CN113379109B - Runoff forecasting method based on prediction model self-adaption - Google Patents

Runoff forecasting method based on prediction model self-adaption Download PDF

Info

Publication number
CN113379109B
CN113379109B CN202110584298.5A CN202110584298A CN113379109B CN 113379109 B CN113379109 B CN 113379109B CN 202110584298 A CN202110584298 A CN 202110584298A CN 113379109 B CN113379109 B CN 113379109B
Authority
CN
China
Prior art keywords
forecasting
sub
basin
value
forecast
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.)
Active
Application number
CN202110584298.5A
Other languages
Chinese (zh)
Other versions
CN113379109A (en
Inventor
刘佳
李传哲
王洋
于福亮
邱庆泰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Institute of Water Resources and Hydropower Research
Original Assignee
China Institute of Water Resources and Hydropower Research
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Institute of Water Resources and Hydropower Research filed Critical China Institute of Water Resources and Hydropower Research
Priority to CN202110584298.5A priority Critical patent/CN113379109B/en
Publication of CN113379109A publication Critical patent/CN113379109A/en
Application granted granted Critical
Publication of CN113379109B publication Critical patent/CN113379109B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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

Abstract

The invention relates to a self-adaptive drainage basin medium and long term runoff forecasting model architecture method, which aims to divide a forecasting drainage basin into different sub drainage basin subareas for forecasting according to the characteristics of the forecasting drainage basin; establishing forecasting factor sets of different sub-watersheds; adopting an ensemble forecasting method for each sub-basin; adjusting parameters of different models in the forecasting method through a self-adaptive method; adopting a river channel calculation method to obtain a final river basin forecast value for the forecast result of the sub-river basin; the deterministic coefficient of the forecasting result is periodically checked to judge whether the forecasting factor needs to be updated or not and the composition of the forecasting method, and the runoff forecasting result obtained by using the method can provide reliable basis for urban flood control or large-scale reservoir inflow forecasting.

Description

Runoff forecasting method based on prediction model self-adaption
Technical Field
The invention relates to a method for applying for the Chinese patent medicine, wherein the application is applied on 2017, 11 and 21, and the application numbers are as follows: 201711163861.1, entitled "a method for constructing adaptive basin middle and long term runoff forecasting model", and its application. The invention relates to a runoff forecasting architecture method, in particular to a runoff forecasting method based on prediction model self-adaption.
Background
The runoff forecasting belongs to the hydrologic forecasting category, is an important component of applied hydrology, is an applied scientific technology for forecasting future runoff change on the basis of mastering objective hydrologic rules, and is a premise for scientific implementation of water resource scheduling, water conservancy flood prevention and drought control. The runoff forecast can be divided into short-term runoff forecast and medium-and long-term runoff forecast according to forecast periods, generally, basin confluence time is taken as a boundary, the forecast period is called short-term forecast when the forecast period is less than or equal to the basin confluence time, and the forecast period is called medium-and long-term forecast when the forecast period is greater than the basin confluence time. The forecast period is a medium forecast in one day, the forecast period is more than one day, the forecast period is a long-term forecast in less than one year, and the forecast period is an ultra-long-term forecast in more than one year.
The medium-long term runoff forecasting is to qualitatively or quantitatively forecast the runoff state in a certain period in the future according to known information. Firstly, rainfall, sea surface temperature, an atmospheric circulation system and a weather system which affect a forecast object are analyzed, forecast factors with strong physical mechanisms and remarkable correlation are selected and determined, and whether the correlation is remarkable or not is calculated by applying a statistical calculation method. And establishing a corresponding forecasting model on the basis, and evaluating and preferably selecting the optimal forecasting model. And finally, analyzing the precision of the model through simulation and trial report of the model, and simultaneously carrying out uncertainty analysis on a deterministic forecast result to give a forecast result, thereby being applied to production practice.
Accurate medium and long term runoff forecasting is an important guarantee for improving the utilization rate of water resources, realizing optimal dispatching operation of a watershed hydropower station and improving the economic benefit of the hydropower station. Particularly, on the background of electric power market reform, the accuracy and the forecast period of medium-long term forecast are improved, a scientific and reasonable drainage basin cascade combined optimization power generation plan is compiled, and the method is particularly important for efficiently developing drainage basin cascade reservoir combined dispatching work.
With the continuous construction of large-scale hydraulic engineering of our country in recent years, play a huge role in relieving water resource shortage and improving ecological environment, runoff forecasting plays a vital role in scientifically, reasonably and efficiently scheduling water resources for the large-scale engineering, the economic benefit, the social benefit and the ecological environment benefit of the hydraulic engineering are fully played, and the realization of the engineering construction target is ensured. How to accurately forecast runoff is the primary subject in the operation, scheduling and management of various engineering water resources and is one of the key problems for determining the success or failure of various engineering. However, due to the influence of human activities, climate changes and other factors, the current forecasting system is in urgent need of improvement in coverage, forecasting factors and other aspects. Particularly, under the condition of changing the underlying surface, the current mechanisms of the runoff producing among all parts of the watershed are correspondingly changed, the simulation effect of the method for neglecting local differences is reduced, and in addition, the current combined forecasting method lacks self-adjustment of weight and cannot adjust and update the forecasting model according to the forecasting result.
Disclosure of Invention
The invention designs a self-adaptive drainage basin medium and long term runoff forecasting model architecture method, which solves the technical problems that the existing forecasting system is in urgent need of perfecting in the aspects of coverage range, forecasting factors and the like, different methods are not established aiming at the difference between medium term forecasting and long term forecasting in the forecasting process, different forecasting models are not established according to the characteristics of different sub drainage basins of a drainage basin, and meanwhile, the forecasting models lack self-evaluation and modification capability and cannot update model blocks in time.
In order to solve the technical problems, the invention adopts the following scheme:
1. a self-adaptive drainage basin medium and long term runoff forecasting model architecture method comprises the following steps:
step 1, collecting basic data of a forecast drainage basin;
step 2, based on the basic data, establishing a linear regression equation between the annual flow sequence x (t) and the time sequence t of the annual flow sequence by using a linear regression method, and further checking the trend of the time sequence;
step 3, dividing the drainage basin into a plurality of sub-drainage basins;
step 4, identifying a forecasting factor;
step 5, establishing a prediction model base;
step 6, judging through a certainty coefficient, selecting a model with a higher certainty coefficient for each sub-basin in a physical cause method, a hydrological statistical method and an artificial intelligence method to form the sub-basin ensemble forecast;
step 7, determining weight values of different methods of a physical cause method, a hydrological statistical method and artificial intelligence according to the result calculated in the step 6, and performing ensemble forecasting;
step 8, a minimum root mean square error algorithm is used for adjusting the weighted value in the step 7, so that the root mean square error between the predicted value and the measured value of each sub-basin is minimized, and the predicted value of each sub-basin is output;
step 9, carrying out river channel calculation to obtain the runoff process of the outlet section of the whole drainage basin according to the forecast value of each sub-drainage basin, and finishing the forecasting process;
step 10, calculating a certainty coefficient for the whole drainage basin forecast result;
step 11, carrying out trend inspection on the certainty coefficient of the day-by-day forecast result of the previous year on a certain day every month;
and step 12, judging whether the forecasting factor set needs to be updated and the forecasting model needs to be reselected.
Further, the basic data in step 1 includes:
basic data A, flow data of days, ten days, months and years of a river basin main control hydrology station, which are used in the step 2 and the step 11;
basic data B, characteristic values of maximum flow, minimum flow and occurrence time of each main control hydrological station in ten days, months and years, and process data of first runoff and last runoff, which are used in the step 3;
basic data C, daily, ten-day, monthly and annual rainfall of the main rainfall station of the drainage basin, which is used in the step 4;
basic data D, collection of 74 circulation indicators, and weather influence factors of numerical forecasting results and reanalysis data of the European middle weather forecasting center ECMWF or the national environmental prediction center NCEP, which are used in the step 4.
Further, in the step 2, a linear regression equation is constructed according to the foundation data a of the basin main control hydrological station provided in the step 1, wherein the linear regression equation gives whether the time series has an increasing or decreasing trend, and the linear regression equation is:
x(t)＝a×t+b；
in the formula: x (t) is a time sequence, t is a corresponding time sequence, a is the slope of a linear equation, the average trend change rate of the characteristic time sequence, and b is an intercept; the values of a and b can be estimated by a least squares method.
Further, the criterion for dividing the drainage basin into several sub-drainage basins in the step 3 is as follows: and 2, merging the sub-drainage basins with the same flow rate change trend of the control station, the same underlying surface condition and the same runoff generating mode of the sub-drainage basins at the upstream of each control station into one sub-drainage basin, and distinguishing different sub-drainage basins from one another, wherein the ten-day, month and year maximum flow rate, the minimum flow rate characteristic value and the occurrence time of each main control hydrological station in the basic data B in the step 1 are used as conditions for judging the underlying surface condition and the runoff generating mode.
Further, in the step 4, forecasting factors are respectively identified for different sub-watersheds, where the forecasting factors include: early precipitation and runoff, 74 items of circulation indexes, meteorological factor data, sea surface temperature, solar activity factors and human activity factors; the solar activity factor selects the relative number of sun black seeds, the associated geomagnetic index and the solar 10cm wave radio current as a forecasting factor; human activities are reflected by the hardened floor area of urban watertight and the dispatching rules of hydropower stations; the meteorological factor data come from the numerical forecast results and reanalysis data of the intermediate-oasis weather forecast center ECMWF or the national environmental forecast center NCEP in the basic data D in the step 1;
and analyzing the correlation degree between different forecasting factors and different sub-basin flows by adopting a correlation analysis method, wherein the calculation formula is as follows:
Figure GDA0003172727010000031
a correlation coefficient; n is the number of data samples; xiThe ith sample value of X;
Yithe ith sample value of Y;
Figure GDA0003172727010000032
sample mean of X;
Figure GDA0003172727010000033
sample mean of Y; x represents the flow of the outlet section of a certain sub-basin, Y represents a certain forecasting factor, and correlation relations between different forecasting factors and the flow of the outlet section of the sub-basin are respectively calculated;
coefficient of correlation RXYHas a value range of [ -1,1 [)]；RXYIf the value is more than 0, the forecasting object Y and the forecasting factor X are in positive correlation; rXYIf the value is less than 0, the forecasting object Y and the forecasting factor X are in negative correlation; rXYEqual to 0, specifies the forecast object Y and the forecastNo correlation between factors X; rXYThe larger the absolute value of (a), the higher the degree of correlation between the prediction object Y and the prediction factor X; for different sub-watersheds, selecting the forecasting factors with the correlation degree ranking at the top by 10 percent as the forecasting factor sets of the different sub-watersheds.
Further, the step 5 of establishing a prediction model library comprises three methods: the method comprises a physical cause method, a hydrological statistical method and an artificial intelligence method, wherein the physical cause method comprises a multivariate linear regression model and a multivariate threshold regression model, the hydrological statistical method comprises a time sequence decomposition model and a rank similarity prediction model, and the artificial intelligence model comprises an artificial neural network model and a support vector machine model.
Further, the deterministic coefficient formula used in step 6 is:
Figure GDA0003172727010000034
in the formula: DC is a deterministic coefficient, y0(i) Is a measured value of yc(i) In order to be able to predict the value,
Figure GDA0003172727010000035
the mean of the measured sequences, m is the length of the data sequence.
Further, the calculation formula in step 7 is as follows:
in step 6, the certainty coefficients of the physical cause method, the hydrological statistical method and the artificial intelligence are respectively A, B and C, and the weight of the simulation result of the physical cause method is
Figure GDA0003172727010000036
The weight of the hydrological statistical method is
Figure GDA0003172727010000037
The artificial intelligence method has the simulated weight of
Figure GDA0003172727010000038
The integration forecast value is:
R＝w1y1+w2y2+w3y3
in the formula, w1，w2，w3Is a weight value of y1，y2，y3For prediction of the value of each method, RiValues are predicted for each sub-basin set.
Further, in step 11, No. 1 a per month performs trend test on the certainty coefficient of the daily forecast result in the previous year (12 months), the test objects of the certainty coefficient are the daily forecast result of the basic data a in step 1 and the daily flow data of the main control hydrological station, a kandel rank correlation test method is adopted, and the calculation formula is as follows:
Figure GDA0003172727010000041
in the formula, U is a certainty coefficient;
Figure GDA0003172727010000042
n is the total length of the deterministic coefficient sequence, xi，xjFor the numerical values in the series, sgn is a sign function, if the number of the returned value is greater than 0, sgn returns 1, if the number is equal to 0, sgn returns 0, and if the number is less than 0, sgn returns-1, and the sign of the numerical parameter determines the returned value of the sgn function; i, j are the numbers of the numerical values in the series from 1 to n; n is the length of the series; τ is constant.
Further, in step 12, for the calculation result of step 11, if | U | > Uα/2When U is larger than 0, the change trend of the deterministic coefficient sequence is obvious, the deterministic coefficient sequence is in an ascending trend, the prediction result is good, and the prediction factor set does not need to be updated and the prediction model does not need to be reselected;
when | U | is greater than Uα/2When U is less than 0, the sequence is in a descending trend, the forecasting result is indicated to have a descending trend, at the moment, the step 4 is returned to re-identify the forecasting factor set, and the steps 6-10 are repeated;
wherein alpha is significance level, and U is obtained by looking up a normal distribution table according to the given significance levelα/2。
The adaptive drainage basin medium-long term runoff forecasting model architecture method has the following beneficial effects:
according to the method, the forecasting basin is divided into different sub-basin partitions for forecasting according to the characteristics of the forecasting basin; establishing forecasting factor sets of different sub-watersheds; adopting an ensemble forecasting method for each sub-basin; adjusting parameters of different models in the forecasting method through a self-adaptive method; adopting a river channel calculation method to obtain a final river basin forecast value for the forecast result of the sub-river basin; the deterministic coefficient of the forecasting result is periodically checked to judge whether the forecasting factor needs to be updated or not and the composition of the forecasting method, and the runoff forecasting result obtained by using the method can provide reliable basis for urban flood control or large-scale reservoir inflow forecasting.
Description of the drawings:
FIG. 1: the invention discloses a flow chart of a self-adaptive medium-and-long-term runoff forecasting model architecture method in a drainage basin.
Detailed Description
The invention will be further illustrated with reference to the following examples:
step 1, collecting basic data of a forecast drainage basin, which mainly comprises the following steps: (1) the watershed mainly controls the date, ten days, month and year flow data of the hydrological station; (2) maximum flow, minimum flow characteristic value and occurrence time of each control hydrological station in ten days, months and years, first runoff and last runoff process data; (3) rainfall in the day, ten days, month and year of the main rainfall station in the drainage basin; (4) collecting 74 circulation indexes and meteorological influence factors such as the ECMWF of the European middle-term weather forecasting center or the numerical forecasting result (reanalysis data) of the NCEP of the national environment forecasting center.
Step 2, for the main control hydrological station, a linear regression method is used for establishing a linear regression equation between the annual flow rate sequence x (t) and the time sequence t of the annual flow rate sequence x (t), and further the trend of the time sequence is checked, whether the time sequence has an increasing or decreasing trend can be given by the method, and the linear regression equation is as follows:
x(t)＝a×t+b
in the formula: x (t) is a time sequence, t is a corresponding time sequence, a is the slope of a linear equation and represents the average trend change rate of the time sequence, and b is an intercept. The values of a and b can be estimated by a least squares method.
Step 3, dividing the drainage basin into a plurality of sub-drainage basins, wherein the division basis mainly comprises that the flow change trend of the control station, the underlying surface condition and the flow production mode of the sub-drainage basins at the upstream of each control station which are judged in the step 2 are combined into one sub-drainage basin, and the sub-drainage basins with different conditions are mutually distinguished;
step 4, identifying the forecasting factors, namely identifying the forecasting factors respectively aiming at different sub-watersheds, wherein the forecasting factors mainly comprise: early precipitation and runoff, 74-item circulation indexes, sea surface temperature, solar activity factors, human activity factors and the like, correlation analysis methods are adopted to analyze the correlation degree between different forecasting factors and different sub-basin flows, and the calculation formula is as follows:
Figure GDA0003172727010000051
in the formula, RXYIs the correlation coefficient between X and Y; n is the number of data samples; xiThe ith sample value of X; y isiThe ith sample value of Y;
Figure GDA0003172727010000052
sample mean of X;
Figure GDA0003172727010000053
is the sample mean of Y.
Coefficient of correlation RXYHas a value range of [ -1,1 [)]。RXYIf the value is more than 0, the forecasting object Y and the forecasting factor X are in positive correlation; rXYIf the value is less than 0, the forecasting object Y and the forecasting factor X are in negative correlation; rXYEqual to 0, indicates no correlation between the predictor Y and the predictor X. RXYThe larger the absolute value of (a), the higher the degree of correlation between the prediction object Y and the prediction factor X. Selecting 10% forecasting factors with the top rank of the correlation degree as forecasting factor sets of different sub-watersheds for different sub-watersheds;
step 5, establishing a prediction model base, wherein the prediction model base mainly comprises three methods: the method comprises a physical cause method, a hydrological statistical method and an artificial intelligence method, wherein the physical cause method comprises a multivariate linear regression model and a multivariate threshold regression model, the hydrological statistical method comprises a time sequence decomposition model and a rank similarity prediction model, and the artificial intelligence model comprises an artificial neural network model and a support vector machine model;
step 6, judging through a certainty coefficient, selecting a model with a higher certainty coefficient for each sub-basin according to a physical cause method, a hydrological statistical method and an artificial intelligence method, and forming a sub-basin ensemble forecast, wherein the certainty coefficient formula is as follows:
Figure GDA0003172727010000054
in the formula: DC is a deterministic coefficient, y0(i) Is a measured value of yc(i) In order to be able to predict the value,
Figure GDA0003172727010000055
the mean of the measured sequences, m is the length of the data sequence.
And 7, determining the weights of different methods of the physical cause method, the hydrological statistical method and the artificial intelligence according to the result calculated in the step 6, and performing ensemble prediction, wherein if the deterministic coefficients of the physical cause method, the hydrological statistical method and the artificial intelligence in the step 6 are respectively A, B and C, the weight of the simulation result of the physical cause method is
Figure GDA0003172727010000056
The weight of the hydrological statistical method is human
Figure GDA0003172727010000057
In the formula, w1，w2，w3Is a weight value of y1，y2，y3For prediction of the value of each method, RiValues are predicted for each sub-basin set.
Step 8, adjusting w in step 7 by using minimum root mean square error algorithm1，w2，w3The error of the root mean square between the predicted value and the measured value of each sub-basin is minimized, and the predicted value is output. The minimum root mean square error algorithm can be implemented with the mat l ab program.
Step 9, carrying out river channel calculation to obtain a runoff process of the outlet section of the drainage basin according to the forecasting result of each sub-drainage basin to finish the forecasting process, wherein the river channel calculation can use a Maskin root method or a neural network method;
step 10, calculating a certainty coefficient for the result of the flow field forecast, wherein the formula is as in step 6;
step 11, carrying out trend test on the certainty coefficient of the day-by-day forecast result of the previous year (12 months) by No. 1 monthly, and adopting a Kandel rank correlation test method, wherein the calculation formula is as follows:
Figure GDA0003172727010000061
in the formula, U is a deterministic coefficient
Figure GDA0003172727010000062
N is the total length of the deterministic coefficient sequence, xi，xjFor values in the series sgn is a sign function, if the number of the return value is greater than 0, sgn returns 1, if the number is equal to 0, then 0 is returned, if the number is less than 0, then-1 is returned, and the sign of the number parameter determines the return value of the sgn function.
Step 12, aiming at step 11, judging whether the forecasting factor set needs to be updated or not and the forecasting model needs to be reselected, if the | U | is more than U |)α/2And when U is larger than 0, the change trend of the deterministic coefficient sequence is obvious, the deterministic coefficient sequence is in an ascending trend, the prediction result is better, and the prediction factor set does not need to be updated and the prediction model does not need to be reselected. When | U | is greater than Uα/2And when the U is less than 0, the sequence is in a descending trend, which indicates that the prediction result is in a descending trend, at the moment, the step 4 is returned to re-identify the prediction factor set, and the steps 6-10 are repeated.

Claims (4)

1. A runoff forecasting method based on prediction model self-adaptation comprises the following steps:
step 1, collecting basic data of a forecast drainage basin;
the basic data in the step 1 comprises:
basic data A, flow data of days, ten days, months and years of a hydrological station mainly controlled by a drainage basin are used in the step 2 and the step 11;
basic data B, characteristic values of maximum flow, minimum flow and occurrence time of each main control hydrological station in ten days, months and years, and process data of first runoff and last runoff, which are used in the step 3;
basic data C, daily, ten-day, monthly and annual rainfall of the main rainfall station of the drainage basin, which is used in the step 4;
basic data D, 74 circulation indexes are collected, and weather influence factors of numerical forecasting results and reanalysis data of the ECMWF (European middle weather forecasting center) or the NCEP (national environmental prediction center) are used in the step 4;
step 2, based on the basic data, establishing a linear regression equation between the annual flow sequence x (t) and the time sequence t of the annual flow sequence by using a linear regression method, and further checking the trend of the time sequence;
step 3, dividing the drainage basin into a plurality of sub-drainage basins;
step 4, identifying a forecasting factor;
in the step 4, forecasting factors are respectively identified for different sub-watersheds, and the forecasting factors include: early precipitation and runoff, 74 items of circulation indexes, meteorological factor data, sea surface temperature, solar activity factors and human activity factors; the solar activity factor selects the relative number of sun black seeds, the associated geomagnetic index and the solar 10cm wave radio current as a forecasting factor; the human activity factor is reflected by the hardened ground area of urban watertight and the dispatching rule of the hydropower station; the meteorological factor data come from the numerical forecast results and reanalysis data of the intermediate-oasis weather forecast center ECMWF or the national environmental forecast center NCEP in the basic data D in the step 1;
and analyzing the correlation degree between different forecasting factors and different forecasting objects by adopting a correlation analysis method, wherein the calculation formula is as follows:
Figure FDA0003536006950000021
in the formula, the correlation coefficient RXYIs the correlation coefficient between X and Y; n is the number of data samples; xiThe ith sample value of X; y isiThe ith sample value of Y;
Figure FDA0003536006950000022
sample mean of X;
Figure FDA0003536006950000023
sample mean of Y; x represents a forecasting factor, Y represents a forecasting object, the forecasting object is the flow of the outlet section of a certain sub-basin, and the correlation relations between different forecasting factors and different forecasting objects are respectively calculated;
coefficient of correlation RXYHas a value range of [ -1,1 [)]；RXYIf the value is more than 0, the forecasting object Y and the forecasting factor X are in positive correlation; rXYIf the value is less than 0, the forecasting object Y and the forecasting factor X are in negative correlation; rXYEqual to 0, indicating no correlation between the predictor Y and the predictor X; rXYThe larger the absolute value of (a), the higher the degree of correlation between the prediction object Y and the prediction factor X; selecting 10% forecasting factors with the top rank of the correlation degree as forecasting factor sets of different sub-watersheds for different sub-watersheds;
step 5, establishing a prediction model base;
the step 5 of establishing a prediction model library comprises three methods: the method comprises a physical cause method, a hydrological statistical method and an artificial intelligence method, wherein the physical cause method comprises a multivariate linear regression model and a multivariate threshold regression model, the hydrological statistical method comprises a time sequence decomposition model and a rank similarity prediction model, and the artificial intelligence model comprises an artificial neural network model and a support vector machine model;
step 6, judging through a certainty coefficient, selecting a model with a higher certainty coefficient for each sub-basin in a physical cause method, a hydrological statistical method and an artificial intelligence method to form the sub-basin ensemble forecast;
the deterministic coefficient formula used in step 6 is:
Figure FDA0003536006950000031
in the formula: DC is a deterministic coefficient, y0(i) Is a measured value of yc(i) In order to be able to predict the value,
Figure FDA0003536006950000032
the mean value of the measured sequence, m is the length of the data sequence;
step 7, determining weight values of different methods of a physical cause method, a hydrological statistical method and artificial intelligence according to the result calculated in the step 6, and performing ensemble forecasting;
step 8, a minimum root mean square error algorithm is used for adjusting the weighted value in the step 7, so that the root mean square error between the predicted value and the measured value of each sub-basin is minimized, and the predicted value of each sub-basin is output;
step 9, carrying out river channel calculation to obtain the runoff process of the outlet section of the whole drainage basin according to the forecast value of each sub-drainage basin, and finishing the forecasting process, wherein the river channel calculation can use a Masjing root method or a neural network method;
step 10, calculating a certainty coefficient for the whole drainage basin forecast result;
the deterministic coefficient formula used in step 10 is:
Figure FDA0003536006950000033
in the formula: DC is a deterministic coefficient, y0(i) Is a measured value of yc(i) In order to be able to predict the value,
Figure FDA0003536006950000034
as a mean of the measured sequenceM is the length of the data sequence;
step 11, carrying out trend inspection on the certainty coefficient of the day-by-day forecast result of the previous year on a certain day every month; in the step 11, No. 1 a per month performs trend test on the certainty coefficient of the daily forecast result in the previous year (12 months), the test objects of the certainty coefficient are the daily forecast result of the basic data a in the step 1 and the daily flow data of the main control hydrological station, a Kandel rank correlation test method is adopted, and the calculation formula is as follows:
Figure FDA0003536006950000035
in the formula, U is a certainty coefficient;
Figure FDA0003536006950000041
Figure FDA0003536006950000042
n is the total length of the deterministic coefficient sequence, xi，xjFor the numerical values in the series, sgn is a sign function, if the number is greater than 0, sgn returns 1, if the number is equal to 0, 0 is returned, if the number is less than 0, 1 is returned, and the sign of the numerical parameter determines the return value of the sgn function; i, j are the numbers of the numerical values in the series from 1 to n; n is the length of the series; τ is a constant;
step 12, judging whether the forecasting factor set needs to be updated and the forecasting model needs to be reselected;
in the step 12, for the calculation result of the step 11, if | U | > Uα/2When U is larger than 0, the change trend of the deterministic coefficient sequence is obvious, the deterministic coefficient sequence is in an ascending trend, the prediction result is good, and the prediction factor set does not need to be updated and the prediction model does not need to be reselected;
when | U | is greater than Uα/2When U is less than 0, the sequence is in a descending trend, the forecasting result is indicated to have a descending trend, at the moment, the step 4 is returned to re-identify the forecasting factor set, and the steps 6-10 are repeated;
wherein alpha is the significance level givenThe significance level of (1) is determined by looking up U through a normal distribution tableα/2。
2. The runoff forecasting method based on the prediction model self-adaptation as claimed in claim 1, wherein: in the step 2, a linear regression equation is constructed according to the basic data a of the basin main control hydrological station provided in the step 1, wherein the linear regression equation gives whether the time series has an increasing or decreasing trend, and the linear regression equation is as follows:
x(t)＝a×t+b；
in the formula: x (t) is a time sequence, t is a corresponding time sequence, a is the slope of a linear equation, the average trend change rate of the characteristic time sequence, and b is an intercept; the values of a and b can be estimated by a least squares method.
3. The runoff forecasting method based on the prediction model self-adaptation as claimed in claim 2, wherein: the standard for dividing the drainage basin into a plurality of sub-drainage basins in the step 3 is as follows: and 2, merging the sub-drainage basins with the same flow rate change trend of the control station, the same underlying surface condition and the same runoff generating mode of the sub-drainage basins at the upstream of each control station into one sub-drainage basin, and distinguishing different sub-drainage basins from one another, wherein the ten-day, month and year maximum flow rate, the minimum flow rate characteristic value and the occurrence time of each main control hydrological station in the basic data B in the step 1 are used as conditions for judging the underlying surface condition and the runoff generating mode.
4. The prediction model adaptive-based runoff forecasting method according to claim 2, wherein: the calculation formula in step 7 is as follows:
in step 6, the certainty coefficients of the physical cause method, the hydrological statistical method and the artificial intelligence are respectively A, B and C, and the weight of the simulation result of the physical cause method is
Figure FDA0003536006950000051
The weight of the hydrological statistical method is
Figure FDA0003536006950000052
The artificial intelligence method has the simulated weight of
Figure FDA0003536006950000053
The integration forecast value is:
R＝w1y1+w2y2+w3y3
in the formula, w1，w2，w3Is a weight value of y1，y2，y3For prediction of the value of each method, RiValues are predicted for each sub-basin set.
CN202110584298.5A 2017-11-21 2017-11-21 Runoff forecasting method based on prediction model self-adaption Active CN113379109B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110584298.5A CN113379109B (en) 2017-11-21 2017-11-21 Runoff forecasting method based on prediction model self-adaption

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711163861.1A CN107992961B (en) 2017-11-21 2017-11-21 Adaptive drainage basin medium and long term runoff forecasting model architecture method
CN202110584298.5A CN113379109B (en) 2017-11-21 2017-11-21 Runoff forecasting method based on prediction model self-adaption

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
CN201711163861.1A Division CN107992961B (en) 2017-11-21 2017-11-21 Adaptive drainage basin medium and long term runoff forecasting model architecture method

Publications (2)

Publication Number Publication Date
CN113379109A CN113379109A (en) 2021-09-10
CN113379109B true CN113379109B (en) 2022-04-19

Family

ID=62031803

Family Applications (3)

Application Number Title Priority Date Filing Date
CN202110584321.0A Active CN113379110B (en) 2017-11-21 2017-11-21 Medium-and-long-term runoff forecast result trend testing method
CN201711163861.1A Active CN107992961B (en) 2017-11-21 2017-11-21 Adaptive drainage basin medium and long term runoff forecasting model architecture method
CN202110584298.5A Active CN113379109B (en) 2017-11-21 2017-11-21 Runoff forecasting method based on prediction model self-adaption

Family Applications Before (2)

Application Number Title Priority Date Filing Date
CN202110584321.0A Active CN113379110B (en) 2017-11-21 2017-11-21 Medium-and-long-term runoff forecast result trend testing method
CN201711163861.1A Active CN107992961B (en) 2017-11-21 2017-11-21 Adaptive drainage basin medium and long term runoff forecasting model architecture method

Country Status (1)

Country Link
CN (3) CN113379110B (en)

Families Citing this family (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113379110B (en) * 2017-11-21 2022-08-05 中国水利水电科学研究院 Medium-and-long-term runoff forecast result trend testing method
CN108876021B (en) * 2018-05-31 2020-06-02 华中科技大学 Medium-and-long-term runoff forecasting method and system
CN108874750A (en) * 2018-06-22 2018-11-23 河海大学 A kind of Calculation of Area Rainfall uncertainty estimation method
CN109059875B (en) * 2018-06-28 2019-11-01 中国水利水电科学研究院 A method of drive perfect model to carry out moon scale Runoff Forecast
CN109523054A (en) * 2018-09-29 2019-03-26 中山大学 A kind of season Runoff Forecast selecting predictors method based on random walk
CN110610256B (en) * 2019-08-02 2022-05-17 华中科技大学 Runoff forecast level evaluation method considering forecast difficulty under different forecast situations
CN110459036B (en) * 2019-09-09 2022-05-17 四川省水利科学研究院 Mountain torrent early warning method based on deep learning
CN110619111B (en) * 2019-09-19 2020-04-28 中国水利水电科学研究院 Natural runoff series consistency correction method
CN110807475B (en) * 2019-10-16 2022-11-18 大连理工大学 Flood classification, identification and forecast method based on certainty coefficient
CN111461453B (en) * 2020-04-13 2021-05-04 中国水利水电科学研究院 Medium-and-long-term runoff ensemble forecasting method based on multi-model combination
CN111665575B (en) * 2020-05-27 2021-01-05 长江水利委员会水文局 Medium-and-long-term rainfall grading coupling forecasting method and system based on statistical power
CN112036604B (en) * 2020-07-29 2022-10-18 大连理工大学 Medium runoff forecasting method considering multiple time sequence process factors
CN112884209B (en) * 2021-01-29 2022-10-11 河海大学 Weather method and mathematical statistics method-based medium and long-term rainfall forecasting method
CN115358134B (en) * 2021-03-17 2023-04-21 南京工业职业技术大学 River basin medium-long term runoff prediction method based on space-time granulation data scene model
CN113254878B (en) * 2021-05-19 2022-05-17 中国电建集团昆明勘测设计研究院有限公司 Method for judging water temperature structure of reservoir in hydraulic and hydroelectric engineering
CN114489170B (en) * 2022-04-15 2022-06-21 天津航天和兴科技有限公司 System and method for predicting and adjusting temperature in sealed cabin
CN115481818B (en) * 2022-10-12 2023-05-30 大连理工大学 Medium-and-long-term runoff forecasting method and system based on time sequence decomposition
CN117132177B (en) * 2023-10-23 2024-01-30 长江三峡集团实业发展(北京)有限公司 Runoff forecasting model construction and runoff forecasting method based on multiple hypothesis test

Family Cites Families (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4146053B2 (en) * 1999-12-13 2008-09-03 富士電機システムズ株式会社 Flow prediction method in dam or river
EP1744185A1 (en) * 2005-07-13 2007-01-17 CIMA-Università degli Studi di Genova Method for estimating a variable geophysical field particularly for forecasting atmospheric precipitation
CN102289570B (en) * 2011-07-23 2015-02-25 浙江大学 Flood forecast method based on rainfall-runoff-flood routing calculation
US9181795B2 (en) * 2011-12-05 2015-11-10 Jehangir Framroze PUNTHAKEY Groundwater management system
CN102867106A (en) * 2012-08-14 2013-01-09 贵州乌江水电开发有限责任公司 Method and system for predicting short-term running water
KR20140103589A (en) * 2013-02-18 2014-08-27 아주대학교산학협력단 Flood estimation method using MAPLE forecasted precipitation data and apparatus thereof
CN104298841B (en) * 2013-07-16 2018-04-13 浙江贵仁信息科技股份有限公司 A kind of Flood Forecasting Method and system based on historical data
CN104281780B (en) * 2014-10-11 2016-03-23 水利部交通运输部国家能源局南京水利科学研究院 Linear resesvoir is detained and confluxes and net watershed (many sub basin) method of confluxing
CN105139093B (en) * 2015-09-07 2019-05-31 河海大学 Flood Forecasting Method based on Boosting algorithm and support vector machines
CN105912770A (en) * 2016-04-08 2016-08-31 中山大学 Real-time hydrologic forecasting system
CN105929466A (en) * 2016-04-14 2016-09-07 国家电网公司 Flood forecast method dealing with influence on human activities
CN106294932B (en) * 2016-07-27 2018-05-18 太原理工大学 The uncertain analysis method influenced of different change condition watershed runoffs
CN106875047B (en) * 2017-01-23 2021-03-16 国网湖南省电力公司 Reservoir basin runoff forecasting method and system
CN107292098A (en) * 2017-06-15 2017-10-24 河海大学 Medium-and Long-Term Runoff Forecasting method based on early stage meteorological factor and data mining technology
CN107274031A (en) * 2017-06-29 2017-10-20 华中科技大学 The hydrologic forecasting method and system of a kind of coupled neural network and VIC distributed hydrological model
CN113379110B (en) * 2017-11-21 2022-08-05 中国水利水电科学研究院 Medium-and-long-term runoff forecast result trend testing method

Also Published As

Publication number Publication date
CN107992961B (en) 2021-04-27
CN113379110A (en) 2021-09-10
CN113379109A (en) 2021-09-10
CN113379110B (en) 2022-08-05
CN107992961A (en) 2018-05-04

Similar Documents

Publication Publication Date Title
CN113379109B (en) Runoff forecasting method based on prediction model self-adaption
CN110619432B (en) Feature extraction hydrological forecasting method based on deep learning
CN107292098A (en) Medium-and Long-Term Runoff Forecasting method based on early stage meteorological factor and data mining technology
CN102867106A (en) Method and system for predicting short-term running water
CN102495937A (en) Prediction method based on time sequence
CN112801342A (en) Adaptive runoff forecasting method based on rainfall runoff similarity
CN113344305B (en) Rapid prediction method for rainstorm waterlogging event
CN114429053B (en) Basin scale WEFE system adaptability simulation optimization method
CN110728409B (en) Flood process type similarity mining and rapid prediction method
CN116485584B (en) Method and system for cooperative regulation and control of WEE of river with large bottom slope in alpine region
CN116070971A (en) Orderly flow regulation and control method and system for river and lake water system
CN110598352B (en) Drainage basin water supply forecasting method
CN112330065A (en) Runoff forecasting method based on basic flow segmentation and artificial neural network model
CN112215389A (en) Method for determining river environment flow process interval
CN111737853A (en) Low-impact development multi-target interval optimization configuration method based on SWMM model
CN110135652B (en) Long-term flood season runoff prediction method
Zhang et al. Projections of the characteristics and probability of spatially concurrent hydrological drought in a cascade reservoirs area under CMIP6
CN113807545A (en) River and lake ecological flow forecasting and early warning method based on deep learning and physical model
Wang et al. A statistical hydrological model for Yangtze river watershed based on stepwise cluster analysis
CN112036604B (en) Medium runoff forecasting method considering multiple time sequence process factors
CN110458722A (en) Flood interval prediction method based on multiple target random vector function connection network
CN115358587A (en) Regional multi-department collaborative infrastructure planning method and system
CN113836807B (en) River and lake ecological flow forecasting and early warning method based on entropy method and long-term and short-term memory neural network
CN113435630A (en) Basin hydrological forecasting method and system with self-adaptive runoff yield mode
Zhang et al. Calibration and uncertainty analysis of a hydrological model based on cuckoo search and the M-GLUE method

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