CN112819234A - Flood forecasting method and system considering initial value correction - Google Patents
Flood forecasting method and system considering initial value correction Download PDFInfo
- Publication number
- CN112819234A CN112819234A CN202110160897.4A CN202110160897A CN112819234A CN 112819234 A CN112819234 A CN 112819234A CN 202110160897 A CN202110160897 A CN 202110160897A CN 112819234 A CN112819234 A CN 112819234A
- Authority
- CN
- China
- Prior art keywords
- flood
- initial value
- stationary phase
- forecasting
- calculation
- 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
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 provides a flood forecasting method and a flood forecasting system considering initial value correction, wherein the method comprises the following steps of carrying out initial flood forecasting according to an established flood forecasting scheme: observing a flood forecasting process diagram and determining a stationary period; calculating a stationary phase error Err, and judging whether the stationary phase error Err is greater than a threshold eta; when the stationary phase error Err is larger than the threshold eta, performing initial value correction by adopting an optimization algorithm; and adopting the optimized initial value to carry out flood forecast. The flood forecasting method and system considering initial value correction judge whether the initial forecasting stationary phase error is larger than a threshold value, if so, an optimization algorithm is adopted to carry out initial value correction, and finally, the corrected initial value is adopted to carry out flood forecasting.
Description
Technical Field
The invention relates to the technical field of flood prediction, in particular to a flood forecasting method and system considering initial value correction.
Background
The flood forecasting precision is influenced by various factors, and the correction of the initial value of the state variable is a feasible method for improving the flood forecasting precision.
An article "flood forecasting method research based on initial value of state variable correction" on the laugh, the courage of laugh, the courage, the liushu, the glowing, and the liucovin at book 40, the book No. 3 of the hydrology magazine in 6 months of 2020 proposes an initial value of state variable correction method, and the method is applied to flood forecasting to improve forecasting accuracy. The method comprises the steps of establishing a target function by utilizing errors of actually measured flow and forecast flow in a stationary period, correcting an initial value of a state variable by adopting an optimization algorithm, and finally carrying out flood forecast by adopting the corrected initial value of the state variable. Because of the restriction limitations of dimension disaster, computer performance and the like, the method does not directly correct the initial values of all the computing units, but adopts a correction coefficient mode, a coefficient is given to each state variable, and the initial value of each unit is multiplied by the corresponding coefficient to serve as the corrected initial value. The method is an indirect correction mode, solves the problems of dimension disaster, computer performance constraint and the like to a certain extent, but also sacrifices the correction precision of the initial value.
Disclosure of Invention
In order to solve the technical problems, the invention provides a flood forecasting method and system considering initial value correction, which judge whether the initial forecasting stationary phase error is larger than a threshold value, if so, an optimization algorithm is adopted to carry out initial value correction, and finally, the corrected initial value is adopted to carry out flood forecasting.
A first object of the present invention is to provide a flood forecasting method considering initial value correction, including initial flood forecasting according to an established flood forecasting scheme, and further including the steps of:
step 1: observing a flood forecasting process diagram and determining a stationary period;
step 2: calculating a stationary phase error Err, and judging whether the stationary phase error Err is greater than a threshold eta;
and step 3: when the stationary phase error Err is larger than the threshold eta, performing initial value correction by adopting an optimization algorithm;
and 4, step 4: and adopting the optimized initial value to carry out flood forecast.
Preferably, the rule for determining the stationary period includes:
(1) the stationary phase is within the preheating phase;
(2) the stationary phase starting moment is the calculation starting moment;
(3) and selecting the time when the main rainfall process does not start and the adjacent measured flow rises by no more than 10% at the end of the stationary period.
In any of the above schemes, preferably, the stationary period is determined by using the concavity and convexity of the function, at the start time of the stationary period, the flow pattern is not convex or concave, the second order difference of the flow function is equal to 0, at the end time of the stationary period, the flow pattern becomes a concave function, and at this time, the second order difference of the flow function is greater than 0.
In any of the above aspects, preferably, the step of the method for determining the plateau period is as follows:
step 11: to smooth the flow process in the preheat period, a sequence Q of measured flows in the preheat period is appliedOtPerforming five-point moving average calculation, and recording as QOta，QOtaThe number of the elements is L;
step 12: calculating QOtaSecond order difference Q ofOta”，QOta"the number of elements is L-2;
step 13: the start time of the stationary period is the start time TB of the preheating period which is 1;
step 14: traverse QOta", when QOta"(li) ═ 0 and QOta"(li +1) ═ 0 and QOta”(li+2)>0 and (Q)Ota(li)-QOta(li-1))/QOta(li-1)<When 10%, it is the stationary phase end time TE ═ li, li ═ 1, 2.
In any of the above schemes, preferably, the stationary phase error is a difference between a measured flow and a predicted flow at each time interval in the stationary phase.
In any of the above schemes, preferably, the stationary phase error calculation uses a normalized root mean square error, and the calculation formula is:
wherein Err is the plateau error; qCtTo forecast flow; qOtThe measured flow is obtained; qAOThe measured flow is the average value of the measured flow; n is the length of the data sequence in the stationary period, i is 1,2, …, n.
In any of the above aspects, it is preferable that the method for determining the threshold η includes the sub-steps of:
step 21: and selecting historical flood. Selecting more than 20 representative floods during selection, wherein the total field number is represented by N;
step 22: respectively carrying out simulated flood forecasting on each flood, taking whether the flood peak/flood volume error is more than 20% as the standard whether the flood is qualified, if so, determining that the flood is not qualified, and if not, determining that the flood is qualified;
step 23: determining the stationary phase of each flood;
step 24: calculating the error of each flood stationary phase;
step 25: form a data set (x)lj,ylj)，xljIs a plateau error; y isljIf the product is qualified, 100 is used for qualification, 0 is used for disqualification, and lj is 1,2, …, N;
step 26: according to xljThe data sets are sorted and drawn from small to large (x)lj,ylj) A relationship graph;
step 27: observing and analyzing the relationship diagram, when xljAnd if the number of unqualified flood fields exceeds 60%, the value is the threshold eta.
In any of the above schemes, preferably, the initial value correction method includes the following sub-steps:
step 31: dividing the drainage basin into m calculation units, respectively calculating the forecast flow of each unit by adopting a forecast model, and superposing to obtain the forecast result of the drainage basin;
step 32: establishing a target function BO by taking the actually measured flow and the forecast flow in the stationary period as objects;
step 33: and (3) adjusting the initial value within the initial value range by adopting an optimization algorithm to perform flood forecast calculation, calculating the objective function value after each adjustment, and considering the searched initial value of the state variable as an optimal value when the objective function value is smaller than a termination condition.
In any of the above schemes, preferably, there are k initial values of the state variables of the prediction model, and therefore, the total number of the initial values of the m calculation units is m × k, and the initial values are expressed in a matrix form:
wherein, wi，jRepresents the j (1. ltoreq. j. ltoreq.k) th initial value of the i (1. ltoreq. i.ltoreq.m) th cell.
In any of the above schemes, preferably, the calculation formula of the objective function BO is:
in any of the above aspects, preferably, the termination conditions are of the types:
1) the difference between the two adjacent calculation target functions is less than epsilon;
2) the optimizing calculation times are less than a certain given value;
3) the optimizing calculation time is less than a given value.
A second object of the present invention is to provide a flood forecasting system considering initial value correction, which includes an initial forecasting module, and further includes the following modules:
a stationary phase determination module: the method is used for observing a flood forecasting process diagram and determining a stationary phase;
a calculation module: the method is used for calculating the stationary phase error Err and judging whether the stationary phase error Err is larger than a threshold eta;
a correction module: the method is used for performing initial value correction by adopting an optimization algorithm when the stationary phase error Err is larger than the threshold eta;
an output module: the initial value after the optimization is used for carrying out flood forecasting;
the system performs flood forecasting with initial value correction taken into account in the method according to the first object.
Preferably, the rule for determining the stationary period includes:
(1) the stationary phase is within the preheating phase;
(2) the stationary phase starting moment is the calculation starting moment;
(3) and selecting the time when the main rainfall process does not start and the adjacent measured flow rises by no more than 10% at the end of the stationary period.
In any of the above schemes, preferably, the stationary period is determined by using the concavity and convexity of the function, at the start time of the stationary period, the flow pattern is not convex or concave, the second order difference of the flow function is equal to 0, at the end time of the stationary period, the flow pattern becomes a concave function, and at this time, the second order difference of the flow function is greater than 0.
In any of the above aspects, preferably, the step of the method for determining the plateau period is as follows:
step 11: to smooth the flow process in the preheat period, a sequence Q of measured flows in the preheat period is appliedOtPerforming five-point moving average calculation, and recording as QOta，QOtaThe number of the elements is L;
step 12: calculating QOtaSecond order difference Q ofOta”，QOta"the number of elements is L-2;
step 13: the start time of the stationary period is the start time TB of the preheating period which is 1;
step 14: traverse QOta", when QOta"(li) ═ 0 and QOta"(li +1) ═ 0 and QOta”(li+2)>0 and (Q)Ota(li)-QOta(li-1))/QOta(li-1)<When 10%, it is the stationary phase end time TE ═ li, li ═ 1, 2.
In any of the above schemes, preferably, the stationary phase error is a difference between a measured flow and a predicted flow at each time interval in the stationary phase.
In any of the above schemes, preferably, the stationary phase error calculation uses a normalized root mean square error, and the calculation formula is:
wherein Err is the plateau error; qCtTo forecast flow; qOtThe measured flow is obtained; qAOThe measured flow is the average value of the measured flow; n is the length of the data sequence in the stationary period, i is 1,2, …, n.
In any of the above aspects, it is preferable that the method for determining the threshold η includes the sub-steps of:
step 21: and selecting historical flood. Selecting more than 20 representative floods during selection, wherein the total field number is represented by N;
step 22: respectively carrying out simulated flood forecasting on each flood, taking whether the flood peak/flood volume error is more than 20% as the standard whether the flood is qualified, if so, determining that the flood is not qualified, and if not, determining that the flood is qualified;
step 23: determining the stationary phase of each flood;
step 24: calculating the error of each flood stationary phase;
step 25: form a data set (x)lj,ylj)，xljIs a plateau error; y isljIf the product is qualified, 100 is used for qualification, 0 is used for disqualification, and lj is 1,2, …, N;
step 26: according to xljThe data sets are sorted and drawn from small to large (x)lj,ylj) A relationship graph;
step 27: observing and analyzing the relationship diagram, when xljAnd if the number of unqualified flood fields exceeds 60%, the value is the threshold eta.
In any of the above schemes, preferably, the initial value correction method includes the following sub-steps:
step 31: dividing the drainage basin into m calculation units, respectively calculating the forecast flow of each unit by adopting a forecast model, and superposing to obtain the forecast result of the drainage basin;
step 32: establishing a target function BO by taking the actually measured flow and the forecast flow in the stationary period as objects;
step 33: and (3) adjusting the initial value within the initial value range by adopting an optimization algorithm to perform flood forecast calculation, calculating the objective function value after each adjustment, and considering the searched initial value of the state variable as an optimal value when the objective function value is smaller than a termination condition.
In any of the above schemes, preferably, there are k initial values of the state variables of the prediction model, and therefore, the total number of the initial values of the m calculation units is m × k, and the initial values are expressed in a matrix form:
wherein, wi，jRepresents the j (1. ltoreq. j. ltoreq.k) th initial value of the i (1. ltoreq. i.ltoreq.m) th cell.
In any of the above schemes, preferably, the calculation formula of the objective function BO is:
in any of the above aspects, preferably, the termination conditions are of the types:
1) the difference between the two adjacent calculation target functions is less than epsilon;
2) the optimizing calculation times are less than a certain given value;
3) the optimizing calculation time is less than a given value.
The invention provides a flood forecasting method and system considering initial value correction, which directly correct the initial values of all computing units, so that the accuracy of initial value correction is improved, and the accuracy of flood forecasting is further improved. Meanwhile, the improved method cancels the calculation of the stationary phase deviation and improves the efficiency.
Drawings
Fig. 1 is a flowchart of a preferred embodiment of a flood forecasting method considering initial value correction according to the present invention.
Fig. 2 is a block diagram of a preferred embodiment of the flood forecasting system considering initial value correction according to the present invention.
Fig. 3 is a diagram illustrating the results of an embodiment of flood forecasting according to the flood forecasting method considering initial value correction of the present invention.
Fig. 4 is a flowchart of an embodiment of a threshold determination method of a flood forecasting method considering initial value correction according to the present invention.
Fig. 5 is a flow chart of another preferred embodiment of the flood forecasting method considering initial value correction according to the present invention.
Fig. 6 is a schematic diagram of an embodiment of the stationary phase outcome of the flood forecasting method considering initial value correction according to the present invention.
Fig. 7 is a comparative diagram of an embodiment of a flood forecasting method considering initial value correction before and after flood correction according to the present invention.
Detailed Description
The invention is further illustrated with reference to the figures and the specific examples.
Example one
As shown in fig. 1 and 2, in a flood forecasting method considering initial value correction, step 100 is executed, and a primary forecasting module 200 performs primary forecasting of flood according to an established flood forecasting scheme.
In step 110, the stationary phase determining module 210 observes the flood forecasting process map and determines the stationary phase. The rules for determining the stationary phase include: (1) the stationary phase is within the preheating phase; (2) the stationary phase starting moment is the calculation starting moment; (3) and selecting the time when the main rainfall process does not start and the adjacent measured flow rises by no more than 10% at the end of the stationary period.
And determining the stationary period by utilizing the concave-convex property of the function, wherein at the beginning moment of the stationary period, the flow graph is not convex or concave, the second-order difference of the flow function is equal to 0, at the ending moment of the stationary period, the flow graph is changed into a concave function, and at the moment, the second-order difference of the flow function is greater than 0. The method for determining the stationary phase comprises the following steps:
step 11: to smooth the flow process during the preheat period, the preheat period is adjustedInner measured flow rate sequence QOtPerforming five-point moving average calculation, and recording as QOta，QOtaThe number of the elements is L;
step 12: calculating QOtaSecond order difference Q ofOta”，QOta"the number of elements is L-2;
step 13: the start time of the stationary period is the start time TB of the preheating period which is 1;
step 14: traverse QOta", when QOta"(li) ═ 0 and QOta"(li +1) ═ 0 and QOta”(li+2)>0 and (Q)Ota(li)-QOta(li-1))/QOta(li-1)<When 10%, it is the stationary phase end time TE ═ li, li ═ 1, 2.
In step 120, the calculation module 220 calculates the stationary phase error Err. The stationary phase error is the difference value between the measured flow and the forecast flow at each time interval in the stationary phase. The error calculation of the stationary phase adopts a normalized root mean square error, and the calculation formula is as follows:
wherein Err is the plateau error; qCtTo forecast flow; qOtThe measured flow is obtained; qAOThe measured flow is the average value of the measured flow; n is the length of the data sequence in the stationary period, i is 1,2, …, n.
In step 130, the calculating module 220 determines whether the stationary period error Err is greater than a threshold η. The method for determining the threshold η comprises the following sub-steps: step 31: and selecting historical flood. Selecting more than 20 representative floods during selection, wherein the total field number is represented by N; step 32: respectively carrying out simulated flood forecasting on each flood, taking whether the flood peak/flood volume error is more than 20% as the standard whether the flood is qualified, if so, determining that the flood is not qualified, and if not, determining that the flood is qualified; step 33: determining the stationary phase of each flood; step 34: calculating the error of each flood stationary phase; step 35: form a data set (x)lj,ylj)，xljIs a plateau error; y isljIf the product is qualified, 100 is used for qualification, 0 is used for disqualification, and lj is 1,2, …, N; step 36: according to xljThe data sets are sorted and drawn from small to large (x)lj,ylj) A relationship graph; step 37: observing and analyzing the relationship diagram, when xljAnd if the number of unqualified flood fields exceeds a certain value and the unqualified flood fields exceed 60%, the value is the threshold eta.
When the stationary phase error Err is not greater than the threshold η, step 135 is executed, without correcting the initial value, using the original initial value to perform flood forecasting.
When the stationary phase error Err is greater than the threshold η, step 140 is executed, and the correction module 230 is configured to perform initial value correction by using an optimization algorithm. The initial value correction method comprises the following substeps:
step 141: and dividing the watershed into m calculation units, respectively calculating the forecast flow of each unit by adopting a forecast model, and superposing to obtain a forecast result of the watershed. The initial values of the state variables of the forecasting model are k, and the initial values of the m computing units are m multiplied by k in total and are expressed by adopting a matrix form:
wherein, wi，jRepresents the j (1. ltoreq. j. ltoreq.k) th initial value of the i (1. ltoreq. i.ltoreq.m) th cell.
Step 142: and establishing a target function BO by taking the actually measured flow and the forecast flow in the stationary period as objects. The calculation formula of the objective function BO is as follows:
step 143: and (3) adjusting the initial value within the initial value range by adopting an optimization algorithm to perform flood forecast calculation, calculating the objective function value after each adjustment, and considering the searched initial value of the state variable as an optimal value when the objective function value is smaller than a termination condition. The types of the termination condition are: 1) the difference between the two adjacent calculation target functions is less than epsilon; 2) the optimizing calculation times are less than a certain given value; 3) the optimizing calculation time is less than a given value.
And executing step 150, and using the optimized initial value to perform flood forecasting by the output module 240.
Example two
A method principle of a flood forecasting method considering initial value correction is as follows: firstly, performing primary forecasting, judging whether the error of the primary forecasting stationary phase is larger than a threshold value, if so, adopting an optimization algorithm to correct the initial value, and finally adopting the corrected initial value to perform flood forecasting.
1. Stationary phase
The stationary period is the stage that only sporadic rainfall occurs in the early stage of each flood, the main rainfall does not start yet, the flow is stable, and no obvious rising occurs. In the flood forecast model calculation, the stationary phase flow comes from the initial value. The flow rate in the stationary phase is directly reflected by the initial value, and if the forecast in the stationary phase is accurate, the initial value is accurate, otherwise, the initial value is inaccurate.
A typical flood forecast production map is shown in fig. 3. According to the forecasting time, the forecasting process is divided into a preheating period and a forecasting period, the preheating period is called from the calculation starting time to the forecasting time, and the forecasting period is called from the forecasting time to the calculation ending time. Measured rainfall, flow and forecast flow exist in the preheating period; only the forecasted flow is in the forecast period.
Stationary phase determination principle:
(1) the stationary phase is within the preheating phase;
(2) the stationary phase starting moment is the calculation starting moment;
(3) and selecting the time when the main rainfall process does not start and the adjacent measured flow rises by no more than 10% at the end of the stationary period.
The stationary phase determining method comprises the following steps:
determining the stationary phase by utilizing the concave-convex property of the function. At the beginning of the stationary period, the flow graph is not convex or concave, the second order difference of the flow function is equal to 0, at the end of the stationary period, the flow graph is changed into a concave function, and at the moment, the second order difference of the flow function is larger than 0. The calculation steps are as follows:
(1) to smooth the flow process in the preheat period, a sequence Q of measured flows in the preheat period is appliedOtPerforming five-point moving average calculation, and recording as QOta，QOtaThe number of the elements is L;
(2) calculating QOtaSecond order difference Q ofOta”，QOta"the number of elements is L-2;
(3) the start time of the stationary period is the start time TB of the preheating period which is 1;
(4) traverse QOta", when QOta"(li) ═ 0 and QOta"(li +1) ═ 0 and QOta”(li+2)>0 and (Q)Ota(li)-QOta(li-1))/QOta(li-1)<When 10%, it is the stationary phase end time TE ═ li, li ═ 1, 2.
Stationary phase error:
the stationary phase error is the difference between the measured flow and the forecast flow at each time interval in the stationary phase.
The stationary phase error calculation adopts a Normalized Root Mean Square Error (NRMSE), and the calculation formula is as follows:
in the formula: err is plateau error; qCtTo forecast flow; qOtThe measured flow is obtained; qAOThe measured flow is the average value of the measured flow; n is the length of the data sequence in the stationary phase.
2. Threshold value
Due to the particularity of flood forecasting, errors inevitably exist, when the flood peak/flood volume errors are larger than a given standard, the flood forecasting of the field is unqualified, and the method provided by the patent only corrects unqualified flood. According to the research on multi-field flood, the stationary phase error and whether the field flood is qualified or not have an obvious correlation, the smaller the stationary phase error is, the higher the qualified probability of the field flood is, and conversely, the larger the stationary phase error is, the lower the qualified probability of the field flood is. And when the stationary phase error is larger than a certain value, the qualified probability of the field flood is obviously reduced, the value is set as a stationary phase threshold eta, when the stationary phase error is larger than the threshold eta, the initial value of the field flood is corrected, otherwise, the initial value of the field flood is not corrected.
The threshold determination method is as follows, and the flow is shown in fig. 4:
(1) and selecting historical flood. Selecting representative flood of more than 20 fields, wherein the total field is represented by N;
(2) respectively carrying out simulated flood forecasting on each flood, taking whether the flood peak/flood volume error is more than 20% as the standard whether the flood is qualified, if so, determining that the flood is not qualified, and if not, determining that the flood is qualified;
(3) determining the stationary phase of each flood;
(4) calculating the error of each flood stationary phase;
(5) form a data set (x)lj,ylj)，xljIs a plateau error; y isljIf the product is qualified, 100 is used for qualification, 0 is used for disqualification, and lj is 1,2, …, N;
(6) according to xljThe data sets are sorted and drawn from small to large (x)lj,ylj) A relationship graph;
(7) observing and analyzing the relationship diagram, when xljAnd if the number of unqualified flood fields exceeds 60%, the value is the threshold eta.
3. Method step
The calculation steps of a flood forecasting method considering initial value correction are as follows, and the flow is shown in fig. 5.
(1) Performing initial flood forecast (the initial value can be obtained by manual setting or continuous calculation by adopting a forecast model) according to the established flood forecast scheme;
(2) observing a flood forecasting process diagram and determining a stationary period;
(3) calculating a stationary phase error Err;
(4) and when the error Err in the stationary period is greater than the threshold eta, performing initial value correction, otherwise, not correcting.
(5) When flood forecasting is performed, a drainage basin is generally divided into m computing units, forecasting flow of each unit is respectively computed by adopting a forecasting model, and forecasting results of the drainage basin are obtained by superposition. The initial values of the state variables of the forecasting model are k, and therefore, the total number of the initial values of the m computing units is m multiplied by k, and the initial values are expressed in a matrix form:
wi，jrepresents the j (1. ltoreq. j. ltoreq.k) th initial value of the i (1. ltoreq. i.ltoreq.m) th cell.
(6) Taking the measured flow and the forecast flow in the stationary period as objects, establishing an objective function:
in the formula QCtTo forecast flow; qOtThe measured flow is obtained; n is the data sequence length.
(7) And (3) adjusting the initial value within the initial value range by adopting an optimization algorithm to perform flood forecast calculation, calculating the objective function value after each adjustment, and considering the searched initial value of the state variable as an optimal value when the objective function value is smaller than a termination condition. The types of termination conditions are:
1) the difference between two adjacent calculation target functions is less than epsilon, and epsilon is constant and can be set to 10-5Or 10-6And the like.
2) The number of optimizing calculations is less than a given value, which may be set to 50, 100, 150, etc.
3) The optimization calculation time is less than a given value and can be set to 60s, 90s, 120s and the like.
(8) The optimization algorithm is adopted for optimization calculation, and assignment of initial conditions of the optimization algorithm is random, and the phenomenon of 'different-parameter and same-effect' of initial values (different-parameter and same-effect: flood forecast calculation results caused by different initial value combinations are the same). This will cause the result of each optimization calculation to be different, and the method adopted in this patent is to perform multiple (10, 20, etc.) optimization calculations, and find out the calculation result of the one time that minimizes the objective function value as the final optimization calculation result.
(9) And (4) adopting the optimized initial value to carry out flood forecasting, wherein the forecasting result value is the final forecasting result.
EXAMPLE III
The process herein provides two improvements over the prior art.
(1) The stationary period of the method is determined by manual observation and is improved into automatic discrimination determination, the improvement overcomes the randomness of manual experience discrimination, and the intelligence of the existing method is improved;
(2) when the initial value correction is carried out, the original correction coefficient is modified into the direct correction initial value matrix W by using the coefficient to correct the initial value. The improvement omits the steps of calculating the deviation U by the existing method, and then correcting the coefficient according to the value range of the judgment coefficient of the deviation U, thereby simplifying the calculation steps of the existing method.
(3) A certain watershed is divided into 4 computing units, and a flood forecasting scheme is compiled by adopting a Xinanjiang model. 63 flood data from 1992 and 2008 were selected to determine the stationary threshold η. Firstly, a stationary period determining method is used for determining a stationary period, a stationary period error is calculated, and then a threshold determining method is used for determining that a stationary period threshold eta is 0.58.
The Xinanjiang model has 7 initial state variables, namely an upper-layer tension water storage WU, a lower-layer tension water storage WL, a deep-layer tension water storage WD, a runoff area ratio FR, a free water storage S, an interflow runoff depth QI and an underground runoff depth QG, and the initial value ranges are shown in the following table 1:
TABLE 1 Xinanjiang model State variable value range table
The drainage basin is divided into 4 computing units, and the initial value matrix is as follows:
and selecting a flood of the drainage basin to carry out initial value correction and forecast again. According to the stationary period determination method, the start time TB of the stationary period is 1, and the end time TE is 6, as shown in fig. 6.
And calculating that the initial forecast stationary phase error of the flood in the field is 0.63 and is greater than a threshold eta, and performing initial value correction for the factors. Selecting Particle Swarm Optimization (PSO) to perform initial value optimization calculation, and setting the iteration termination condition as that the difference between two calculated objective functions is less than epsilon 10-5The optimization iteration calculation times are 100 times, and the optimization iteration calculation time is 90 s; the total number of calculations is 10, and the calculation result in which the objective function value is minimized once is taken as the final preferred initial value. The results of the calculations for the prior art method and the method herein before and after the initial value correction are shown in table 2 below:
TABLE 2 initial value correction and comparison table
As can be seen from table 2, since the coefficient correction is adopted in the conventional method, the initial values of the corrected units have a plurality of same values, for example, WU of units 1,2 and 3 is 12, WD of units 2, 3 and 4 is 50.2, FR of units 2 and 3 is 0.8, FR of units 1 and 4 is 0.6, and S of units 3 and 4 is 6.8. The method directly corrects each initial value of each unit, so that the initial values are different, and correction results are more accurate.
And (3) performing flood forecast calculation again on the initial values corrected by the two methods respectively, wherein the error statistics of flood forecast results are shown in a table 3, and the result pair is shown in a table 7.
TABLE 3 statistical table of correction results
As can be seen from table 3, the correction result of the method is better than that of the existing method in terms of flood peak and flood volume, and the certainty coefficient is close to that of the existing method, which indicates that the correction method has higher correction accuracy. .
For a better understanding of the present invention, the foregoing detailed description has been given in conjunction with specific embodiments thereof, but not with the intention of limiting the invention thereto. Any simple modifications of the above embodiments according to the technical essence of the present invention still fall within the scope of the technical solution of the present invention. In the present specification, each embodiment is described with emphasis on differences from other embodiments, and the same or similar parts between the respective embodiments may be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
Claims (10)
1. A flood forecasting method considering initial value correction comprises the steps of carrying out initial flood forecasting according to an established flood forecasting scheme, and is characterized by further comprising the following steps of:
step 1: observing a flood forecasting process diagram and determining a stationary period;
step 2: calculating a stationary phase error Err, and judging whether the stationary phase error Err is greater than a threshold eta;
and step 3: when the stationary phase error Err is larger than the threshold eta, performing initial value correction by adopting an optimization algorithm;
and 4, step 4: and adopting the optimized initial value to carry out flood forecast.
2. A flood forecasting method taking into account initial value modifications, as claimed in claim 1, characterized in that said stationary phase determination rule satisfies the following condition:
(1) the stationary phase is within the preheating phase;
(2) the stationary phase starting moment is the calculation starting moment;
(3) and selecting the time when the main rainfall process does not start and the adjacent measured flow rises by no more than 10% at the end of the stationary period.
3. The flood forecasting method considering initial value correction according to claim 2, wherein the stationary phase error is a difference between a measured flow rate and a forecasted flow rate at each time interval in a stationary phase.
4. A flood forecasting method taking into account initial value modification, as claimed in claim 3, characterized in that the stationary phase error calculation uses normalized root mean square error, the calculation formula being:
wherein Err is the plateau error; qCtTo forecast flow; qOtThe measured flow is obtained; qAOThe measured flow is the average value of the measured flow; n is the length of the data sequence in the stationary period, i is 1,2, …, n.
5. A flood forecasting method taking into account initial value modifications, as claimed in claim 4, characterized in that said threshold η is determined by means of the following sub-steps:
step 21: and selecting historical flood. Selecting representative flood of more than 20 fields during selection, wherein the total field number is represented by N;
step 22: respectively carrying out simulated flood forecasting on each flood, taking whether the flood peak/flood volume error is more than 20% as the standard whether the flood is qualified, if so, determining that the flood is not qualified, and if not, determining that the flood is qualified;
step 23: determining the stationary phase of each flood;
step 24: calculating the error of each flood stationary phase;
step 25: form a data set (x)lj,ylj)，xljIs a plateau error; y isljIf the product is qualified, 100 is used for qualification, 0 is used for disqualification, and lj is 1,2, …, N;
step 26: according to xljThe data sets are sorted and drawn from small to large (x)lj,ylj) A relationship graph;
step 27: observing and analyzing the relationship diagram, when xljAnd if the number of unqualified flood fields exceeds 60%, the value is the threshold eta.
6. A flood forecasting method considering initial value modification according to claim 5, characterized in that the method of initial value modification comprises the sub-steps of:
step 31: dividing the drainage basin into m calculation units, respectively calculating the forecast flow of each unit by adopting a forecast model, and superposing to obtain the forecast result of the drainage basin;
step 32: establishing a target function BO by taking the actually measured flow and the forecast flow in the stationary period as objects;
step 33: and (3) adjusting the initial value within the initial value range by adopting an optimization algorithm to perform flood forecast calculation, calculating the objective function value after each adjustment, and considering the searched initial value of the state variable as an optimal value when the objective function value is smaller than a termination condition.
7. A flood forecasting method taking into account initial value modifications, as claimed in claim 6, characterized in that the initial values of the state variables of the forecasting model are k, and thus, the total number of the initial values of the state variables of the m calculation units is m x k, and they are expressed in the form of a matrix:
wherein, wi，jRepresents the j (1. ltoreq. j. ltoreq.k) th initial value of the i (1. ltoreq. i.ltoreq.m) th cell.
8. The flood forecasting method considering initial value modification of claim 7, wherein the objective function BO is calculated by the formula:
9. a flood forecasting method taking into account initial value modifications, as claimed in claim 8, characterized in that said termination conditions are of the type:
1) the difference between the two adjacent calculation target functions is less than epsilon;
2) the optimizing calculation times are less than a certain given value;
3) the optimizing calculation time is less than a given value.
10. A flood forecasting system considering initial value modification, comprising an initial forecasting module, characterized by further comprising the following modules:
a stationary phase determination module: the method is used for observing a flood forecasting process diagram and determining a stationary phase;
a calculation module: the method is used for calculating the stationary phase error Err and judging whether the stationary phase error Err is larger than a threshold eta;
a correction module: the method is used for performing initial value correction by adopting an optimization algorithm when the stationary phase error Err is larger than the threshold eta;
an output module: the initial value after the optimization is used for carrying out flood forecasting;
the system performs flood forecasting taking into account initial value corrections according to the method of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110160897.4A CN112819234B (en) | 2021-02-05 | 2021-02-05 | Flood forecasting method and system considering initial value correction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110160897.4A CN112819234B (en) | 2021-02-05 | 2021-02-05 | Flood forecasting method and system considering initial value correction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112819234A true CN112819234A (en) | 2021-05-18 |
| CN112819234B CN112819234B (en) | 2022-10-14 |
Family
ID=75861676
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110160897.4A Active CN112819234B (en) | 2021-02-05 | 2021-02-05 | Flood forecasting method and system considering initial value correction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112819234B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114612801A (en) * | 2022-05-10 | 2022-06-10 | 环球数科集团有限公司 | Flood early warning method based on high-resolution remote sensing satellite geometric positioning model |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160255112A1 (en) * | 2012-02-23 | 2016-09-01 | Markport Limited | Message flooding prevention in messaging networks |
| CN107341346A (en) * | 2017-06-26 | 2017-11-10 | 华中科技大学 | A kind of hydrologic forecasting method |
| CN107688702A (en) * | 2017-08-24 | 2018-02-13 | 西安理工大学 | A kind of river flood traffic camouflaging rule analogy method based on wolf pack algorithm |
| CN109710879A (en) * | 2017-08-17 | 2019-05-03 | 中国水利水电科学研究院 | A kind of optimized treatment method and device of forecast system of controlling flood |
-
2021
- 2021-02-05 CN CN202110160897.4A patent/CN112819234B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160255112A1 (en) * | 2012-02-23 | 2016-09-01 | Markport Limited | Message flooding prevention in messaging networks |
| CN107341346A (en) * | 2017-06-26 | 2017-11-10 | 华中科技大学 | A kind of hydrologic forecasting method |
| CN109710879A (en) * | 2017-08-17 | 2019-05-03 | 中国水利水电科学研究院 | A kind of optimized treatment method and device of forecast system of controlling flood |
| CN107688702A (en) * | 2017-08-24 | 2018-02-13 | 西安理工大学 | A kind of river flood traffic camouflaging rule analogy method based on wolf pack algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 李匡 等: "基于状态变量初值修正的洪水预报方法研究", 《水文》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114612801A (en) * | 2022-05-10 | 2022-06-10 | 环球数科集团有限公司 | Flood early warning method based on high-resolution remote sensing satellite geometric positioning model |
| CN114612801B (en) * | 2022-05-10 | 2022-07-12 | 环球数科集团有限公司 | Flood early warning method based on high-resolution remote sensing satellite geometric positioning model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112819234B (en) | 2022-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108897719B (en) | Meteorological data missing values complementing method based on self-adapted genetic algorithm | |
| CN110033164B (en) | Risk assessment and decision method for reservoir group combined flood control scheduling | |
| CN112819234B (en) | Flood forecasting method and system considering initial value correction | |
| CN113378484B (en) | Plain river network area flood process forecasting method based on machine learning | |
| CN116596044B (en) | Power generation load prediction model training method and device based on multi-source data | |
| CN105631532A (en) | Power system load prediction method using fuzzy decision-based neural network model | |
| CN113361195B (en) | Power distribution network line ultra-short term load prediction method based on LSTM and similar days | |
| CN111709454B (en) | Multi-wind-field output clustering evaluation method based on optimal copula model | |
| CN115186857A (en) | Neural network reservoir water level prediction method based on ensemble learning | |
| CN112883632A (en) | Lithium battery equivalent circuit model parameter identification method based on improved ant colony algorithm | |
| CN107657349B (en) | Method for extracting scheduling rules of staged power generation of reservoir | |
| CN115271253A (en) | Water-wind power generation power prediction model construction method and device and storage medium | |
| CN106598918A (en) | Non-uniform designed flood calculation method based on quantile regression | |
| CN112016839A (en) | Flood disaster prediction and early warning method based on QR-BC-ELM | |
| CN114819322A (en) | Method for forecasting lake inflow flow of lake | |
| CN112711896B (en) | Complex reservoir group optimal scheduling method considering multi-source forecast error uncertainty | |
| CN107341346B (en) | A kind of hydrologic forecasting method | |
| CN108470214A (en) | A kind of Bounded Errors method for parameter estimation based on range optimization algorithm | |
| US20050171882A1 (en) | System and method for making private equity commitments | |
| CN113486556B (en) | Improved efficient automatic history fitting method for oil and gas reservoir | |
| Guthrie | Optimal adaptation to uncertain climate change | |
| CN110782080B (en) | Electric power system structure optimization method based on population performance sorting selection | |
| CN114357865A (en) | Hydropower station runoff and associated source load power year scene simulation and prediction method thereof | |
| CN114266186A (en) | Air conditioner energy consumption data missing value filling method, equipment and medium | |
| CN116090678B (en) | Data processing method, device and equipment |
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 |