CN110471950A - A kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure - Google Patents
A kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure Download PDFInfo
- Publication number
- CN110471950A CN110471950A CN201910653280.9A CN201910653280A CN110471950A CN 110471950 A CN110471950 A CN 110471950A CN 201910653280 A CN201910653280 A CN 201910653280A CN 110471950 A CN110471950 A CN 110471950A
- Authority
- CN
- China
- Prior art keywords
- flood
- data
- model
- mind
- forecasting
- 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 64
- 238000012360 testing method Methods 0.000 claims abstract description 15
- 238000012706 support-vector machine Methods 0.000 claims abstract description 12
- 238000013459 approach Methods 0.000 claims abstract description 10
- 238000013501 data transformation Methods 0.000 claims abstract description 6
- 238000011160 research Methods 0.000 claims abstract description 5
- 238000012549 training Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 16
- 238000005259 measurement Methods 0.000 claims description 13
- 238000011156 evaluation Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 7
- 239000002689 soil Substances 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000002156 mixing Methods 0.000 claims description 4
- 230000007704 transition Effects 0.000 claims description 4
- 238000009825 accumulation Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims 1
- 230000008901 benefit Effects 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 12
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 11
- 239000013598 vector Substances 0.000 description 10
- 238000012545 processing Methods 0.000 description 7
- 238000001556 precipitation Methods 0.000 description 6
- 238000004140 cleaning Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 3
- 230000007812 deficiency Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000013277 forecasting method Methods 0.000 description 3
- 238000006424 Flood reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000008020 evaporation Effects 0.000 description 2
- 238000001704 evaporation Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/21—Design, administration or maintenance of databases
- G06F16/215—Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Probability & Statistics with Applications (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Marketing (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Educational Administration (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 invention discloses a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedures, firstly, the history hydrographic data of each website in collection research basin, is then stored in hydrology historical data base；Secondly, carrying out data cleansing, data transformation, data set division to hydrology historical data；Again, construct the Real-time Flood Forecasting model in " three stages ", first stage forecasts flood using the data-driven model based on support vector machines, second stage carries out intelligent adjustment using state of the Markov approach to initial forecasting model, then phase III identical historical flood flow rate mode during searching in flood pattern library with real-time flood carries out the adjustment of configuration to the flood hydrograph of real-time prediction；Then test set data assessment model of mind performance is used；Finally, carrying out real-time prediction.It is the effective tool of middle and small river flood forecasting the invention has the benefit that flood peak precision and Flood Tendency can effectively be forecast.
Description
Technical field
The present invention relates to data-driven water flow forecasting technique fields, and in particular to a kind of middle and small river Real-time Flood Forecasting intelligence
It can model prediction method.
Background technique
Middle Flood of small drainage area have the characteristics that it is sudden it is strong, the concentration time is fast, leading time is short.Timely and effectively middle and small river
Flood damage evaluation can help the mankind effectively defending flood, reduction flood loss, be important non-engineering measure of preventing and reducing natural disasters
One of.Flood forecasting generally uses the model of mind two ways of the hydrological model based on runoff process and data-driven at present,
And two models complement each other in actually forecast.Data-driven models does not consider the physical mechanism of hydrologic process substantially, be with
The optimal mathematical relationship established between inputoutput data is the black box submethod of target.Middle and small river has the complicated hydrology special
Property, the nonlinear influencing factors such as boundary condition and active mankind's activity, for existing intelligent Flood Forecasting Model in middle river
Adaptability and precision deficiency problem in real-time prediction are flowed, takes real time correction technology suitably to correct prediction error and be necessary and arranges
It applies.
Timely and effectively middle and small river flood damage evaluation, be at present it is most effective it is feasible prevent and reduce natural disasters non-engineering measure it
One.In recent two decades, using data driven technique model of mind predictions and simulations Nonlinear Hydrological application and capture
It is made great progress in terms of the noise that complex data is concentrated.Classical data-driven modeling method mainly has artificial neural network
Network, support vector machines (SVM), fuzzy logic, evolutionary computation etc..In the above-mentioned methods, each Flood Forecasting Model is all to compare to have
The forecasting procedure of effect is substantially that Watershed Runoff forms physical process by the hydrological model that actual measurement rainfall and Streamflow Data construct
One generalization, error is inevitable, therefore flood forecasting is to the appropriate amendment of prediction error work using real time correction technology
A kind of necessary measure.
Summary of the invention
For existing intelligent Flood Forecasting Model adaptability and precision deficiency problem, this hair in middle and small river real-time prediction
Bright purpose is the Real-time Flood Forecasting method for proposing to be based on " three stages ", can effectively forecast that flood peak precision and flood become
Gesture is the effective tool of middle and small river flood forecasting.
To achieve the goals above, the present invention is to realize by the following technical solutions:
A kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure, comprising the following steps:
Then the hydrographic data being collected into is stored in water by step 1, the hydrographic data of the middle small watershed of collection research
Literary historical data base；
Step 2 pre-processes the hydrographic data in the hydrology historical data base, and the pretreatment includes data
Cleaning, data transformation, data set divide；
Step 3 constructs real-time prediction model of mind, and the first stage of the model of mind initially models, using based on branch
Hold the data-driven model forecast flood of vector machine；
Step 4, the second stage situation optimization of the model of mind, using Markov approach to initial forecasting model
State carry out intelligent adjustment；
Step 5, the phase III forecasting process wire shaped control of the model of mind, uses water in flood pattern library
The polynary blending algorithm of text carries out method for measuring similarity, searching and real-time flood to the time sequence model of each flooding schedule factor
Then the identical historical flood flow rate mode of process carries out the adjustment of configuration to the flood hydrograph of real-time prediction；
Step 6 uses test set data assessment model of mind performance；
Step 7, real-time prediction input hydrographic data pretreated in the step 2 as model of mind, future
24 hours basin Outlet Section flows are model of mind output.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in said step 1, collection is ground
The hydrographic data for the middle small watershed studied carefully includes history rainfall data in basin, the historical traffic data of basin Outlet Section and basin
History evaporates data, and the hydrographic data being collected into then is stored in hydrology historical data base.
Above-mentioned a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 2, to described
Hydrology historical data is pre-processed, and the data cleansing includes removal apparent peel off noise data and repeated data, and is mended
Full missing data；Data cleansing and completion are realized automatically by Principle of Statistics.
Above-mentioned a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 2, to described
Hydrology historical data is pre-processed, and the data are transformed to convert the data by normalized fashion suitable for model training
Mode, specific method be standard normalization.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 2, normalization
Treated, and data set is divided into training set and test set, wherein data set division methods are to choose a wherein flood data
As test set, remaining flood play is as training set；Training dataset training prevents model excessively quasi- using cross-validation method
It closes.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 3, the branch
It is Radial basis kernel function that the model for holding vector machine, which chooses kernel function,；The parameter of the initial modelling phase support vector machines of model of mind is adopted
It is automatically generated with the library LIBSVM grid optimizing function SVMcgForRegress function.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 4, building is real
When forecast model of mind, model of mind second stage situation optimization, using Markov approach to the state of initial forecasting model
Carry out intelligent adjustment；The described intelligent adjustment refer to Markov approach on the basis of the forecast result of the step 3,
By calculating state transition probability matrix, the residual values of forecast subsequent time are obtained to assess the matching journey of predicted value and observation
Degree corrects predicted value result.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 5, the intelligence
Can the model phase III building flood pattern library be based on the polynary one-dimensional Hydrological Time Series data with correlation, it is described
Hydrological Time Series data include period areal rainfall time series data, accumulation period areal rainfall data, rainfall trend time sequence
Column data and soil moisture content time series data, then using the polynary blending algorithm of the hydrology to each flooding schedule factor when
Between sequence pattern carry out similarity measurement；The flood pattern library of the model of mind phase III building matches the history come
Flood discharge mode carries out adjustment in shape to the forecast result that the model of mind second stage Markov corrects；It is described
The correction that the adjustment of configuration carries out using least square method final forecast result is carried out to the flood hydrograph of real-time prediction.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure uses survey in the step 6
Examination collection data assessment model of mind performance, three stages for specially assessing the model of mind using four kinds of evaluation criterias are big vast in real time
Water forecast result, respectively flood peak peak value relative error QmaxRelative error Δ t between [%], peak are currentmax[h], root-mean-square error
RMSE, coefficient of determination R2。
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure, in the step 6, described four
Kind evaluation criteria formula is as follows:
1) flood peak peak value relative error Qmax[%]:
Wherein, Qmax--- crest discharge actual value, for the historical data of ground hydrometric station measurement；--- flood peak stream
Predicted value is measured, is three stage Real-time Flood Forecasting results of the model of mind；
2) relative error Δ t between peak is currentmax[h]:
Wherein, tmax--- practical flood time of peak,--- forecasting runoff time of peak；
3) root-mean-square error RMSE:
4) coefficient of determination R2:
Wherein, Qi--- in i moment flood-discharge measurement value,--- in i moment flood discharge forecast value,--- flood
Water flow average observed value,--- flood discharge is averaged predicted value, N --- test sample sum, i is 1 to n-hour.
A kind of above-mentioned middle and small river Real-time Flood Forecasting model of mind forecasting procedure is practical forecast in the step 7
Link, the hydrographic data newly arrived are also needed by pretreatment, then using data after processing as the input of forecasting model, future
24 hours basin Outlet Section flows are model output, i.e. 24 hours futures of model prediction flow.
The invention has the benefit that
(1) present invention utilizes deep learning algorithm, using a kind of Real-time Flood Forecasting method for being based on " three stages ", with biography
The method of system is compared, and is taken real time correction technology and is suitably corrected prediction error, is solved existing intelligent Flood Forecasting Model and is existed
Adaptability and precision deficiency problem in middle and small river real-time prediction.
(2) the intelligent Real-time Flood Forecasting model in " three stages " proposed by the present invention.First stage is initial modeling, mainly
Use the data-driven model based on support vector machines as the initial forecast stage；Second stage is situation optimization, which adopts
Intelligent adjustment is carried out with state of the Markov approach to initial forecasting model；Phase III is shape control, is being constructed
Flood pattern library in find with the most like historical flood flow rate mode of real-time flood process, utilize obtained similar flow mould
Formula carries out the adjustment of configuration to the flood pattern of real-time prediction." three stages " model is that a gradual increment is intelligently pre-
Report model, forecasting model can in real time to dynamic flood data carry out effective Feedback and in time adjustment model state with finally it is defeated
Out, accomplish the high accuracy prediction of middle and small river.
Detailed description of the invention
The following describes the present invention in detail with reference to the accompanying drawings and specific embodiments:
Fig. 1 is experiment flow figure of the invention；
Fig. 2 is the prosperousization basin water system that the present invention studies and station net distribution map；
Fig. 3 is the Heihe River basin water system that the present invention studies and station net distribution map；
Fig. 4 initially models the support vector regression forecast principle schematic diagram used for And Real-time Forecasting Model of the invention；
Fig. 5 is that And Real-time Forecasting Model difference of the invention forecasts stage forecast result performance comparison figure.
Specific embodiment
To be easy to understand the technical means, the creative features, the aims and the efficiencies achieved by the present invention, below with reference to
Specific embodiment, the present invention is further explained.
As shown in Figure 1, firstly, the history hydrographic data of each website in collection research basin, is then stored in hydrology historical data
Library；Secondly, carrying out data cleansing, data transformation, data set division etc. to hydrology historical data；Again, it constructs " three stages "
Real-time Flood Forecasting model, wherein the first stage forecasts flood, second-order using the data-driven model based on support vector machines
Duan Caiyong Markov approach carries out intelligent adjustment to the state of initial forecasting model, and the phase III seeks in flood pattern library
The historical flood flow rate mode most like with real-time flood process is looked for, whole shape then is carried out to the flood hydrograph of real-time prediction
The adjustment of state；Then test set data assessment model of mind performance is used；Finally, using the Real-time Flood Forecasting mould in " three stages "
Type carries out real-time prediction.
The specific implementation steps are as follows by the present invention:
Step 1: selecting prosperousization basin and Heihe River basin as experimental basin, basin water system as shown in Figures 2 and 3 with
It stands net distribution map, 28 flood data of 31 floods of prosperousization 1998-2010 and Heihe 2003-2010 is collected, wherein each water
Literary survey station flood season data are the hydrographic data of 1/hour of granularity, and every data includes each precipitation station rainfall in basin, and basin goes out
Mouth section flow, basin evaporation capacity.Totally 6 rainfall survey stations, a basin export discharge site (measuring flow in prosperousization basin
And rainfall) an and evaporation survey station, i.e., every data includes 9 dimension information；Equally, totally 8 rainfall survey stations in black basin, one
Basin exports discharge site (measuring flow and rainfall), i.e., every data includes 10 dimension information, and then the above hydrographic data is deposited
Enter hydrology historical data base；
Step 2: taking out data from historical data base and go forward side by side line number Data preprocess, including data cleansing, data transformation,
Data set division etc..Data cleansing of the present invention is using the cleaning for having supervision, specific cleaning way are as follows: in the finger of hydrology domain expert
The characteristics of leading down, and combining statistical method analysis data simultaneously defines data cleansing rule.Main cleaning way includes: number
According to the inspection and processing of null value；The detection and processing of data illegal value；The detection and processing of inconsistent data；Duplicated records
Detection and processing, ultimately produce data cleansing report.Wherein, the completion of assigning null data uses method for automatically completing, this side
Method fills a null value according to the value distribution situation of data centralized recording, specifically automatically by Principle of Statistics
It is substituted for single missing data using the average value of same website surrounding time, continuous missing data uses same time phase
The average value of adjacent survey station point substitutes.
Carrying out data transformation to hydrology historical data is exactly by data conversion or to be unified for and be appropriate for data mining
Form, having for relating generally to simply are summarized and are assembled to obtain statistical information by the data concentrated to hydrographic data；
Followed by data interpolating is handled, and each hydrographic data is processed by way of linear interpolation the input of same time granularity；Most
The standardization processing of data is zoomed in and out to each hydrology attributive character data afterwards, make it possible to drop into a specific region it
Between, make the feature between different dimensions numerically and have certain comparative, mode, specific formula are normalized using Min-max are as follows:
Wherein,For the data after normalized, xobsFor original survey station measurement data to be processed,For original survey
Minimum value in measurement data of standing,For the maximum value in original survey station measurement data, the data value after normalized is 0
To between 1.
Data set after normalized is divided into training set and test set, wherein data set division methods are to choose it
In a flood data as test set, remaining flood play is as training set；It is only repeated for Flood Forecasting Model
The label for the flood sample trained obtains branch very high in this case, but encounters the sample that do not trained then
Forecast result can be very poor, and training dataset training prevents model over-fitting using cross-validation method, uses 10 times in specific experiment
Cross validation.
Step 3: the intelligent prediction model first stage initially models, and is driven using the data based on support vector regression (SVM)
Movable model forecasts flood.SVM is asked based on structural risk minimization principle in the flood forecasting for solving to belong to nonlinear regression
When topic, SVM be nonlinear regression problem in lower dimensional space is mapped to higher dimensional space by kernel function by hydrology historical sample, and
Optimum regression function is solved in higher dimensional space, the linear regression for thereby realizing higher dimensional space corresponds to the non-thread of lower dimensional space
Property return, while preferable Generalization Ability is obtained by the complexity of Controlling model and approximation accuracy
Give a historical flood training sample set D={ (x1,y1),(x2,y2),…,(xl,yl), xk∈Rn, it is that n is tieed up
Flood forecasting factor input vector, yiIt is corresponding flood forecasting value, l is the flood sample size of training, k=1,2 ..., l.
Flood forecasting problem is exactly to pass through training sample set, finds the mapping relations between input and output, i.e. y=f (x).Support to
The basic thought of amount machine forecasting model be exactly by by the input of sample from luv space be mapped to the space of a more higher-dimension into
Row linear regression, the regression model that learns make f (x) and y as close possible to regression model is as follows:
F (x)=w φ (x)+b (2)
φ () is a Nonlinear Mapping in formula, and w is weight vectors, w ∈ Rn, b is biasing, b ∈ R.Assuming that supporting vector
Recurrence can tolerate between f (x) and y that maximum deviation is ε, therefore calculating is damaged when the absolute value of the bias between f (x) and y is greater than ε
It loses.As shown in figure 4, constructing the intervallum that a width is 2 ε in the two sides of f (x), when training sample is in this intervallum, then
The sample predictions are correct.What wherein support vector machines forecasting model kernel function was chosen is Radial basis kernel function, form are as follows:
The bandwidth (width) of σ in formula --- kernel function, characterizes the width of closed boundary, xm,xnWhen to be respectively m and n
Carve flood forecasting factor input vector.
In the model of mind initial modelling phase, optimal parameter is sought using the library LIBSVM grid when supporting vector machine model is trained
Major function SVMcgForRegress function automatically generates.Wherein the selection of SVMcgForRegress function optimal parameter has punishment
Parameter c and RBF nuclear parameter g, the value range of the two is all [2cmin,2cmax] in, wherein default cmin=-8, cmax=8.
Step 4, model of mind second stage situation optimization, using Markov approach to the state of initial forecasting model
Carry out intelligent adjustment.Assuming that flood forecasting leading time is T, initial modelling phase supporting vector machine model flood forecasting result is
Y={ y1,y2,…,yT, practical flood value is Y '={ y1′,y′2,…,y′T, pass through ξd=Yd-Yd' obtain predicted value and observation
Residual sequence ξ={ ξ of value1,ξ2,…,ξT}.D-th of moment, the state of residual sequence is
Πd=[yd+min(ξf),yd+max(ξf)] (4)
State transition probability matrix formula is
pd=p (ξL+1∈Πf|ξL∈Πd) (5)
D, f=1,2 ... in formula, n, n are residual sequence number of states, SdFor state Π in residual sequence ξdSample number
Amount, SdfIt (u) is state ΠdState Π is converted to through u stepfSample size.
By the state transition probability matrix that L forecast result sample of preliminary forecasting model calculates, available L
Then+1 residual values correct the predicted value of L+1 by the residual valuesFormula is as follows
Step 5, the control of model of mind phase III forecasting process wire shaped are found and flood in real time in flood pattern library
Then the most like historical flood flow rate mode of water process carries out the adjustment of configuration to the flood hydrograph of real-time prediction.
The building in flood pattern library is that the formation based on the similarity measurement to peb process, in view of flood is each flooding schedule factor sequence
Column mode interaction as a result, therefore reflecting peb process early period using the similitude of flooding schedule factor sequence mode
Similitude.
When each flooding schedule of the polynary blending algorithm of the hydrology includes: period areal rainfall because of Time Sub-series in the present patent application
Between sequence, accumulation period areal rainfall, rainfall trend time series, soil moisture content time series.Wherein period areal rainfall refers to
The area weight sum of products FR (t that the same hourly precipitation amount of each precipitation station in basin and the precipitation station can influencep)；Rainfall trend
Time series RT such as following formula
RT={ FR (t1),FR(t2),FR(t3),...,FR(tp),...,FR(tq), (0 < p≤q) (8)
In formula, FR (tp) it is tpThe rainfall intensity of hour；For tpZ rainfall website hourly precipitation amount of hour
Average value, q are rainfall sequence total time length, tpAt the time of for precipitation time series length being p.
Soil moisture content WMtIt may be expressed as:
WMt={ WM (t1),WM(t2),…,WM(tp),WM(tm) (0 < p≤r) (10)
In formula: WMtFor soil moisture content time series, WM (t1) it is t1The soil moisture content at moment, WM (tp) it is tpMoment
Soil moisture content, r is length of time series.
Then the characterization factor similitude for calculating each play flood two-by-two respectively using Euclidean distance, finds out similar flood
Process carries out the correction of flood discharge in shape for intelligent prediction model second stage forecast result on this basis, here
Using least square method.
Step 6 assesses " three stages " using test set data assessment model of mind performance, and using four kinds of evaluation criterias
Real-time Flood Forecasting is as a result, respectively flood peak peak value relative error QmaxRelative error Δ t between [%], peak are currentmaxIt is [h], square
Root error RMSE, coefficient of determination R2.Four kinds of evaluation criteria formula are as follows:
1) flood peak peak value relative error Qmax[%]:
Wherein Qmax--- crest discharge actual value, the historical data of ground hydrometric station measurement,--- crest discharge is pre-
Report value, three stage Real-time Flood Forecasting results of model of mind.
2) relative error Δ t between peak is currentmax[h]:
Wherein tmax--- practical flood time of peak,--- forecasting runoff time of peak.
3) root-mean-square error RMSE:
4) coefficient of determination R2:
Wherein Qi--- in i moment flood-discharge measurement value,--- in i moment flood discharge forecast value,--- flood
Water flow average observed value,--- flood discharge is averaged predicted value, N --- test sample sum.
Wherein " three stages " Real-time Flood Forecasting method is as shown in the table in the analog result of prosperousization and Heihe River basin.
Step 7, this step are real-time prediction link, are inputted pretreated current hydrographic data as model of mind,
Following 24 hours basin Outlet Section flows are model of mind output.The hydrographic data newly arrived is also needed by pretreatment, so
Afterwards using data after processing as the input of forecasting model, following 24 hours basin Outlet Section flows are model output, i.e. model
24 hours futures flow of forecast.As shown in figure 5, showing prosperousization and one, Heihe Real-time Flood Forecasting respectively as a result, wherein Fig. 5
In (a) figure be forecast result of the flood of prosperousization basin 20000828 after second stage corrects, (b) figure is Heihe in Fig. 5
Forecast result of the flood of basin 20100821 after second stage corrects, (c) figure is prosperousization basin 20000828 in Fig. 5
Forecast result of the flood after the phase III corrects, (d) figure is No. 20100821 floods of Heihe River basin by third rank in Fig. 5
Forecast result after Duan Jiaozheng.The result shows that having between the pattern base correction of phase III is current to the peak of the flood hydrograph of forecast
Biggish precision improves.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.Industry description
Merely illustrate the principles of the invention, without departing from the spirit and scope of the present invention, the present invention also have various change and
It improves, these changes and improvements all fall within the protetion scope of the claimed invention.The claimed scope of the invention is by appended power
Sharp claim and its equivalent thereof.
Claims (10)
1. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure, which comprises the following steps:
Then step 1, the hydrographic data of the middle small watershed of collection research are gone through the hydrographic data the being collected into deposit hydrology
History database；
Step 2 pre-processes the hydrographic data in the hydrology historical data base, it is described pretreatment include data cleansing,
Data transformation, data set divide；
Step 3 constructs real-time prediction model of mind, and the first stage of the model of mind initially models, using based on support to
The data-driven model of amount machine forecasts flood；
Step 4, the second stage situation optimization of the model of mind, using Markov approach to the shape of initial forecasting model
State carries out intelligent adjustment；
Step 5, the phase III forecasting process wire shaped control of the model of mind are more using the hydrology in flood pattern library
First blending algorithm carries out method for measuring similarity to the time sequence model of each flooding schedule factor, finds and real-time flood process
Then identical historical flood flow rate mode carries out the adjustment of configuration to the flood hydrograph of real-time prediction；
Step 6 uses test set data assessment model of mind performance；
Step 7, real-time prediction are inputted hydrographic data pretreated in the step 2 as model of mind, and future 24 is small
When basin Outlet Section flow be model of mind output.
2. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In said step 1, the hydrographic data of the middle small watershed of collection research includes that history rainfall data in basin, basin outlet are disconnected
The historical traffic data and basin history in face evaporate data, and the hydrographic data being collected into then is stored in hydrology historical data
Library.
3. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In the step 2, the hydrology historical data is pre-processed, the data cleansing includes removing the noise data that peels off
And repeated data, and completion missing data；Data cleansing and completion are realized automatically by Principle of Statistics.
4. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In the step 2, the hydrology historical data is pre-processed, the data are transformed to count by normalized fashion
According to the mode being converted into suitable for model training, specific method is standard normalization.
5. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 4, it is characterised in that:
In the step 2, the data set after normalized is divided into training set and test set, wherein data set division methods are
Wherein a flood data are as test set for selection, and remaining flood play is as training set；Training dataset training is using friendship
Fork proof method prevents model over-fitting.
6. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In the step 3, it is Radial basis kernel function that the model of the support vector machines, which chooses kernel function,；Model of mind initially models
The parameter of stage support vector machines is automatically generated using the library LIBSVM grid optimizing function SVMcgForRegress function.
7. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In the step 4, real-time prediction model of mind, model of mind second stage situation optimization, using Markov approach are constructed
Intelligent adjustment is carried out to the state of initial forecasting model；The intelligent adjustment refers to Markov approach in the step 3
Forecast result on the basis of, by calculate state transition probability matrix, obtain forecast subsequent time residual values it is pre- to assess
The matching degree of report value and observation corrects predicted value result.
8. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In the step 5, the flood pattern library of the model of mind phase III building is based on polynary one-dimensional with correlation
Hydrological Time Series data, the Hydrological Time Series data include period areal rainfall time series data, accumulation period face rain
Data, rainfall trend time series data and soil moisture content time series data are measured, is then calculated using the polynary fusion of the hydrology
Method carries out similarity measurement to the time sequence model of each flooding schedule factor；The flood of the model of mind phase III building
Pattern base matches the forecast knot that the historical flood flow rate mode come corrects the model of mind second stage Markov
Fruit carries out adjustment in shape；The adjustment that the flood hydrograph to real-time prediction carries out configuration uses least square method
Carry out the correction of final forecast result.
9. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 1, it is characterised in that:
In the step 6, using test set data assessment model of mind performance, specially using described in four kinds of evaluation criteria assessments
Three stage Real-time Flood Forecastings of model of mind are as a result, respectively flood peak peak value relative error QmaxIt is opposite between [%], peak are current to miss
Poor Δ tmax[h], root-mean-square error RMSE, coefficient of determination R2。
10. a kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure according to claim 9, feature exist
In: in the step 6, four kinds of evaluation criteria formula are as follows:
1) flood peak peak value relative error Qmax[%]:
Wherein, Qmax--- crest discharge actual value, for the historical data of ground hydrometric station measurement；--- crest discharge is pre-
Report value is three stage Real-time Flood Forecasting results of the model of mind；
2) relative error Δ t between peak is currentmax[h]:
Wherein, tmax--- practical flood time of peak,--- forecasting runoff time of peak；
3) root-mean-square error RMSE:
4) coefficient of determination R2:
Wherein, Qi--- in i moment flood-discharge measurement value,--- in i moment flood discharge forecast value,--- flood stream
Average observed value is measured,--- flood discharge is averaged predicted value, N --- test sample sum, i is 1 to n-hour.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910653280.9A CN110471950B (en) | 2019-07-19 | 2019-07-19 | Intelligent model forecasting method for real-time flood forecasting of medium and small rivers |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910653280.9A CN110471950B (en) | 2019-07-19 | 2019-07-19 | Intelligent model forecasting method for real-time flood forecasting of medium and small rivers |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110471950A true CN110471950A (en) | 2019-11-19 |
| CN110471950B CN110471950B (en) | 2022-09-16 |
Family
ID=68508280
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910653280.9A Active CN110471950B (en) | 2019-07-19 | 2019-07-19 | Intelligent model forecasting method for real-time flood forecasting of medium and small rivers |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110471950B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111260159A (en) * | 2020-02-26 | 2020-06-09 | 刘祖敏 | Meteorological-hydrological coupling flood measuring and reporting method |
| CN111583587A (en) * | 2020-04-20 | 2020-08-25 | 成都万江港利科技股份有限公司 | Small watershed prediction system |
| CN111598724A (en) * | 2020-05-19 | 2020-08-28 | 四川革什扎水电开发有限责任公司 | Time-interval integration method for day-ahead prediction of warehousing flow of small and medium-sized reservoirs |
| CN111814407A (en) * | 2020-07-28 | 2020-10-23 | 安徽沃特水务科技有限公司 | Flood forecasting method based on big data and deep learning |
| CN112084461A (en) * | 2020-08-10 | 2020-12-15 | 河海大学 | KNN real-time correction method based on historical flood learning |
| CN112348246A (en) * | 2020-11-02 | 2021-02-09 | 大连理工大学 | SSA-based standardized noise reduction method and application thereof in flood forecast of mountain drainage basin |
| CN112785043A (en) * | 2020-12-31 | 2021-05-11 | 河海大学 | Flood forecasting method based on time sequence attention mechanism |
| CN112801416A (en) * | 2021-03-10 | 2021-05-14 | 长沙理工大学 | LSTM watershed runoff prediction method based on multi-dimensional hydrological information |
| CN113642699A (en) * | 2021-06-30 | 2021-11-12 | 天津大学 | Intelligent river flood forecasting system |
| CN114707753A (en) * | 2022-04-25 | 2022-07-05 | 河海大学 | Regional LSTM flood forecasting method |
| CN115271154A (en) * | 2022-06-07 | 2022-11-01 | 中国长江电力股份有限公司 | Nonlinear regression flood element prediction method based on polynomial and partial least square coupling |
| CN115510631A (en) * | 2022-09-15 | 2022-12-23 | 长江水利委员会水文局 | Flood process line design method and system considering multi-time flood forms |
| CN116933217A (en) * | 2023-09-19 | 2023-10-24 | 广州华水生态科技有限公司 | Method for forecasting flood of medium and small river based on digital twin and AI technology |
| CN117350180A (en) * | 2023-12-06 | 2024-01-05 | 南京光时代数字科技有限公司 | Method for constructing flood rapid prediction model set under complex working condition |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012036368A1 (en) * | 2010-09-13 | 2012-03-22 | 대한민국(기상청장) | System for monitoring rainfall and water level in real-time and monitoring method using same |
| US20180373993A1 (en) * | 2017-06-23 | 2018-12-27 | University Of Alaska Fairbanks | Method Of Predicting Streamflow Data |
| CN109615011A (en) * | 2018-12-14 | 2019-04-12 | 河海大学 | A kind of middle and small river short time flood forecast method based on LSTM |
| CN109754573A (en) * | 2019-01-21 | 2019-05-14 | 北京理工大学 | A kind of middle and small river flood forecasting pre-warning facility and its operation method |
-
2019
- 2019-07-19 CN CN201910653280.9A patent/CN110471950B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012036368A1 (en) * | 2010-09-13 | 2012-03-22 | 대한민국(기상청장) | System for monitoring rainfall and water level in real-time and monitoring method using same |
| US20180373993A1 (en) * | 2017-06-23 | 2018-12-27 | University Of Alaska Fairbanks | Method Of Predicting Streamflow Data |
| CN109615011A (en) * | 2018-12-14 | 2019-04-12 | 河海大学 | A kind of middle and small river short time flood forecast method based on LSTM |
| CN109754573A (en) * | 2019-01-21 | 2019-05-14 | 北京理工大学 | A kind of middle and small river flood forecasting pre-warning facility and its operation method |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111260159A (en) * | 2020-02-26 | 2020-06-09 | 刘祖敏 | Meteorological-hydrological coupling flood measuring and reporting method |
| CN111260159B (en) * | 2020-02-26 | 2023-06-06 | 刘祖敏 | Meteorological hydrologic coupling flood forecasting method |
| CN111583587A (en) * | 2020-04-20 | 2020-08-25 | 成都万江港利科技股份有限公司 | Small watershed prediction system |
| CN111598724B (en) * | 2020-05-19 | 2022-07-22 | 四川革什扎水电开发有限责任公司 | Time-interval integration method for day-ahead prediction of warehousing flow of small and medium reservoirs |
| CN111598724A (en) * | 2020-05-19 | 2020-08-28 | 四川革什扎水电开发有限责任公司 | Time-interval integration method for day-ahead prediction of warehousing flow of small and medium-sized reservoirs |
| CN111814407B (en) * | 2020-07-28 | 2023-10-13 | 安徽沃特水务科技有限公司 | Flood forecasting method based on big data and deep learning |
| CN111814407A (en) * | 2020-07-28 | 2020-10-23 | 安徽沃特水务科技有限公司 | Flood forecasting method based on big data and deep learning |
| CN112084461A (en) * | 2020-08-10 | 2020-12-15 | 河海大学 | KNN real-time correction method based on historical flood learning |
| CN112084461B (en) * | 2020-08-10 | 2024-05-14 | 河海大学 | KNN real-time correction method based on historical flood learning |
| CN112348246A (en) * | 2020-11-02 | 2021-02-09 | 大连理工大学 | SSA-based standardized noise reduction method and application thereof in flood forecast of mountain drainage basin |
| CN112785043B (en) * | 2020-12-31 | 2022-08-30 | 河海大学 | Flood forecasting method based on time sequence attention mechanism |
| CN112785043A (en) * | 2020-12-31 | 2021-05-11 | 河海大学 | Flood forecasting method based on time sequence attention mechanism |
| CN112801416A (en) * | 2021-03-10 | 2021-05-14 | 长沙理工大学 | LSTM watershed runoff prediction method based on multi-dimensional hydrological information |
| CN113642699A (en) * | 2021-06-30 | 2021-11-12 | 天津大学 | Intelligent river flood forecasting system |
| CN114707753A (en) * | 2022-04-25 | 2022-07-05 | 河海大学 | Regional LSTM flood forecasting method |
| CN115271154A (en) * | 2022-06-07 | 2022-11-01 | 中国长江电力股份有限公司 | Nonlinear regression flood element prediction method based on polynomial and partial least square coupling |
| CN115271154B (en) * | 2022-06-07 | 2023-12-29 | 中国长江电力股份有限公司 | Nonlinear regression flood element prediction method based on polynomial and partial least square coupling |
| CN115510631B (en) * | 2022-09-15 | 2024-03-19 | 长江水利委员会水文局 | Flood process line design method and system considering multiple flood forms |
| CN115510631A (en) * | 2022-09-15 | 2022-12-23 | 长江水利委员会水文局 | Flood process line design method and system considering multi-time flood forms |
| CN116933217A (en) * | 2023-09-19 | 2023-10-24 | 广州华水生态科技有限公司 | Method for forecasting flood of medium and small river based on digital twin and AI technology |
| CN116933217B (en) * | 2023-09-19 | 2024-01-30 | 广州华水生态科技有限公司 | Method for forecasting flood of medium and small river based on digital twin and AI technology |
| CN117350180B (en) * | 2023-12-06 | 2024-02-27 | 南京光时代数字科技有限公司 | Method for constructing flood rapid prediction model set under complex working condition |
| CN117350180A (en) * | 2023-12-06 | 2024-01-05 | 南京光时代数字科技有限公司 | Method for constructing flood rapid prediction model set under complex working condition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110471950B (en) | 2022-09-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110471950A (en) | A kind of middle and small river Real-time Flood Forecasting model of mind forecasting procedure | |
| Fahimi et al. | Application of soft computing based hybrid models in hydrological variables modeling: a comprehensive review | |
| Balakrishnan et al. | Crop production-ensemble machine learning model for prediction | |
| Lipu et al. | Artificial intelligence based hybrid forecasting approaches for wind power generation: Progress, challenges and prospects | |
| CN113705877B (en) | Real-time moon runoff forecasting method based on deep learning model | |
| Alibabaei et al. | Irrigation optimization with a deep reinforcement learning model: Case study on a site in Portugal | |
| Khan et al. | Evaluating the performance of several data mining methods for predicting irrigation water requirement | |
| Calp | A hybrid ANFIS-GA approach for estimation of regional rainfall amount | |
| Al Sadeque et al. | A deep learning approach to predict weather data using cascaded LSTM network | |
| Li et al. | A method of rainfall runoff forecasting based on deep convolution neural networks | |
| Bushara et al. | Computational intelligence in weather forecasting: A review | |
| Saab et al. | Review on generating optimal operation for dam and reservoir water system: simulation models and optimization algorithms | |
| Kumar et al. | Comparative analysis of backpropagation and RBF neural network on monthly rainfall prediction | |
| Elzain et al. | Innovative approach for predicting daily reference evapotranspiration using improved shallow and deep learning models in a coastal region: A comparative study | |
| Sang et al. | Ensembles of gradient boosting recurrent neural network for time series data prediction | |
| Bhattacharyya et al. | Long term prediction of rainfall in Andhra Pradesh with Deep learning | |
| Saini et al. | Deep-LSTM Model for Wheat Crop Yield Prediction in India | |
| Al-Fawa’Reh et al. | Intelligent Methods for flood forecasting in Wadi al Wala, Jordan | |
| KR20230066927A (en) | Method and apparatus for predicting power generation | |
| CN114492945A (en) | Short-term photovoltaic power prediction method, medium and equipment in electric power market background | |
| CN113723670A (en) | Photovoltaic power generation power short-term prediction method with variable time window | |
| Adeyemo | Soft Computing techniques for weather and Climate change studies | |
| Hu et al. | Application of the Support Vector Machine on precipitation-runoff modelling in Fenhe River | |
| Mazhar et al. | Do we really have to consider data mining techniques for meteorological data | |
| Terzi | Monthly river flow forecasting by data mining process |
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 |