CN111754099B - Real-time flood control scheduling method based on three-stage risk hedging rule - Google Patents

Abstract

The invention discloses a real-time flood control scheduling method based on a three-stage risk hedging rule, which comprises the following steps of S1, determining a reliable prediction period by historical prediction information, and dividing the whole flood prediction period into three stages; s2, constructing a real-time optimization scheduling model based on a three-stage risk hedging rule, and dividing the model into an outer-layer two-stage model and an inner-layer two-stage model; s3, solving an outer layer two-stage model, and determining a flood control reservoir capacity for a reliable forestation period; s4, solving an inner layer two-stage model on the basis of the flood control reservoir capacity for the reliable forestation period, and determining the flood control reservoir capacity for the current period; and S5, calculating the delivery flow at the current moment according to the reservoir capacity water level curve and the reservoir flood discharge capacity curve. The advantages are that: the runoff forecasting information is fully utilized, the forecasting uncertainty is simulated in more detail by carrying out three-stage division on the forecasting time, and the flood control capacity of the reservoir is excavated and the utilization of the reservoir to flood resources is improved.

Description

Real-time flood control scheduling method based on three-stage risk hedging rule

Technical Field

The invention relates to the technical field of reservoir flood control dispatching management, in particular to a real-time flood control dispatching method based on three-stage risk hedging rules.

Background

Flood extreme events frequently occur under global climate change, on one hand, flood peak and flood retention can be reduced by utilizing reservoir real-time scheduling, on the other hand, flood can be fully regulated and stored, and the water resource utilization efficiency is improved. For most large reservoirs, the reservoirs simultaneously have multiple scheduling targets such as flood control, power generation, water supply and the like. During flood, the reservoir flood control aims require the reservoir to maintain low water level and low water storage as much as possible, while the reservoir power generation and other interesting aims require the reservoir to maintain high water level as much as possible, and obvious competition relations exist among the aims. Therefore, the method has practical significance in improving the real-time flood control scheduling to improve the utilization efficiency of flood resources as much as possible on the premise of ensuring the safety of the flood.

Because the forecast information based on reservoir dispatching has uncertainty characteristics, the real-time flood control dispatching inevitably has decision risk. When the predicted flood is earlier than actual, the reservoir may not be available for water due to premature pre-drainage, whereas when the predicted flood is later than actual, the reservoir has smaller flood control capacity and insufficient flood control capacity. To improve this problem, it is critical to make a runoff forecast and make full use of the forecast information. In the past research, the real-time scheduling mode of adopting two-stage risk hedging rules is widely focused and accepted in academia by integrating the current and future forecast uncertainties. The risk hedging mode aims at minimizing flood loss of the whole field by balancing marginal risks of the current stage and the future stage, and a reservoir water discharge scheme of the current period is determined in real time, so that the method has a certain economic theoretical basis. Researches show that the scheduling mode has remarkable effect of improving flood control and water resource utilization efficiency of the multi-target reservoir.

However, with the continuous improvement of the runoff forecasting technology, the drawbacks of the traditional two-stage forecasting time interval dividing mode are gradually highlighted. The two-stage division mode assumes that the relative prediction error of the future stage does not change with time, has the same prediction level, and the relative prediction error of reservoir storage can be simulated by the same probability density function. However, in flood control forecasting practice, the forecasting period is generally divided into three stages according to the forecasting mode. The first stage is the current scheduling period, corresponds to the observed warehouse-in flow, and the forecast error of the stage is mainly an observation error; the second phase corresponds to the warehouse-in flow forecasted by the observed rainfall and the land hydrologic model, and the forecasting error of the phase mainly comes from the hydrologic model; the third stage is based on the rainfall forecast and the hydrologic model forecast, and the forecast error of the third stage comprises the errors of the meteorological model and the hydrologic model. Due to the significant difference in the source of the forecast information, the uncertainty of the forecast flows at different stages is also significantly different. Therefore, the traditional two-stage prediction time interval division mode cannot well describe the dynamic change process of prediction errors along with time, and obvious defects exist in aspects of flood resource utilization and reservoir flood control capacity.

Disclosure of Invention

The invention aims to provide a real-time flood control scheduling method based on a three-stage risk hedging rule, so that the problems existing in the prior art are solved.

In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:

a real-time flood control scheduling method based on three-stage risk hedging rules, which comprises the following steps of,

s1, determining a reliable forecasting period by historical forecasting information, and dividing the whole flood forecasting period into three stages;

s2, constructing a real-time optimization scheduling model based on a three-stage risk hedging rule, and dividing the model into an outer-layer two-stage model and an inner-layer two-stage model;

s3, solving an outer layer two-stage model, and determining a flood control reservoir capacity for a reliable forestation period;

s4, solving an inner layer two-stage model on the basis of the flood control reservoir capacity for the reliable forestation period, and determining the flood control reservoir capacity for the current period;

and S5, combining the flood control reservoir capacity for the reliable foreseeing period and the flood control reservoir capacity for the current period, and calculating the delivery flow at the current moment according to the reservoir capacity water level curve and the reservoir flood discharge capacity curve.

Preferably, step S1 is specifically to set, according to the history forecast information, a period that does not include rainfall forecast information in the entire flood forecast period as a reliable forecast period; dividing the flood forecast period into three phases based on the reliable forecast period, wherein the current period in the reliable forecast period is the first phase; the future phase within the reliable foreseeable period is the second phase; the period after the reliable foreseeable period is the third phase, i.e. the future phase outside the reliable foreseeable period.

Preferably, step S2 specifically includes,

s21, simulating the runoff forecast uncertainty by a random variable meeting Gaussian distribution, and taking the calculation formula of the embankment accident risk of the forecast uncertainty into consideration,

wherein R represents a risk of embankment accident; q and Q ₀ Representing the actual value and the forecast value of the forecast runoff respectively; q (Q) _{d max} Is a river dikeMaximum safe overflow is prevented; sigma is a runoff forecast variance statistic, and represents uncertainty of runoff forecast; p represents probability;

considering that sigma varies with the forecast period, the forecast error is calculated by adopting a piecewise function, specifically,

σ(t+T _x )＝a(t+T _x )·Q _in (t+T _x ) (2)

wherein t represents the current calculation time; t (T) _x Is a dynamic foresight period; Δt is long in scheduling time period, namely long in interval time between scheduling operations of two times; a (t+T) _x ) As coefficients, the dynamic prediction error level is represented; constant a ₀ 、a ₁ 、a ₂ And b ₀ Determining from historical forecast information, i.e. counting the accuracy of the historical forecast information, estimating a at the node of the corresponding forecast period from the counted relative error of the forecast runoff ₀ 、a ₁ 、a ₂ And b ₀ ；T ₀ For the duration of a reliable prediction period; q (Q) _in The flow rate is the reservoir storage flow rate;

s22, based on a three-stage prediction period division mode, with the aim of minimizing the total event risk of the embankment, constructing a real-time optimization scheduling model based on a three-stage risk hedging rule,

min[R ₁ +(1-R ₁ )R ₂ +(1-R ₁ )(1-R ₂ )R ₃ ] (4)

wherein R is ₁ 、R ₂ And R is ₃ Representing the risk of embankment accident in three stages respectively, and determining the allocated flood control reservoir capacity; three-stage flood control reservoir capacity distribution value V _i (i=1, 2, 3) satisfying the following constraint;

V _i ≥(S _lim -S ₀ ) (5a)

V _i ≤(S _max -S ₀ ) (5b)

V ₁ +V ₂ +V ₃ ＝(S _ini -S ₀ ) (5c)

wherein S is _lim Limiting the water storage capacity of the reservoir corresponding to the water level for flood control; s is S _max The water storage capacity of the reservoir corresponding to the flood control high water level is obtained; s is S ₀ The water storage capacity of the reservoir corresponding to the water level is regulated for the current flood; s is S _ini The reservoir water storage capacity corresponding to the primary water level in the current scheduling period;

s23, decomposing a real-time optimized scheduling model based on a three-stage risk hedging rule, wherein the real-time optimized scheduling model comprises the following steps,

r is recorded ₁₂ ＝R ₁ +(1-R ₁ )R ₂ Then

R ₁ +(1-R ₁ )R ₂ +(1-R ₁ )(1-R ₂ )R ₃ ＝R ₁₂ +(1-R ₁₂ )R ₃ (7)

To implement equation (4), R is first minimized ₁₂ +(1-R ₁₂ )R ₃ Determining R ₁₂ Corresponding distribution storage capacity; re-minimizing R ₁ +(1-R ₁ )R ₂ Further determine R ₁ Corresponding distribution storage capacity; real-time optimized scheduling model based on three-stage risk hedging rule can be decomposed into an outer two-stage model min [ R ] ₁₂ +(1-R ₁₂ )R ₃ ]And inner layer two-stage model min [ R ] ₁ +(1-R ₁ )R ₂ ]。

Preferably, in step S3, the total water supply amount of a plurality of time periods can be adopted for decision analysis during the distribution of the outer two-stage flood control reservoir capacity; according to the reservoir dispatching time period length deltat, the outer layer two-stage model for determining the flood control reservoir capacity distribution of all time periods in the reliable foreseeable period is,

min[R ₁₂ +(1-R ₁₂ )R ₃ ] (8)

R ₁₂ ＝F ₁₂ (Q _{in_12} Δt+q ₁₂ Δt-V ₁₂ -Q _{d max} Δt) (9a)

R ₃ ＝F ₃ (Q _{in_3} Δt+q ₃ Δt-V ₃ -Q _{d max} Δt) (9b)

wherein, Q _in (t) represents the storage flow of the reservoir at the time t; q (t) represents the interval lateral inflow; sigma (sigma) ₁₂ Sum sigma ₃ Respectively representing the uncertainty of runoff forecasting within and outside the reliable forecasting period; v (V) ₁₂ And V ₃ For decision variables, respectively representing flood control reservoir capacities distributed outside a reliable forestation period; f and F represent a standard normal probability distribution function and a probability density function, respectively; v (V) ₁₂ And V ₃ The following constraint conditions are satisfied,

-V ₁₂ +(S _lim -S ₀ )≤0 (10a)

V ₁₂ -(S _max -S ₀ )≤0 (10b)

V ₁₂ +V ₃ -(S _ini -S ₀ )＝0 (10c)

since the nonlinear optimization model represented by formulas (8) and (10) satisfies the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁₂ )(1-R ₂ )-λ ₁ +λ ₂ +f ₂ (V ₃ )(1-R ₁ )＝0 (11a)

λ ₁ [-V ₁₂ +(S _lim -S ₀ )]＝0 (11b)

λ ₂ [V ₁₂ -(S _max -S ₀ )]＝0 (11c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0。

Preferably according to lambda ₁ And lambda (lambda) ₂ Different values of (a) correspond to different water-incoming situations, there are several risk hedging schemes,

a1, if lambda ₁ ＞0，V ₁₂ ＝S _lim -S ₀ ，λ ₂ =0; then there is

f ₁ (V ₁₂ )(1-R ₂ )＜f ₂ (V ₃ )(1-R ₁ )

At this time, the third stage outside the reliable prediction period predicts that the flood risk is larger, namely the flood is about to come, and the reservoir should be pre-drained to the flood limit water level as much as possible so as to ensure enough flood control reservoir capacity;

a2, if lambda ₂ ＞0，V ₁₂ ＝S _max -S ₀ ，λ ₁ =0; then there is

f ₁ (V ₁₂ )(1-R ₂ )＞f ₂ (V ₃ )(1-R ₁ )

At this time, flood in a reliable foreseeing period is larger, and all flood control reservoir capacities are applied to the current flood blocking;

a3, when lambda ₁ ＝λ ₂ ＝0，S _lim -S ₀ ＜V ₁₂ ＜S _max -S ₀ The method comprises the steps of carrying out a first treatment on the surface of the Then there is

f ₁ (V ₁₂ )(1-R ₂ )＝f ₂ (V ₃ )(1-R ₁ )

At this time, flood at the inner and outer stages of the period is reliably predicted to have larger risks, and the distribution of the flood control reservoir capacity follows the risk hedging rule, namely the marginal risks of the two stages are equal.

Preferably, step S4 is specifically to further determine an inner two-stage risk hedging rule according to the allocated reliable prediction period flood control reservoir capacity based on the outer two-stage risk hedging rule; the two-stage model of the inner layer is that,

min[R ₁ +(1-R ₁ )R ₂ ] (12)

wherein i=1, 2; q (Q) _{in_1} ＝Q _in (t)，Q _{in_2} ＝maxQ _in (t+j)；q ₁ ＝q(t)，q ₂ ＝maxq(t+j)，j＝1,2,3,4；

V which can be distributed in a foreseeable period according to different functions of the reservoir in different stages ₁₂ Whether the value is positive or not, two different conditions exist,

b1, when V ₁₂ When the water retention capacity is more than or equal to 0, the reservoir holds flood, and the decision variable V ₁ And V ₂ The following constraints are satisfied,

-V ₁ ≤0 (14a)

-V ₂ ≤0 (14b)

V ₁ +V ₂ ＝V ₁₂ (14c)

since the nonlinear optimization models represented by formulas (12) and (14) satisfy the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁ )(1-R ₂ )-λ ₁ +f ₂ (V ₂ )(1-R ₁ )+λ ₂ ＝0 (15a)

λ ₁ V ₁ ＝0 (15b)

λ ₂ V ₂ ＝0 (15c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0；

B2, when V ₁₂ When the pressure is less than 0, the reservoir is pre-drained, and the decision variable V is determined ₁ And V ₂ The following constraints are satisfied,

V ₁ ≤0 (16a)

V ₂ ≤0 (16b)

V ₁ +V ₂ ＝V ₁₂ (16c)

since the nonlinear optimization models represented by formulas (12) and (16) satisfy the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁ )(1-R ₂ )+λ ₁ +f ₂ (V ₂ )(1-R ₁ )-λ ₂ ＝0 (17a)

λ ₁ V ₁ ＝0 (17b)

λ ₂ V ₂ ＝0 (17c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0；

Preferably, for case B1, according to λ ₁ And lambda (lambda) ₂ Different values of (c) there are several risk hedging schemes,

c1, if lambda ₁ ＞0，V ₁ ＝0，V ₂ ＝V ₁₂ ，λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＜f ₂ (V ₂ )(1-R ₁ )

At this time, the flood in the second stage is larger than that in the current period, the reservoir does not block flood in the current period, and the reservoir capacity V ₁₂ Assigned to the second stage;

c2, if lambda ₂ ＞0，V ₂ ＝0，V ₁ ＝V ₁₂ ，λ ₁ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＞f ₂ (V ₂ )(1-R ₁ )

At this time, flood is larger at the current moment, and all storage capacities V ₁₂ Flood control for the current period;

c3, lambda ₁ ＝λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＝f ₂ (V ₂ )(1-R ₁ )

At this time, the storage capacity to be distributed in two stages is determined according to the marginal risk equality.

Preferably, for case B2, according to λ ₁ And lambda (lambda) ₂ Different values of (c) there are several risk hedging schemes,

d1, if lambda ₁ ＞0，V ₁ ＝0，V ₂ ＝V ₁₂ ，λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＞f ₂ (V ₂ )(1-R ₁ )

At this time, as flood does not arrive yet, the risk is small, and the current period is not pre-leaked;

d2, if lambda ₂ ＞0，V ₂ ＝0，V ₁ ＝V ₁₂ ，λ ₁ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＜f ₂ (V ₂ )(1-R ₁ )

At this time, because of the proximity of the flood, the risk is large, and the current period is about to be pre-leaked to the flood limit water level;

d3, when lambda ₁ ＝λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＝f ₂ (V ₂ )(1-R ₁ )

At this time, the pre-leak amount for the current period is determined by the two-stage marginal risk equality.

Preferably, in step S5, reservoir flood control calculation is performed according to water balance and water level-reservoir capacity curve, and the current delivery flow is calculated as follows,

dS/dt＝Q _in -Q _out (18)

S＝f(Z) (19)

wherein S is the water storage capacity of the reservoir; q (Q) _out The flow for reservoir delivery; z is the reservoir level; f (Z) represents the water level-reservoir capacity curve of the reservoir.

The beneficial effects of the invention are as follows: 1. the method provided by the invention provides a concept of a reliable prediction period according to whether a rainfall prediction error is included, the reliable prediction period is divided into three phases, a real-time scheduling optimization model of a three-phase risk hedging rule is constructed on the basis, and the three-phase model is decomposed into two-phase models for solving, so that the method is effective and feasible for improving the flood control capacity and the flood resource utilization efficiency of the reservoir. 2. The method provided by the invention fully utilizes runoff forecast information, and the forecast uncertainty is simulated in more detail by carrying out three-stage division on the forecast time, so that the flood control capacity of the reservoir is excavated, and meanwhile, the utilization of the reservoir to flood resources is improved.

Drawings

FIG. 1 is a schematic flow chart of a method according to an embodiment of the invention;

FIG. 2 is a schematic diagram of a reliable prediction period and a forecast three-stage division in an embodiment of the present invention;

FIG. 3 is a schematic diagram of a river dike failure probability density function in an embodiment of the invention;

FIG. 4 is a schematic diagram of three-stage division and two-stage division of inner and outer layers of a prediction period in an embodiment of the present invention;

FIG. 5 is a schematic diagram of a typical flood process of a 12-stroke water reservoir in a vantage water reservoir in an embodiment of the present invention;

fig. 6 is a schematic diagram of a partial session flood scheduling result based on a three-stage risk hedging rule in an embodiment of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the detailed description is presented by way of example only and is not intended to limit the invention.

Example 1

As shown in fig. 1, a real-time flood control scheduling method based on a three-stage risk hedging rule, the method comprising,

s1, determining a reliable forecasting period by historical forecasting information, and dividing the whole flood forecasting period into three stages;

s2, constructing a real-time optimization scheduling model based on a three-stage risk hedging rule, and dividing the model into an outer-layer two-stage model and an inner-layer two-stage model;

s3, solving an outer layer two-stage model, and determining a flood control reservoir capacity for a reliable forestation period;

s4, solving an inner layer two-stage model on the basis of the flood control reservoir capacity for the reliable forestation period, and determining the flood control reservoir capacity for the current period;

and S5, combining the flood control reservoir capacity for the reliable foreseeing period and the flood control reservoir capacity for the current period, and calculating the delivery flow at the current moment according to the reservoir capacity water level curve and the reservoir flood discharge capacity curve.

In this embodiment, the flood control scheduling method mainly includes reliable prediction period determination and prediction period division, real-time optimization scheduling model and division of three-stage risk hedging rules, outer-layer two-stage model solving, inner-layer two-stage model solving, and current time delivery flow calculating. The method comprises the following steps:

1. reliable prediction period determination and prediction period partitioning

In this embodiment, along with the growth of the forecast period, the forecast error also increases obviously, and the uncertainty of the forecast warehouse-in presents a typical nonlinear growth trend along with the forecast period. In a complete flood forecast, different information sources can cause the in-store pre-report to show significant uncertainty differences. Since the accuracy of the meteorological model is far less than that of the hydrological model, the uncertainty is significantly greater when the rainfall forecast information is contained during the forecast period. Therefore, the invention considers the average prediction precision level of the hydrologic model, provides a concept of reliable prediction period to distinguish whether the prediction period of rainfall prediction information is included or not, and can also clearly determine the prediction information corresponding to three different prediction approaches. As can be seen from fig. 2, the runoff forecast error is relatively small and the growth is small for the reliable forecast period, that is, the reliable forecast period does not include rainfall forecast information.

Relative to a normal flood forecast process (duration T ₁ ) The reliable prediction period is relatively short (duration T ₀ ) The prediction error is also significantly smaller (the relative error is typically less than 20%) within the reliable prediction period. As in fig. 2, the flood forecast is divided into three phases based on the reliable forecast period: the current phase reliably predicts the future phase within the period and reliably predicts the future phase outside the period.

Determining the content of the corresponding step S1 in the reliable forecast period, specifically, setting a period which does not contain rainfall forecast information in the whole flood forecast period as the reliable forecast period according to the historical forecast information; dividing the flood forecast period into three phases based on the reliable forecast period, wherein the current period in the reliable forecast period is the first phase; the future phase within the reliable foreseeable period is the second phase; the period after the reliable foreseeable period is the third phase, i.e. the future phase outside the reliable foreseeable period.

The history forecast information mainly comprises basic data such as reservoir capacity curves and river flood control capacity.

2. Real-time optimization scheduling model and division of three-stage risk hedging rule

In this embodiment, the real-time optimized scheduling model and the partitioning of the three-stage risk hedging rule correspond to the content of step S2, which specifically includes,

s21, the runoff forecast uncertainty can be generally simulated by random variables meeting Gaussian distribution, as shown in figure 3, taking river flood control as an example, taking a embankment accident risk calculation formula of the forecast uncertainty as,

wherein R represents a risk of embankment accident; q and Q ₀ Respectively representing the actual value and the forecast value of the forecast runoff, m ³ /s；Q _dmax For the maximum safe overflow of river dikes, m ³ S; sigma is the variance statistic of the runoff forecast, characterizes the uncertainty of the runoff forecast, m ³ S; p represents probability;

in the previous real-time scheduling using the two-stage risk hedging rule, the uncertainty σ of the runoff forecast of the future period is usually assumed to be constant, however, in the three-stage model of the invention, the variation of σ along with the forecast period is considered, and the forecast error is calculated by adopting a piecewise function, specifically,

σ(t+T _x )＝a(t+T _x )·Q _in (t+T _x ) (2)

wherein, t is shown in tableShowing the current calculation time h; t (T) _x H is a dynamic foreseeing period; Δt is long in scheduling time period, namely long in interval time of two scheduling operations, and h; a (t+T) _x ) As coefficients, the dynamic prediction error level is represented; constant a ₀ 、a ₁ 、a ₂ And b ₀ Determining from historical forecast information, i.e. counting the accuracy of the historical forecast information, estimating a at the node of the corresponding forecast period from the counted relative error of the forecast runoff ₀ 、a ₁ 、a ₂ And b ₀ ；T ₀ For the duration of a reliable prediction period; q (Q) _in The flow rate is the reservoir storage flow rate;

s22, in real-time flood control dispatching, the reservoir is generally regulated from the flood limit water level and does not exceed the flood control high water level; based on the three-stage prediction period division mode, with the aim of minimizing the total jetty accident risk, constructing a real-time optimized scheduling model (objective function) based on the three-stage risk hedging rule,

min[R ₁ +(1-R ₁ )R ₂ +(1-R ₁ )(1-R ₂ )R ₃ ] (4)

wherein R is ₁ 、R ₂ And R is ₃ Representing the risk of embankment accident in three stages respectively, and determining the allocated flood control reservoir capacity; three-stage flood control reservoir capacity distribution value V _i (i=1, 2, 3) satisfying the following constraint;

V _i ≥(S _lim -S ₀ ) (5a)

V _i ≤(S _max -S ₀ ) (5b)

V ₁ +V ₂ +V ₃ ＝(S _ini -S ₀ ) (5c)

wherein S is _lim Reservoir water storage capacity m corresponding to flood control limit water level ³ ；S _max For the reservoir water storage capacity corresponding to the flood control high water level, m ³ ；S ₀ Reservoir water storage capacity corresponding to water level of current flood, m ³ ；S _ini For the reservoir water storage capacity corresponding to the primary water level in the current dispatching period, m ³ ；

S23, considering that the difficulty in directly solving the three-stage optimization model represented by the formulas (4) and (5) is high, the invention decomposes the three-stage model, namely, the real-time optimization scheduling model based on the three-stage risk hedging rule,

r is recorded ₁₂ ＝R ₁ +(1-R ₁ )R ₂ Then

R ₁ +(1-R ₁ )R ₂ +(1-R ₁ )(1-R ₂ )R ₃ ＝R ₁₂ +(1-R ₁₂ )R ₃ (7)

To implement equation (4), R is first minimized ₁₂ +(1-R ₁₂ )R ₃ Determining R ₁₂ Corresponding distribution storage capacity; re-minimizing R ₁ +(1-R ₁ )R ₂ Further determine R ₁ Corresponding distribution storage capacity; real-time optimized scheduling model based on three-stage risk hedging rule can be decomposed into an outer two-stage model min [ R ] ₁₂ +(1-R ₁₂ )R ₃ ]And inner layer two-stage model min [ R ] ₁ +(1-R ₁ )R ₂ ]The method comprises the steps of carrying out a first treatment on the surface of the That is, the three-phase model may be decomposed into two-step implementations:

step one, min [ R ] ₁₂ +(1-R ₁₂ )R ₃ ]；

Step two, min [ R ] ₁ +(1-R ₁ )R ₂ ]；

The three-stage model is decomposed into a first-step implementation and a second-step implementation, which is essentially implemented by performing risk hedging analysis twice successively between three forecasting periods. The first step is to allocate the optimal flood control storage capacity in the reliable foreseeing period and outside the reliable foreseeing period, and the second step is to allocate the optimal flood control storage capacity in the current period and the future period in the reliable foreseeing period, as shown in fig. 4.

3. Outer two-stage model solution

In this embodiment, in combination with the first step, the content of the step S3 corresponds to that, for the outer two-stage model, since the runoff prediction outside the reliable prediction period depends on the rainfall prediction result, and the runoff prediction result error in this period is relatively large, when the outer two-stage flood control reservoir capacity is allocated, the total water supply in multiple periods can be considered for decision analysis; according to the reservoir dispatching time period length deltat, the outer layer two-stage model for determining the flood control reservoir capacity distribution of all time periods in the reliable foreseeable period is,

min[R ₁₂ +(1-R ₁₂ )R ₃ ] (8)

R ₁₂ ＝F ₁₂ (Q _{in_12} Δt+q ₁₂ Δt-V ₁₂ -Q _{d max} Δt) (9a)

R ₃ ＝F ₃ (Q _{in_3} Δt+q ₃ Δt-V ₃ -Q _{d max} Δt) (9b)

wherein, Q _in (t) represents the flow rate of the reservoir in storage at the time t, m ³ /s；σ ₁₂ Sum sigma ₃ Respectively represent the uncertainty of runoff forecasting within and outside the reliable forecasting period, m ³ S; q (t) represents the interval lateral inflow, m ³ /s；V ₁₂ And V ₃ For decision variables, representing the flood control reservoir capacity m allocated to the outside of the reliable forestation period ³ The method comprises the steps of carrying out a first treatment on the surface of the F and F represent a standard normal probability distribution function and a probability density function, respectively; decision variable V ₁₂ And V ₃ The following constraint conditions are satisfied,

-V ₁₂ +(S _lim -S ₀ )≤0 (10a)

V ₁₂ -(S _max -S ₀ )≤0 (10b)

V ₁₂ +V ₃ -(S _ini -S ₀ )＝0 (10c)

since the nonlinear optimization model represented by formulas (8) and (10) satisfies the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁₂ )(1-R ₂ )-λ ₁ +λ ₂ +f ₂ (V ₃ )(1-R ₁ )＝0 (11a)

λ ₁ [-V ₁₂ +(S _lim -S ₀ )]＝0 (11b)

λ ₂ [V ₁₂ -(S _max -S ₀ )]＝0 (11c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0。

In the present embodiment, according to lambda ₁ And lambda (lambda) ₂ Different values of (a) correspond to different water-incoming situations, there are several risk hedging schemes,

a1, if lambda ₁ ＞0，V ₁₂ ＝S _lim -S ₀ ，λ ₂ =0; then there is

f ₁ (V ₁₂ )(1-R ₂ )＜f ₂ (V ₃ )(1-R ₁ )

At this time, the third stage outside the reliable prediction period predicts that the flood risk is larger, namely the flood is about to come, and the reservoir should be pre-drained to the flood limit water level as much as possible so as to ensure enough flood control reservoir capacity;

a2, if lambda ₂ ＞0，V ₁₂ ＝S _max -S ₀ ，λ ₁ =0; then there is

f ₁ (V ₁₂ )(1-R ₂ )＞f ₂ (V ₃ )(1-R ₁ )

At this time, flood in a reliable foreseeing period is larger, and all flood control reservoir capacities are applied to the current flood blocking;

a3, when lambda ₁ ＝λ ₂ ＝0，S _lim -S ₀ ＜V ₁₂ ＜S _max -S ₀ The method comprises the steps of carrying out a first treatment on the surface of the Then there is

f ₁ (V ₁₂ )(1-R ₂ )＝f ₂ (V ₃ )(1-R ₁ )

At this time, flood at the inner and outer stages of the period is reliably predicted to have larger risks, and the distribution of flood control storage capacity follows a risk hedging rule, namely, the marginal risks of the two stages are equal (a storage capacity distribution scheme meeting the equal marginal risks can be determined in a trial-and-error mode).

4. Inner two-stage model solution

In this embodiment, in combination with the second step, the content corresponding to the step S4 is specifically that, based on the outer layer two-stage risk hedging rule, the inner layer two-stage risk hedging rule is further determined according to the allocated reliable prediction period flood control reservoir capacity; the two-stage model of the inner layer is that,

min[R ₁ +(1-R ₁ )R ₂ ] (12)

wherein i=1, 2; q (Q) _{in_1} ＝Q _in (t)，Q _{in_2} ＝maxQ _in (t+j)；q ₁ ＝q(t)，q ₂ ＝maxq(t+j)，j＝1,2,3,4；

V which can be distributed in a foreseeable period according to different functions of the reservoir in different stages ₁₂ Whether the value is positive or not, two different conditions exist,

b1, when V ₁₂ When the water retention capacity is more than or equal to 0, the reservoir holds flood, and the decision variable V ₁ And V ₂ The following constraints are satisfied,

-V ₁ ≤0 (14a)

-V ₂ ≤0 (14b)

V ₁ +V ₂ ＝V ₁₂ (14c)

since the nonlinear optimization models represented by formulas (12) and (14) satisfy the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁ )(1-R ₂ )-λ ₁ +f ₂ (V ₂ )(1-R ₁ )+λ ₂ ＝0 (15a)

λ ₁ V ₁ ＝0 (15b)

λ ₂ V ₂ ＝0 (15c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0；

For case B1, according to lambda ₁ And lambda (lambda) ₂ Different values of (c) there are several risk hedging schemes,

c1, if lambda ₁ ＞0，V ₁ ＝0，V ₂ ＝V ₁₂ ，λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＜f ₂ (V ₂ )(1-R ₁ )

At this time, the flood in the second stage is larger than that in the current period, the reservoir does not block flood in the current period, and the reservoir capacity V ₁₂ Assigned to the second stage;

c2, if lambda ₂ ＞0，V ₂ ＝0，V ₁ ＝V ₁₂ ，λ ₁ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＞f ₂ (V ₂ )(1-R ₁ )

At this time, flood is larger at the current moment, and all storage capacities V ₁₂ Flood control for the current period;

c3, lambda ₁ ＝λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＝f ₂ (V ₂ )(1-R ₁ )

At this time, the storage capacity to be distributed in two stages is determined according to the marginal risk equality.

B2, when V ₁₂ When the pressure is less than 0, the reservoir is pre-drained, and the decision variable V is determined ₁ And V ₂ The following constraints are satisfied,

V ₁ ≤0 (16a)

V ₂ ≤0 (16b)

V ₁ +V ₂ ＝V ₁₂ (16c)

since the nonlinear optimization models represented by formulas (12) and (16) satisfy the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁ )(1-R ₂ )+λ ₁ +f ₂ (V ₂ )(1-R ₁ )-λ ₂ ＝0 (17a)

λ ₁ V ₁ ＝0 (17b)

λ ₂ V ₂ ＝0 (17c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0；

For case B2, according to lambda ₁ And lambda (lambda) ₂ Different values of (c) there are several risk hedging schemes,

d1, if lambda ₁ ＞0，V ₁ ＝0，V ₂ ＝V ₁₂ ，λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＞f ₂ (V ₂ )(1-R ₁ )

At this time, as flood does not arrive yet, the risk is small, and the current period is not pre-leaked;

d2, if lambda ₂ ＞0，V ₂ ＝0，V ₁ ＝V ₁₂ ，λ ₁ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＜f ₂ (V ₂ )(1-R ₁ )

At this time, because of the proximity of the flood, the risk is large, and the current period is about to be pre-leaked to the flood limit water level;

d3, when lambda ₁ ＝λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＝f ₂ (V ₂ )(1-R ₁ )

At this time, the pre-leak amount for the current period is determined by the two-stage marginal risk equality. And then, by executing the step S5, the delivery flow at the current moment can be calculated and acquired.

5. Calculating the current time of delivery flow

In this embodiment, the calculation of the current delivery flow corresponds to the content of step S5, specifically, the calculation of reservoir flood control is implemented according to the water balance and the water level-reservoir capacity curve without considering reservoir evaporation and leakage, and the calculation of the current delivery flow is as follows,

dS/dt＝Q _in -Q _out (18)

S＝f(Z) (19)

wherein S is the water storage capacity of the reservoir; q (Q) _out The flow for reservoir delivery; z is the reservoir level; f (Z) represents the water level-reservoir capacity curve of the reservoir.

Example two

In this embodiment, the effectiveness of the method of the present invention will be specifically described by taking the upstream wan reservoir in the Ganjiang river in the Jiangxi province as an example. The main tasks of the Van reservoirs (E114 DEG 41', N26 DEG 33') comprise water supply, power generation, flood control, shipping and the like, and the basic information of the reservoirs is shown in the following table 1.

Table 1 basic information table for wan an reservoir

	Design value	Description of the invention
			Flood limit water level	85.0m	Corresponding stock capacity is 319 million m 3
Normal running water level in flood season	88.0m	Corresponding stock capacity is 472 million m 3
			Flood control high water level	93.6m	Corresponding storage capacity is 889.2 million m 3
Current running mode of reservoir	-	Forecasting of pre-diarrhea
			River course flood control ability	8800m 3 /s	-
Reliable prediction period	24.0h	The scheduling time period is 6h long

The invention selects representative flood of 12 occasions 10 years after 1995 to conduct real-time scheduling analysis so as to verify the effectiveness of the method, as shown in figure 5.

The real-time flood control scheduling method based on the three-stage risk hedging rule is adopted, and the obtained scheduling result is shown in table 2 and fig. 6.

Table 2 flood control scheduling results based on three-stage risk hedging rules

/>

As can be seen from the table 2, after the scheduling rule provided by the invention is adopted, the flood control capacity of the river channel is obviously improved, and the power generation benefit is also obviously improved. As seen in fig. 6, after the three-stage risk hedging rule is adopted, the flood control capacity of the reservoir is more fully utilized, and the peak value of the storage flood is obviously weakened. The flood blocked by the reservoir not only reduces the downstream flood prevention risk, but also raises the running water level of the reservoir, and increases the power generation benefit.

In conclusion, the real-time flood control scheduling based on the three-stage risk hedging rule fully utilizes runoff forecast information, the forecast uncertainty is simulated in more detail by carrying out three-stage division on forecast time, and the flood control capacity of the reservoir is excavated and the utilization of flood resources of the reservoir is improved. The invention is effective and feasible through the verification of the wanan embodiment.

By adopting the technical scheme disclosed by the invention, the following beneficial effects are obtained:

the invention provides a real-time flood control scheduling method based on a three-stage risk hedging rule. Meanwhile, the method fully utilizes runoff forecasting information, the forecasting uncertainty is simulated in more detail by carrying out three-stage division on forecasting time, and the flood control capacity of the reservoir is excavated and the utilization of the reservoir to flood resources is improved.

The foregoing is merely a preferred embodiment of the present invention and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present invention, which is also intended to be covered by the present invention.

Claims (7)

1. A real-time flood control scheduling method based on three-stage risk hedging rules is characterized in that: the method may include the steps of,

s1, determining a reliable forecasting period by historical forecasting information, and dividing the whole flood forecasting period into three stages; step S1, specifically, according to historical forecast information, setting a period which does not contain rainfall forecast information in the whole flood forecast period as a reliable forecast period; dividing the flood forecast period into three phases based on the reliable forecast period, wherein the current period in the reliable forecast period is the first phase; the future phase within the reliable foreseeable period is the second phase; the time period after the reliable foreseeing period is a third stage, namely a future stage outside the reliable foreseeing period;

s2, constructing a real-time optimization scheduling model based on a three-stage risk hedging rule, and dividing the model into an outer-layer two-stage model and an inner-layer two-stage model; step S2 specifically includes the following,

s21, simulating the runoff forecast uncertainty by a random variable meeting Gaussian distribution, and taking the calculation formula of the embankment accident risk of the forecast uncertainty into consideration,

wherein R represents a risk of embankment accident; q and Q ₀ Representing the actual value and the forecast value of the forecast runoff respectively; q (Q) _dmax The maximum safe overflow of the river dike is achieved; sigma is a runoff forecast variance statistic, and represents uncertainty of runoff forecast; p represents probability;

considering that sigma varies with the forecast period, the forecast error is calculated by adopting a piecewise function, specifically,

σ(t+T _x )＝a(t+T _x )·Q _in (t+T _x ) (2)

wherein t represents the current calculation time; t (T) _x Is a dynamic foresight period; Δt is long in scheduling time period, namely long in interval time between scheduling operations of two times; a (t+T) _x ) As coefficients, the dynamic prediction error level is represented; constant a ₀ 、a ₁ 、a ₂ And b ₀ Determining from historical forecast information, i.e. counting the accuracy of the historical forecast information, estimating a at the node of the corresponding forecast period from the counted relative error of the forecast runoff ₀ 、a ₁ 、a ₂ And b ₀ ；T ₀ For the duration of a reliable prediction period; q (Q) _in The flow rate is the reservoir storage flow rate;

s22, based on a three-stage prediction period division mode, with the aim of minimizing the total event risk of the embankment, constructing a real-time optimization scheduling model based on a three-stage risk hedging rule,

min[R ₁ +(1-R ₁ )R ₂ +(1-R ₁ )(1-R ₂ )R ₃ ] (4)

wherein R is ₁ 、R ₂ And R is ₃ Representing the risk of embankment accident in three stages respectively, and determining the allocated flood control reservoir capacity; three-stage flood control reservoir capacity distribution value V _i (i=1, 2, 3) satisfying the following constraint;

V _i ≥(S _lim -S ₀ ) (5a)

V _i ≤(S _max -S ₀ ) (5b)

V ₁ +V ₂ +V ₃ ＝(S _ini -S ₀ ) (5c)

wherein S is _lim Limiting the water storage capacity of the reservoir corresponding to the water level for flood control; s is S _max The water storage capacity of the reservoir corresponding to the flood control high water level is obtained; s is S ₀ The water storage capacity of the reservoir corresponding to the water level is regulated for the current flood; s is S _ini The reservoir water storage capacity corresponding to the primary water level in the current scheduling period;

s23, decomposing a real-time optimized scheduling model based on a three-stage risk hedging rule, wherein the real-time optimized scheduling model comprises the following steps,

r is recorded ₁₂ ＝R ₁ +(1-R ₁ )R ₂ Then

R ₁ +(1-R ₁ )R ₂ +(1-R ₁ )(1-R ₂ )R ₃ ＝R ₁₂ +(1-R ₁₂ )R ₃ (7)

To implement equation (4), R is first minimized ₁₂ +(1-R ₁₂ )R ₃ Determining R ₁₂ Corresponding distribution storage capacity; re-minimizing R ₁ +(1-R ₁ )R ₂ Further determine R ₁ Corresponding distribution storage capacity; real-time optimized scheduling model based on three-stage risk hedging rule can be decomposed into an outer two-stage model min [ R ] ₁₂ +(1-R ₁₂ )R ₃ ]And inner layer two-stage model min [ R ] ₁ +(1-R ₁ )R ₂ ]；

S3, solving an outer layer two-stage model, and determining a flood control reservoir capacity for a reliable forestation period;

s4, solving an inner layer two-stage model on the basis of the flood control reservoir capacity for the reliable forestation period, and determining the flood control reservoir capacity for the current period;

and S5, combining the flood control reservoir capacity for the reliable foreseeing period and the flood control reservoir capacity for the current period, and calculating the delivery flow at the current moment according to the reservoir capacity water level curve and the reservoir flood discharge capacity curve.

2. The real-time flood control scheduling method based on three-stage risk hedging rule according to claim 1, wherein: step S3, specifically, when the outer two-stage flood control reservoir capacity is distributed, the total water supply amount of a plurality of time periods can be adopted for decision analysis; according to the reservoir dispatching time period length deltat, the outer layer two-stage model for determining the flood control reservoir capacity distribution of all time periods in the reliable foreseeable period is,

min[R ₁₂ +(1-R ₁₂ )R ₃ ] (8)

R ₁₂ ＝F ₁₂ (Q _{in_12} Δt+q ₁₂ Δt-V ₁₂ -Q _dmax Δt) (9a)

R ₃ ＝F ₃ (Q _{in_3} Δt+q ₃ Δt-V ₃ -Q _dmax Δt) (9b)

wherein,Q _in (t) represents the storage flow of the reservoir at the time t; q (t) represents the interval lateral inflow; sigma (sigma) ₁₂ Sum sigma ₃ Respectively representing the uncertainty of runoff forecasting within and outside the reliable forecasting period; v (V) ₁₂ And V ₃ For decision variables, respectively representing flood control reservoir capacities distributed outside a reliable forestation period; f and F represent a standard normal probability distribution function and a probability density function, respectively; v (V) ₁₂ And V ₃ The following constraint conditions are satisfied,

-V ₁₂ +(S _lim -S ₀ )≤0 (10a)

V ₁₂ -(S _max -S ₀ )≤0 (10b)

V ₁₂ +V ₃ -(S _ini -S ₀ )＝0 (10c)

since the nonlinear optimization model represented by formulas (8) and (10) satisfies the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁₂ )(1-R ₂ )-λ ₁ +λ ₂ +f ₂ (V ₃ )(1-R ₁ )＝0 (11a)

λ ₁ [-V ₁₂ +(S _lim -S ₀ )]＝0 (11b)

λ ₂ [V ₁₂ -(S _max -S ₀ )]＝0 (11c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0。

3. The real-time flood control scheduling method based on three-stage risk hedging rule according to claim 2, wherein: according to lambda ₁ And lambda (lambda) ₂ Different values of (a) correspond to different water-incoming situations, there are several risk hedging schemes,

a1, if lambda ₁ ＞0，V ₁₂ ＝S _lim -S ₀ ，λ ₂ ＝0；Then there is

f ₁ (V ₁₂ )(1-R ₂ )＜f ₂ (V ₃ )(1-R ₁ )

At this time, the third stage outside the reliable prediction period predicts that the flood risk is larger, namely the flood is about to come, and the reservoir should be pre-drained to the flood limit water level as much as possible so as to ensure enough flood control reservoir capacity;

a2, if lambda ₂ ＞0，V ₁₂ ＝S _max -S ₀ ，λ ₁ =0; then there is

f ₁ (V ₁₂ )(1-R ₂ )＞f ₂ (V ₃ )(1-R ₁ )

At this time, flood in a reliable foreseeing period is larger, and all flood control reservoir capacities are applied to the current flood blocking;

a3, when lambda ₁ ＝λ ₂ ＝0，S _lim -S ₀ ＜V ₁₂ ＜S _max -S ₀ The method comprises the steps of carrying out a first treatment on the surface of the Then there is

f ₁ (V ₁₂ )(1-R ₂ )＝f ₂ (V ₃ )(1-R ₁ )

At this time, flood at the inner and outer stages of the period is reliably predicted to have larger risks, and the distribution of the flood control reservoir capacity follows the risk hedging rule, namely the marginal risks of the two stages are equal.

4. A real-time flood control scheduling method based on three-stage risk hedging rules according to claim 3, characterized in that: step S4, specifically, on the basis of the outer layer two-stage risk hedging rule, further determining the inner layer two-stage risk hedging rule according to the allocated reliable prediction period flood control reservoir capacity; the two-stage model of the inner layer is that,

min[R ₁ +(1-R ₁ )R ₂ ] (12)

wherein i=1, 2; q (Q) _{in_1} ＝Q _in (t)，Q _{in_2} ＝maxQ _in (t+j)；q ₁ ＝q(t)，q ₂ ＝maxq(t+j)，j＝1,2,3,4；

V which can be distributed in a foreseeable period according to different functions of the reservoir in different stages ₁₂ Whether the value is positive or not, two different conditions exist,

b1, when V ₁₂ When the water retention capacity is more than or equal to 0, the reservoir holds flood, and the decision variable V ₁ And V ₂ The following constraints are satisfied,

-V ₁ ≤0 (14a)

-V ₂ ≤0 (14b)

V ₁ +V ₂ ＝V ₁₂ (14c)

since the nonlinear optimization models represented by formulas (12) and (14) satisfy the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁ )(1-R ₂ )-λ ₁ +f ₂ (V ₂ )(1-R ₁ )+λ ₂ ＝0 (15a)

λ ₁ V ₁ ＝0 (15b)

λ ₂ V ₂ ＝0 (15c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0；

B2, when V ₁₂ When the pressure is less than 0, the reservoir is pre-drained, and the decision variable V is determined ₁ And V ₂ The following constraints are satisfied,

V ₁ ≤0 (16a)

V ₂ ≤0 (16b)

V ₁ +V ₂ ＝V ₁₂ (16c)

since the nonlinear optimization models represented by formulas (12) and (16) satisfy the sufficiency and necessity of the KKT condition; then there is

-f ₁ (V ₁ )(1-R ₂ )+λ ₁ +f ₂ (V ₂ )(1-R ₁ )-λ ₂ ＝0 (17a)

λ ₁ V ₁ ＝0 (17b)

λ ₂ V ₂ ＝0 (17c)

Wherein lambda is ₁ And lambda (lambda) ₂ Lambda is the Lagrangian multiplier ₁ ≥0，λ ₂ ≥0；

5. The real-time flood control scheduling method based on three-stage risk hedging rule according to claim 4, wherein: for case B1, according to lambda ₁ And lambda (lambda) ₂ Different values of (c) there are several risk hedging schemes,

c1, if lambda ₁ ＞0，V ₁ ＝0，V ₂ ＝V ₁₂ ，λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＜f ₂ (V ₂ )(1-R ₁ )

At this time, the flood in the second stage is larger than that in the current period, the reservoir does not block flood in the current period, and the reservoir capacity V ₁₂ Assigned to the second stage;

c2, if lambda ₂ ＞0，V ₂ ＝0，V ₁ ＝V ₁₂ ，λ ₁ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＞f ₂ (V ₂ )(1-R ₁ )

At this time, flood is larger at the current moment, and all storage capacities V ₁₂ Flood control for the current period;

c3, lambda ₁ ＝λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＝f ₂ (V ₂ )(1-R ₁ )

At this time, the storage capacity to be distributed in two stages is determined according to the marginal risk equality.

6. The real-time flood control scheduling method based on three-stage risk hedging rule according to claim 5, wherein: for case B2, according to lambda ₁ And lambda (lambda) ₂ Different values of (c) there are several risk hedging schemes,

d1, if lambda ₁ ＞0，V ₁ ＝0，V ₂ ＝V ₁₂ ，λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＞f ₂ (V ₂ )(1-R ₁ )

At this time, as flood does not arrive yet, the risk is small, and the current period is not pre-leaked;

d2, if lambda ₂ ＞0，V ₂ ＝0，V ₁ ＝V ₁₂ ，λ ₁ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＜f ₂ (V ₂ )(1-R ₁ )

At this time, because of the proximity of the flood, the risk is large, and the current period is about to be pre-leaked to the flood limit water level;

d3, when lambda ₁ ＝λ ₂ =0; then there is

f ₁ (V ₁ )(1-R ₂ )＝f ₂ (V ₂ )(1-R ₁ )

At this time, the pre-leak amount for the current period is determined by the two-stage marginal risk equality.

7. The real-time flood control scheduling method based on three-stage risk hedging rule according to claim 1, wherein: step S5 is that reservoir flood control calculation is realized according to water balance and a water level-reservoir capacity curve, the delivery flow at the current moment is calculated as follows,

dS/dt＝Q _in -Q _out (18)

S＝f(Z) (19)

wherein S is the water storage capacity of the reservoir; q (Q) _out The flow for reservoir delivery; z is the reservoir level; f (Z) represents the water level-reservoir capacity curve of the reservoir.