrix Cov (X)=∑；If E (X)=μ, enables X*=X- μ, i.e.,
There is E (X*)=0.
F=(F1, F2…Fm)TIt is unobservable random variable, mean vector E (F)=0, covariance matrix Cov (F)
=Im.ε=(ε1, ε2…εp)TIt is also unobservable random variable, mean vector E (ε) -0, covariance matrixIt is diagonal matrix, and Cov (F, ε)=0, i.e. ε are uncorrelated to F.Factor Analysis Model
It can be indicated by formula (2).
Matrix form is X=AF+s
Wherein, A=(aij)p×mWith F=(F1, F2…Fm)TFor common factor, ε=(ε1, ε2…εp)TFor error.A is the factor
Load, aijIndicate i-th of variable XiIn j-th of common factor FjOn load.
With Principal Component Analysis, Factor Analysis Model is solved.In Factor Analysis Model, the covariance of stochastic variable X
For ∑, correlation matrix R, because X is normalized matrix, therefore
R=∑=AAT+Dε
If the characteristic value of R is λ1≥λ2≥…≥λp> 0, corresponding unit orthogonal eigenvectors are e1, e2... ep, enable U
=(e1, e2... ep), it can obtain:
As common factor FiNumber be p when, error 0 can obtain R=AAT, therefore it is desirableI.e. jth column factor loading be j-th of principal component coefficient ejWithAchievement.As m≤p
When, m column structure requirement loading matrix before takingReach by the accumulative variance contribution ratio of common factor
Percentage (taking 80%) choose m.
For 5 class drainage characteristics attribute factors, principal component analysis is carried out according to above method, extracts 5 class drainage characteristics
Principal component.
4, decision Tree algorithms data set is constructed:
Machine learning algorithm data set includes training set, verifying collection and test set.A kind of machine learning as supervised
Algorithm, the data set of decision tree include independent variable and dependent variable.Independent variable is poor by the principal component of basin each in step 3 type-collection
The distance between value and basin composition, dependent variable is the deterministic coefficient for having data watershed parameters cross validation to obtain.In addition it needs
It is noted that the distance between basin refers to the geometric distance between the central point of basin.It is constituted from independent variable and dependent variable
50% group of data is randomly selected in data set, as verifying collection, to be left 25% as training set, 25% and be used as test set.
5, decision tree is constructed, optimum decision tree is generated:
The algorithm flow of decision tree is as shown in Figure 3.Complete decision tree is constructed with training set first.What step 4 constructed
In data set, independent variable and dependent variable are continuous variable, carry out branch to sample with square error approach (MSE), building is complete
Decision tree.The calculation formula of square error approach such as formula (3):
In formula, C and D are two samples after branch, nkFor the value of k-th of sample dependent variable in C branch,For C branch
In all samples mean of dependent variable；niFor the value of i-th of sample dependent variable in D branch,It is equal for all dependent variables in D branch
Value；Err is square error, takes the smallest point of err as cut-off and constructs decision tree.
To prevent decision tree over-fitting, beta pruning is carried out to decision tree with verifying collection.Beta pruning uses CCP (Cost
Complexity Pruning) beta pruning method, tree T is first obtained according to heuristic rule during beta pruningmaxParametric family { T1,
T2,······,TL, then estimate according to the error rate of tree, Best tree T is selected in parametric familyi.For given tree Ti
Subtree Tit, the balanced growth for defining the error rate of each leaf node isIn formula, epTo be leaf by subtree beta pruning
Wrong classification rate after node, euTo set TiWrong classification rate,For subtree TitThe number of the leaf node of subordinate.Tree
Ti+1It is by setting TiCutting has the minimum limb increased to obtain the error rate of each leaf node, that is, cutting those has minimum δ value
Node selects the standard of Best tree minimum for error rate.
The decision tree after beta pruning is further tested with test set, generates optimum decision tree.
6, according to decision tree classification rule, basin optimal classification rule is obtained；
Decision tree classification rule refers to that step 5 obtains class condition of the optimum decision tree from root node to leaf node.From because
In many leaf nodes of variable, the maximum leaf node of dependent variable value is found out, using the classifying rules of this leaf node as optimal classification
Rule.
7, unga(u)ged basin parameter obtains:
Calculating respectively has between the principal component difference in data basin and basin in unga(u)ged basin and decision tree building data set
Distance, be that Choosing Hydrological Reference Basin is chosen in unga(u)ged basin by the optimal classification rule of step 6.Same unga(u)ged basin passes through this mistake
The possible more than one of the Choosing Hydrological Reference Basin that journey is found asks each mean parameter as no money for there is the case where multiple Choosing Hydrological Reference Basins
Stream field parameter value；The parameter value of the case where for only one Choosing Hydrological Reference Basin, the Choosing Hydrological Reference Basin can be applied directly to non-avaible
Basin.So far, unga(u)ged basin parameter identification is completed.
Above in conjunction with attached drawing, an exemplary description of the invention, it is clear that realization of the invention is not by aforesaid way
Limitation, as long as use the inventive concept and technical scheme of the present invention carry out various improvement, or it is not improved will be of the invention
Conception and technical scheme directly apply to other occasions, be within the scope of the invention.
Claims (9)
1. a kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning, it is characterised in that: including following
Step:
Step 1, data preparation: including the Underlying Surface Data and long Series of Water vigour of style image data in data basin, unga(u)ged basin
Underlying Surface Data；
Step 2, acquisition have data Watershed Hydrologic Models parameter set: carrying out rate to hydrological model using long Series of Water vigour of style image data
Fixed and verifying obtains；
Step 3, drainage characteristics attribute Principle component extraction: to every class drainage characteristics, establishing Factor Analysis Model, with principal component point
Analysis method solving model obtains the principal component score value of each drainage characteristics type；
Step 4, building machine learning decision Tree algorithms data set: the data in data set include respectively having data basin principal component to obtain
Divide the distance between difference, basin, the deterministic coefficient of watershed parameters cross validation, wherein deterministic coefficient is as decision tree structure
The dependent variable built, distance is as independent variable between principal component scores difference and basin；
Step 5, building decision tree, generate optimum decision tree；
Step 6, acquisition basin optimal classification rule regular according to decision tree classification；
Step 7, unga(u)ged basin parameter obtain: calculating unga(u)ged basin and respectively have data basin principal component difference, non-avaible stream
Domain and respectively there is distance between data basin, there is data basin to study and compare as optimal according to step 6 gained optimal classification Rules Filtering
Basin obtains Hydro-Model Parameter Calibration Technology.
2. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 1, feature
Be: in step 1, the Underlying Surface Data includes constructing topography and geomorphology data, the land use data, soil that hydrological model needs
The climate characteristic data and river basin authorities data in earth categorical data and basin.
3. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 1, feature
Be: in step 1, the long Series of Water vigour of style image data for having data basin refers to 10 years or more rainfall runoff data, number
The time is represented according to that should include that large, medium and small water is various；Flood play, humid region are no less than 50, many 25 of arid area.
4. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 3, feature
It is: in step 2, is tested with 2/3 session data calibration Hydro-Model Parameter Calibration Technology, rear 1/3 data before long Series of Water vigour of style image data
Demonstrate,prove Hydro-Model Parameter Calibration Technology.
5. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 1, feature
Be: in step 2, Hydro-Model Parameter Calibration Technology calibration and verifying precision select deterministic coefficient to evaluate, it is desirable that calibration and verifying
Deterministic coefficient mean value is all larger than 0.7.
6. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 1, feature
Be: in step 3, the drainage characteristics in drainage characteristics attribute Principle component extraction includes watershed unit geomorphic feature, soil types
With the climate characteristic and basin structural character of soil texture feature, land use and vegetative coverage feature, basin.
7. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 1, feature
Be: in step 4, machine learning data set includes training set, verifying collection and test set.
8. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 7, feature
It is: in step 5, complete decision tree is constructed by training set, beta pruning and optimization, test set is carried out to decision tree by verifying collection
Decision tree is advanced optimized, optimum decision tree is finally obtained.
9. the Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning according to claim 1, feature
Be: in step 6, decision tree classification rule refers to the branching rule of decision tree, the quantity of the rule and the leaf section of decision tree
It counts identical；Basin optimal classification rule refers to the maximum decision tree classification rule of dependent variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910106649.4A CN109840873A (en) | 2019-02-02 | 2019-02-02 | A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910106649.4A CN109840873A (en) | 2019-02-02 | 2019-02-02 | A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109840873A true CN109840873A (en) | 2019-06-04 |
Family
ID=66884564
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910106649.4A Pending CN109840873A (en) | 2019-02-02 | 2019-02-02 | A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109840873A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110442937A (en) * | 2019-07-24 | 2019-11-12 | 武汉大学 | A kind of valley water security management method merging satellite remote sensing and machine learning techniques |
| CN110659823A (en) * | 2019-09-21 | 2020-01-07 | 四川大学工程设计研究院有限公司 | Similar watershed analysis method, model, system and computer storage medium |
| CN111680828A (en) * | 2020-05-21 | 2020-09-18 | 广州鑫泓设备设计有限公司 | Method for carrying out mountain torrent early warning based on time-space variable source mixed runoff production |
| CN111914487A (en) * | 2020-08-14 | 2020-11-10 | 贵州东方世纪科技股份有限公司 | Data-free regional hydrological parameter calibration method based on antagonistic neural network |
| CN111914488A (en) * | 2020-08-14 | 2020-11-10 | 贵州东方世纪科技股份有限公司 | Data regional hydrological parameter calibration method based on antagonistic neural network |
| CN111914465A (en) * | 2020-08-14 | 2020-11-10 | 贵州东方世纪科技股份有限公司 | Data-free regional hydrological parameter calibration method based on clustering and particle swarm optimization |
| CN112131989A (en) * | 2020-09-15 | 2020-12-25 | 河海大学 | Millimeter wave rain measurement model parameter obtaining method based on space rainfall data |
| CN113887635A (en) * | 2021-10-08 | 2022-01-04 | 河海大学 | Basin similarity classification method and classification device |
| CN114662310A (en) * | 2022-03-22 | 2022-06-24 | 中国水利水电科学研究院 | Machine learning-based method and device for regionalization of data-free small watershed parameters |
| CN114943361A (en) * | 2022-03-15 | 2022-08-26 | 水利部交通运输部国家能源局南京水利科学研究院 | Method for estimating evapotranspiration of reference crops in data-lacking areas |
| CN117763970A (en) * | 2023-12-26 | 2024-03-26 | 水利部交通运输部国家能源局南京水利科学研究院 | Image deep learning-based hydrologic model parameter reconstruction method for scarce data area |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107330086A (en) * | 2017-07-04 | 2017-11-07 | 中国科学院新疆生态与地理研究所 | A kind of method for improving non-avaible High aititude River Basin Hydrology process simulation precision |
| CN108108860A (en) * | 2018-02-22 | 2018-06-01 | 河海大学 | A kind of four steps coupling MEDIUM OR LONG RANGE HYDROLOGIC FORECAST METHOD |
-
2019
- 2019-02-02 CN CN201910106649.4A patent/CN109840873A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107330086A (en) * | 2017-07-04 | 2017-11-07 | 中国科学院新疆生态与地理研究所 | A kind of method for improving non-avaible High aititude River Basin Hydrology process simulation precision |
| CN108108860A (en) * | 2018-02-22 | 2018-06-01 | 河海大学 | A kind of four steps coupling MEDIUM OR LONG RANGE HYDROLOGIC FORECAST METHOD |
Non-Patent Citations (2)
| Title |
|---|
| S. RAGETTLI ET AL.: "Modeling flash floods in ungauged mountain catchments of China: A decision tree learning approach for parameter regionalization", 《JOURNAL OF HYDROLOGY》 * |
| YALI WANG ET AL.: "Forecasting and Providing Warnings of Flash Floods for Ungauged Mountainous Areas Based on a Distributed Hydrological Model", 《WATER》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110442937A (en) * | 2019-07-24 | 2019-11-12 | 武汉大学 | A kind of valley water security management method merging satellite remote sensing and machine learning techniques |
| CN110659823A (en) * | 2019-09-21 | 2020-01-07 | 四川大学工程设计研究院有限公司 | Similar watershed analysis method, model, system and computer storage medium |
| CN111680828A (en) * | 2020-05-21 | 2020-09-18 | 广州鑫泓设备设计有限公司 | Method for carrying out mountain torrent early warning based on time-space variable source mixed runoff production |
| CN111914487B (en) * | 2020-08-14 | 2023-04-07 | 贵州东方世纪科技股份有限公司 | Data-free regional hydrological parameter calibration method based on antagonistic neural network |
| CN111914488A (en) * | 2020-08-14 | 2020-11-10 | 贵州东方世纪科技股份有限公司 | Data regional hydrological parameter calibration method based on antagonistic neural network |
| CN111914465A (en) * | 2020-08-14 | 2020-11-10 | 贵州东方世纪科技股份有限公司 | Data-free regional hydrological parameter calibration method based on clustering and particle swarm optimization |
| CN111914487A (en) * | 2020-08-14 | 2020-11-10 | 贵州东方世纪科技股份有限公司 | Data-free regional hydrological parameter calibration method based on antagonistic neural network |
| CN111914488B (en) * | 2020-08-14 | 2023-09-01 | 贵州东方世纪科技股份有限公司 | Data area hydrologic parameter calibration method based on antagonistic neural network |
| CN111914465B (en) * | 2020-08-14 | 2024-04-12 | 贵州东方世纪科技股份有限公司 | Clustering and particle swarm optimization-based method for calibrating hydrologic parameters of non-data region |
| CN112131989A (en) * | 2020-09-15 | 2020-12-25 | 河海大学 | Millimeter wave rain measurement model parameter obtaining method based on space rainfall data |
| CN112131989B (en) * | 2020-09-15 | 2021-07-23 | 河海大学 | Millimeter wave rain measurement model parameter obtaining method based on space rainfall data |
| CN113887635A (en) * | 2021-10-08 | 2022-01-04 | 河海大学 | Basin similarity classification method and classification device |
| CN114943361A (en) * | 2022-03-15 | 2022-08-26 | 水利部交通运输部国家能源局南京水利科学研究院 | Method for estimating evapotranspiration of reference crops in data-lacking areas |
| CN114662310A (en) * | 2022-03-22 | 2022-06-24 | 中国水利水电科学研究院 | Machine learning-based method and device for regionalization of data-free small watershed parameters |
| CN114662310B (en) * | 2022-03-22 | 2023-03-03 | 中国水利水电科学研究院 | Machine learning-based method and device for regionalizing small data domain parameters without data |
| CN117763970A (en) * | 2023-12-26 | 2024-03-26 | 水利部交通运输部国家能源局南京水利科学研究院 | Image deep learning-based hydrologic model parameter reconstruction method for scarce data area |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109840873A (en) | A kind of Cross Some Region Without Data Hydro-Model Parameter Calibration Technology fields method based on machine learning | |
| CN109858647B (en) | Regional flood disaster risk evaluation and estimation method coupled with GIS and GBDT algorithm | |
| Nourani et al. | Estimation of prediction interval in ANN-based multi-GCMs downscaling of hydro-climatologic parameters | |
| CN101480143B (en) | Method for predicating single yield of crops in irrigated area | |
| Bowden et al. | Input determination for neural network models in water resources applications. Part 1—background and methodology | |
| CN109242203A (en) | A kind of water quality prediction of river and water quality impact factors assessment method | |
| Thissen et al. | Dealing with uncertainties in fresh water supply: Experiences in the Netherlands | |
| Rodríguez-Vázquez et al. | Rainfall–runoff modelling using genetic programming | |
| Mediero et al. | Probabilistic calibration of a distributed hydrological model for flood forecasting | |
| Tigkas et al. | Comparative study of evolutionary algorithms for the automatic calibration of the Medbasin-D conceptual hydrological model | |
| Prakash et al. | Improved higher lead time river flow forecasts using sequential neural network with error updating | |
| CN106097351A (en) | A kind of based on multiobject adaptive threshold image partition method | |
| Rocha et al. | Early prediction of durum wheat yield in Spain using radial basis functions interpolation models based on agroclimatic data | |
| CN117787081A (en) | Hydrological model parameter uncertainty analysis method based on Morris and Sobol methods | |
| CN117763970B (en) | Image deep learning-based hydrologic model parameter reconstruction method for scarce data area | |
| Karimi et al. | Monthly rainfall prediction using ARIMA and gene expression programming: A case study in Urmia, Iran | |
| Lu et al. | Application of grey relational analysis for evaluating road traffic safety measures: advanced driver assistance systems against infrastructure redesign | |
| Chumbe-Llimpe et al. | Comparison of the machine learning and AquaCrop models for quinoa crops. | |
| CN113673777B (en) | Desert succession prediction method under climate change condition | |
| CN116563714A (en) | Method and system for automatically distinguishing growth stage of rice | |
| Mahmoudi et al. | Regionalization of rainfall intensity–duration–frequency (IDF) curves with L-moments method using neural gas networks | |
| Jiang et al. | Discharge estimation based on machine learning | |
| CN107871174A (en) | A kind of price of medicinal material Forecasting Methodology | |
| Barua et al. | Drought assessment and forecasting: A case study on the Yarra River catchment in Victoria, Australia | |
| Wu et al. | Application of K-means cluster and rough set in classified real-time flood forecasting |
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: 20190604 |
|
| RJ01 | Rejection of invention patent application after publication |