CN103088784B - Cascade reservoir flood control water level real-time dynamic control method - Google Patents

Abstract

The invention discloses a cascade reservoir flood control water level real-time dynamic control method. The cascade reservoir flood control water level real-time dynamic control method comprises the following steps: (1) building a numerical meteorological hydrological forecast model of the flood season of a river basin, forecasting the flood process of the river basin of the future 1- 7 days in a rolling mode, (2) building a 'large scale system polymerization idea'-based random long-term optimization scheduling graph model, using a self-adaptive genetic algorithm to make a long-term optimization scheduling graph, (3) building a 'large scale system polymerization decomposition idea'-based cascade reservoir flood control water level real-time dynamic control model according to the coupling principle of long-term and short-term scheduling, and using a sequential optimization method to optimize and obtain a cascade reservoir flood control level real-time dynamic control scheme. The cascade reservoir flood control water level real-time dynamic control method can conduct unified schedule to all reservoirs in the upstream and the downstream of a cascade reservoir group, improve the power benefit of the cascade reservoir to the maximum under the premise that flood control safety of the cascade reservoir is guaranteed, is suitable for cascade reservoirs or reservoir groups flood resource scheduling, and can be widely applied to river basin cascade reservoir flood control water level real-time dynamic control.

Description

A kind of step reservoir flood real-time dynamic control method of restricting water supply

Technical field

The invention belongs to step reservoir scheduling field, particularly a kind of step reservoir flood real-time dynamic control method of restricting water supply.

Background technology

Enter 21 century, building up and coming into operation along with the large quantities of hydro plant with reservoirs of China, Chinese Water Conservancy hydroelectric project entered into by the crucial transitional period of building to management operating, carrying out hydropower station group combined dispatching is the major action of complying with " energy-saving power generation " and " utilization of flood resources " Times ' Demand, has important theory value and realistic meaning.Dynamic control of limitation level in flood season is one of important non-engineering measure realizing " utilization of flood resources ".Along with the significantly lifting of the medium-term and long-term numerical value weather forecast of increase and the basin technology of reservoir quantity (dimension) in water reservoir system, need the information considered more and more, the restrict water supply dynamic control of position of flood also will become more complicated.

At present both at home and abroad few to the restrict water supply method of position research of step reservoir flood, existing research method majority is the flood of single reservoir to be restricted water supply to a research method is nested simply enters step reservoir, does not consider the restrict water supply mutual coordination problem of position of flood between storage capacity compensation problem between upstream and downstream reservoir and reservoir.The Guo Sheng of Wuhan University practices teach problem group the restrict water supply dynamic control problem of position of step reservoir flood has been carried out to systematic research, has successively proposed successively progressive Coordination Model of multi-reservoir flood control compensation combined dispatching based on forecast and storage capacity compensation ^[1], the progressive compensation scheduling model successively of the step reservoir dynamic control of limitation level in flood season based on forecast and storage capacity compensation ^[2]with the step reservoir flood based on " polymerization reservoir " restrict water supply a co-design with use scheduling model ^[3,4].Consider that actual reservoir operation is the rolling process forward of " forecast, decision-making, enforcement, forecast, decision-making again, implement again " ^[5], known existing step reservoir dynamic control of limitation level in flood season model is only paid attention to current regimen situation of change, does not take the long term variations of warehouse-in runoff into account.

The list of references relating in literary composition is as follows:

[1] Li Wei, Guo Shenglian, Guo Fuqiang, etc. Hydropower Plant Reservoir group controls flood and compensates combined dispatching scale-model investigation and application [J]. Journal of Hydraulic Engineering, 2007,38 (7): 826-831.

[2] Li Wei, Guo Shenglian, Liu Pan, etc. the scale-model investigation of step reservoir dynamic control of limitation level in flood season and utilization [J]. hydroelectric generation journal, 2008,27 (2): 22-28.

[3] Guo Shenglian, Chen Jionghong, Liu Pan. a kind of step reservoir flood limit water level combined application dispatching method: China, CN201110067570.9[P] .2011-9-14.

[4] Guo Shenglian, Chen Jionghong, Li Fei, etc. Qingjian River step reservoir flood restrict water supply a co-design and utilization [J]. hydroelectric generation journal, 2012,31 (4): 6-11.

[5] Qiu Lin, Chen Shouyu. Hydropower Plant Reservoir Real time optimal dispatch model and application thereof [J]. Journal of Hydraulic Engineering, 1997,57 (3): 74-77

Summary of the invention

For the deficiencies in the prior art, the present invention is based on the meteorological hydrological forecast of basin numerical value, a kind of compensation of the storage capacity between upstream and downstream reservoir, step reservoir flood long-term and that short term scheduling is coupled real-time dynamic control method of restricting water supply of considering has been proposed.

For solving the problems of the technologies described above, the present invention adopts following technical scheme:

A step reservoir flood real-time dynamic control method of restricting water supply, comprises the following steps:

Step 1, sets up the step reservoir basin meteorological hydrologic forecast model of numerical value in flood season, and the forecast basin peb process in 1～7 day future that rolls; The meteorological hydrologic forecast model of described numerical value is comprised of numerical value Meteorological Forecast Model and hydrologic forecast model;

Step 2, based on " large system polymerization thought ", build the randomness Long-term Optimal Dispatch graph model of step reservoir, and adopt self-adapted genetic algorithm to obtain the Long-term Optimal Dispatch figure of step reservoir, based on Long-term Optimal Dispatch figure, obtain the Long-term Optimal Dispatch strategy of polymerization reservoir; Described Long-term Optimal Dispatch graph model is the randomness Optimized model building based on " large system polymerization thought ";

Step 3, according to the principles in coupling of the Long-term Optimal Dispatch of polymerization reservoir and short term scheduling and the meteorological hydrologic forecast model of numerical value, the step reservoir flood of structure based on " large system polymerization Idea of Classification " Real-time dynamic control model of restricting water supply, obtains a step reservoir flood real-time dynamic control case of restricting water supply according to a step reservoir flood Real-time dynamic control model of restricting water supply.

The meteorological hydrologic forecast model of numerical value in step 1 is based on numerical value weather forecast and distributedly ooze ability hydrological model under variable and set up, and valid time can reach 1～7 day, and numerical value weather forecast is used for forecasting the Meteorological Characteristics such as rainfall, temperature.

The randomness Long-term Optimal Dispatch graph model based on " large system polymerization thought " structure step reservoir in step 2 further comprises sub-step:

2-1a obtains virtual polymerization reservoir based on " large system polymerization thought " polymerization step reservoir;

2-2a with the period at the beginning of accumulation of energy and be carved into while facing and can represent polymerization reservoir running status, the period Mo accumulation of energy of take is decision variable, build relate to polymerization reservoir adjacent time interval enter can correlation randomness Long-term Optimal Dispatch model, and definite constraints.

The Long-term Optimal Dispatch figure that employing self-adapted genetic algorithm in step 2 obtains step reservoir further comprises sub-step:

2-1b adopts genetic algorithm to generate at random the initial schedule line of polymerization reservoir;

2-2b initial schedule line produces new scheduling line through individual variation, intersection and selection, calculates the fitness of polymerization reservoir initial schedule line and new scheduling line; Described fitness is the target function value of randomness Long-term Optimal Dispatch graph model, the annual average power generation that described object function is step reservoir.

Whether the fitness judgement new scheduling line of 2-3b based on scheduling line restrains, if convergence, described new scheduling line is the Long-term Optimal Dispatch figure of polymerization reservoir, otherwise repeating step 2-2b.

In step 2, the Long-term Optimal Dispatch strategy of gained polymerization reservoir is:

s _*(t+1)＝Opt(u(t),s(t),t)

In formula,

S _*(t+1) for the Long-term Optimal Dispatch strategy of polymerization reservoir t+1 period;

(u (t), s (t), t) dispatches t period optimal policy for polymerization reservoir to Opt for a long time.

Long-term Optimal Dispatch in step 3 and the principles in coupling of Short-term Optimal Operation are expressed as:

$s(T+1)=\frac{{\mathrm{\τ}}_t}{T_y}\mathrm{Opt}(u\left(t\right),s\left(t\right),t)+\frac{T_y-{\mathrm{\τ}}_t}{T_y}\mathrm{Opt}(u(t+1),s(t+1),t+1)$

Wherein,

T is last period of short term scheduling;

S (T+1) is the period Mo accumulation of energy of " polymerization reservoir ";

T _yfor valid time, its value is 1～7 day;

τ _tfor at leading time T _yinside belong to the time span in the Long-term Optimal Dispatch t period.

S (t), s (t+1) are polymerization reservoir t, accumulation of energy at the beginning of the t+1 period;

Opt (u (t), s (t), t), (u (t+1), s (t+1) t+1) are respectively polymerization reservoir and dispatch for a long time t, t+1 period optimal policy Opt;

U (t), u (t+1) enter polymerization reservoir t, t+1 period energy, $u(t+1)=\frac{1}{T_y-{\mathrm{\τ}}_t}\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{I}_j^y\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{2i}\left(t\right)\right)\mathrm{\Δt},$ be j reservoir and the interval prediction process that becomes a mandarin, L is the number of reservoir in step reservoir, c _{j, i}for waterpower incidence matrix value, K _ibe the power factor of i reservoir, H _2i(t) be the average productive head of i reservoir t period, Δ t is that calculation interval is long, and i, j are the numbering of each reservoir in step reservoir.

Compared with prior art, the present invention has the following advantages and effect:

1, the present invention can carry out United Dispatching to each reservoir of step reservoir upstream and downstream, guaranteeing under the prerequisite of step reservoir flood control safety, by long-term generation schedule and short-term Real-Time Scheduling efficient coupling, can improve to greatest extent the emerging sharp benefit of step reservoir, suitable application in the scheduling of step reservoir or multi-reservoir utilization of flood resources, can be widely used in the cascaded reservoirs flood position of restricting water supply and dynamically control in real time;

2, prior art all turns to optimization aim with multi annual average benefit maximum, and the present invention emphasizes, based on the basin meteorological hydrological forecast of numerical value in flood season, cascaded reservoirs to be put in storage to the valid time T of runoff _yextend to 7 days, the flood of seeking the benefit value maximum of step reservoir within a 7 days valid times real-time dynamic control case of restricting water supply, has more practicality in practice.

Accompanying drawing explanation

Fig. 1 is the step reservoir flood of the present invention in real time dynamically control flow chart of position of restricting water supply;

Fig. 2 is the polymerization Long-term Optimal Regulation for Reservoir figure in this concrete enforcement.

The specific embodiment

The present invention is based on the meteorological hydrological forecast of numerical value, principles in coupling according to long-term and short term scheduling, set up a step reservoir flood Real-time dynamic control model of restricting water supply, and based on a step reservoir flood Real-time dynamic control model of restricting water supply, each reservoir of step reservoir upstream and downstream is carried out to United Dispatching, guaranteeing under the prerequisite of step reservoir flood control safety, the step reservoir flood of seeking a comprehensive utilization benefit maximum real-time dynamic control case of restricting water supply, its idiographic flow refers to Fig. 1.

Below by embodiment, and by reference to the accompanying drawings, the present invention will be further described.

A step reservoir flood real-time dynamic control method of restricting water supply, comprises the following steps:

Step 1, based on numerical value weather forecast and hydrological model, set up the step reservoir basin meteorological hydrologic forecast model of numerical value in flood season, the meteorological hydrologic forecast model of numerical value can roll and forecast the peb process in the leading time of basin, and the valid time of the meteorological hydrologic forecast model of numerical value can reach 1～7.

At present, flood forecasting prediction generally be take " throughfall " as forecast basis, and leading time is limited; And numerical value Meteorological Models can obtain the weather informations such as following contingent rainfall in advance, by coupling weather forecast result and hydrological distribution model, can extend the leading time of flood forecasting, for flood decision tries to gain time precious to one, for correctly making Flood Control Dispatch decision-making, provide scientific basis, can reduce or remit flood loss, increase generating retaining etc., obtain huge economic benefit and social benefit.

Reliable numerical value weather forecast precision, for realizing the coupling of weather forecast and hydrological forecast, effectively extend flood forecasting leading time scientific basis is provided.The present embodiment has been set up Japan, Germany and artificial multiple numerical value Meteorological Models and has been oozed the coupling mechanism of ability (VIC) hydrological model with distributed under variable, thereby obtains the step reservoir basin numerical value meteorology hydrologic forecast model in flood season.Utilize the various weather forecasts output of numerical value Meteorological Models, as, time segment length be 1h rainfall, temperature process, drive distributed VIC hydrological model, continuous analog and the real-time prediction of realization to step reservoir river basin flood process.

In the situation that leading time is 7 days, tradition flood forecasting has almost been lost the prediction ability to flood peak, but the meteorological hydrologic forecast model of numerical value still can forecast the Flood Information that basin may occur future, therefore, the meteorological hydrologic forecast model of numerical value just can be predicted flood generation event before 7 days, can realize the meteorological real-time hydrological forecasting in basin, thereby effectively extend the leading time of flood forecasting, for carrying out the work such as Analysis on flood control situation, reservoir operation, play positive effect.

Step 2, based on " large system polymerization thought ", build the randomness Long-term Optimal Dispatch graph model of step reservoir, and adopt self-adapted genetic algorithm to obtain the Long-term Optimal Dispatch figure of step reservoir, based on Long-term Optimal Dispatch figure, polymerization reservoir is carried out to operation simulation and obtain Long-term Optimal Dispatch strategy.

This step is further comprising the steps:

1) build the randomness Long-term Optimal Dispatch graph model of step reservoir

First this step obtains virtual polymerization reservoir based on " large system polymerization thought " polymerization step reservoir, and inquires into the Optimized Operation figure of polymerization reservoir, thereby provides long-term schedule information for the Real-Time Scheduling of step reservoir.

Polymerization reservoir running status by the period at the beginning of accumulation of energy and be carved into the vector representation that can form while facing, take period Mo accumulation of energy as decision variable, foundation consider polymerization reservoir adjacent time interval enter can correlation randomness Long-term Optimal Dispatch model (take 5～October of flood season as schedule periods, week is scheduling slot t) contrary recurrence equation, as follows:

$\left\{\begin{array}{c}{F}_t(u\left(t\right),s\left(t\right))=\max \{R_t(u\left(t\right),s\left(t\right),s(t+1))+\underset{k=1}{\overset{M}{\mathrm{\Σ}}}{p}_k(t+1)F_{t+1}^{*}(u(t+1),s(t+1))\}\\ F_t^{*}\left(s\left(t\right)\right)=\underset{k=1}{\overset{M}{\mathrm{\Σ}}}{p}_k\left(t\right)F_t^{*}(u\left(t\right),s\left(t\right))\end{array}\right.---\left(1\right)$

In formula,

At the beginning of s (t), s (t+1) are respectively the polymerization reservoir t period, last accumulation of energy;

U (t), u (t+1) enter polymerization reservoir t, t+1 period energy;

R _t(u (t), s (t), s (t+1)) is polymerization reservoir t stage period benefit;

M is the dispersed number that enters energy the t period;

K is the discrete sequence number that enters energy the period;

P _k(t), p _k(t+1) being respectively that polymerization reservoir t, the k equal portions of t+1 period enter can transition probability, wherein, and p _k(t)=P (u (t) u (t-1)), p _k(t+1)=P (u (t+1)/u (t)), P (x) is probability-distribution function, u (t-1) enters energy the polymerization reservoir t-1 period;

be respectively polymerization reservoir t, t+1 period to the optimum remaining benefits of dispatching the end of term;

F _t(u (t), s (t)) is for the t period is to the remaining benefits of dispatching the end of term;

for the t period enters optimum remaining benefits in energy to the M that dispatches the end of term is discrete.

To describe the polymerization process of step reservoir in detail below:

In step reservoir, because the water of upper pond can be able to be reused by lower storage basin, when calculating each reservoir and contain electric weight, should be multiplied by the conversion coefficient sum of the whole step reservoirs of this reservoir and downstream thereof so, can be by accumulation of energy s (t) at the beginning of formula (2) the calculating polymerization reservoir t period:

$s\left(t\right)=\underset{j=1}{\overset{L}{\mathrm{\Σ}}}(V_j\left(t\right)-V_j\left(0\right))\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{1i}\left(t\right)\right)---\left(2\right)$

In formula,

V _j(t) be water retention capacity at the beginning of j reservoir t period;

V _j(0) be j reservoir minimum capacity of a reservoir;

L is the number of reservoir in step reservoir;

I, j are reservoir numbering in step reservoir;

H _1i(t) be i the reservoir average water head that retaining state has at the beginning of the t period, wherein, φ _ibe the water level storage-capacity curve of i reservoir, H _1i(0) be the downstream head of i reservoir; V _i(t), V _i(0) be respectively water retention capacity and minimum capacity of a reservoir at the beginning of i reservoir t period;

K _iit is the power factor of i reservoir;

C _{j, i}for the concrete value of waterpower incidence matrix, its value rule is shown in formula (3)～(5):

c _j，i＝λ(X _j，i,Y _j,i)i,j∈[1,L]（3）

$X_{j,i}=\left\{\begin{array}{c}1\\ 0\end{array}\right.,Y_{j,i}=\left\{\begin{array}{c}1\\ 0\end{array}\right.---\left(4\right)$

In formula (3)～(5),

When reservoir j is during in reservoir i downstream, X _j,ivalue is 0, otherwise is 1;

When reservoir i, j have hydraulic connection, Y _j,ivalue is 1, otherwise is 0;

Work as X _{j, i}=Y _j,i=1 o'clock, c _{j, i}be 1, otherwise, c _{j, i}be 0.

Can calculate equally that the polymerization reservoir t period enters can u (t):

$u\left(t\right)=\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{I}_j\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{2i}\left(t\right)\right)\mathrm{\Δt}---\left(6\right)$

In formula,

L is the number of reservoir in step reservoir;

I, j are reservoir numbering in step reservoir;

K _iit is the power factor of i reservoir;

C _{j, i}for the concrete value of waterpower incidence matrix, its value rule is shown in formula (3)～(5);

I _j(t) be the average reservoir inflow of j reservoir t period, I _j(t)=Q _j-1(t)+QJ _j(t), Q _j-1(t) be the generating flow of j-1 reservoir t period, QJ _j(t) be the local inflow between j-1 reservoir and j reservoir, j-1 reservoir and j reservoir are adjacent reservoir;

H _2i(t) be the average productive head of i reservoir t period, φ _ibe the water level storage-capacity curve of i reservoir, V _i(t), V _i(t+1) be respectively i reservoir t, water retention capacity at the beginning of the t+1 period, be i reservoir level of tail water flow curve, Q _i(t) be the generating flow of i reservoir t period, H _2i(0) be the head loss of i reservoir;

Δ t is that calculation interval is long.

It is as follows that polymerization reservoir goes out energy r (t) in the t period:

$r\left(t\right)=\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{Q}_j\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{2i}\left(t\right)\right)\mathrm{\Δt}---\left(7\right)$

In formula,

Q _j(t) be the generating flow of j reservoir t period;

L is the number of reservoir in step reservoir;

I, j are reservoir numbering in step reservoir;

H _2i(t) be the average productive head of i reservoir t period, φ _ibe the water level storage-capacity curve of i reservoir, V _i(t), V _i(t+1) be respectively i reservoir t, water retention capacity at the beginning of the t+1 period, be i reservoir level of tail water flow curve, Q _i(t) be the generating flow of i reservoir t period, H _2i(0) be the head loss of i reservoir;

C _{j, i}for the concrete value of waterpower incidence matrix, its value rule is shown in formula (3)～(5);

K _iit is the power factor of i reservoir;

Δ t is that calculation interval is long.

It is as follows that polymerization reservoir period t abandons energy w (t):

$w\left(t\right)=\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{W}_j\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{2i}\left(t\right)\right)\mathrm{\Δt}---\left(8\right)$

In formula,

W _j(t) be the discharge of abandoning of j reservoir period t;

H _2i(t) be the average productive head of i reservoir t period, wherein, φ _ibe the water level storage-capacity curve of i reservoir, V _i(t), V _i(t+1) be respectively i reservoir t, water retention capacity at the beginning of the t+1 period, be i reservoir level of tail water flow curve, Q _i(t) be the generating flow of i reservoir t period, H _2i(0) be the head loss of i reservoir;

C _{j, i}for the concrete value of waterpower incidence matrix, its value rule is shown in formula (3)～(5);

K _iit is the power factor of i reservoir;

L is the number of reservoir in step reservoir;

I, j are reservoir numbering in step reservoir;

Δ t is that calculation interval is long.

Step reservoir converts virtual polymerization reservoir to and depends on to a great extent net water head, and the variation of net water head is usually very large, therefore need to construct the actual power flow function of polymerization reservoir in the t period.If the scheduling rule of polymerization reservoir is piecewise linear function figure, as shown in Figure 2, in this figure, abscissa represents polymerization reservoir t period utilizable energy power s (t)+u (t), and ordinate represents the actual power generation d (t) of polymerization reservoir t period, for dropping on a B (x _b, y _b) and C (x _c, y _c) between point, can adopt formula (9) to calculate the actual power generation of polymerization reservoir t period:

$d\left(t\right)=y_B+\frac{y_C-y_B}{x_C-x_B}((s\left(t\right)+u\left(t\right))-x_B)---\left(9\right)$

In formula,

D (t) is the actual power generation of polymerization reservoir t period, d (t)=D (t) Δ t, and D (t) exerted oneself for the step reservoir t period, and Δ t is that calculation interval is long;

S (t)+u (t), for accumulation of energy at the beginning of the polymerization reservoir t period and t period enter energy sum, can be regarded as t period available energy.

So far, step reservoir has been polymerized to 1 virtual polymerization reservoir, polymerization reservoir key element comprises s (t), s (t+1), u (t), r (t), w (t) and d (t)).

The constraints of determining the randomness Long-term Optimal Dispatch model of polymerization reservoir is as follows:

(1) polymerization reservoir energy balance constraint, can be represented by following formula:

s(t+1)＝s(t)+u(t)-r(t)-w(t)（10）

In formula,

At the beginning of s (t), s (t+1) are respectively the polymerization reservoir t period, last accumulation of energy;

U (t) enters energy the polymerization reservoir t period;

W (t) abandons energy the polymerization reservoir t period;

R (t) goes out energy for polymerization reservoir in the t period.

(2) polymerization reservoir fraction constraint, can be represented by following formula:

D'(t)＝D(t)-σA(NF-D(t))（11）

In formula,

D ' is (t) for taking into account the exerting oneself of step reservoir t period of fraction constraint;

NF is that step reservoir guarantees to exert oneself;

D (t) exerted oneself for the step reservoir t period;

A is greater than 0 penalty coefficient, and the order of magnitude is 10 ³～10 ⁶;

σ is 0 or 1 variable, and its value rule is: during D (t) >=NF, σ is 0, otherwise is 1.

(3) polymerization reservoir accumulation of energy constraint, can be represented by following formula:

0≤s(t)≤su(t)（12）

In formula,

S (t) is accumulation of energy at the beginning of the polymerization reservoir t period;

Su (t) is for the maximum accumulation of energy of polymerization reservoir t period, and in flood season, su (t) is the flood of each reservoir in the polymerization reservoir accumulation of energy summation that position corresponding storage capacity emptying to minimum capacity of a reservoir has of restricting water supply.

(4) polymerization reservoir goes out and can retrain, and can be represented by following formula:

rl(t)≤r(t)≤ru(t)（13）

In formula,

R (t) goes out energy the polymerization reservoir t period;

Rl (t) for the minimum of polymerization reservoir t period go out can, by the irrigation of each reservoir, water supply, shipping or ecological requirement, determined;

Ru (t) for the maximum of polymerization reservoir t period go out can, by the downstream flood control of each reservoir require, discharge capacity determines.

(5) polywater library partition condition, can be represented by following formula:

s(0)＝s ₀（14）

s(T+1)＝s _T+1（15）

In formula,

S ₀for the just accumulation of energy of polymerization reservoir schedule periods, it is the accumulation of energy summation that in polymerization reservoir, each reservoir operation beginning given storage capacity emptying to minimum capacity of a reservoir has;

S _t+1for polymerization reservoir is in scheduling end of term accumulation of energy, it is the accumulation of energy summation that in polymerization reservoir, the given storage capacity emptying of each reservoir operation end of term has to minimum capacity of a reservoir.

(6) reservoir variable nonnegativity restrictions.

2) the randomness Long-term Optimal Dispatch graph model based on step reservoir obtains the Long-term Optimal Dispatch figure of step reservoir.

Can adopt self-adapted genetic algorithm to obtain the Long-term Optimal Dispatch figure of step reservoir.Can first preset the shape of initial schedule line, as shown in Figure 2, then adopt genetic algorithm to encode to the horizontal stroke of key point, ordinate.Scheduling line to day part, as take week as scheduling slot, only need 4 control point A, B, the horizontal stroke of C, D, ordinate that coding variable is set, namely 8 floating-point encoding variablees.Utilize genetic coding to provide at random the feasible solution of scheduling line, polymerization reservoir moves according to scheduling graph, and statistics operation result, selects optimum scheduling graph, by genetic operators such as intersection, variations, obtains new improvement scheduling line, iterates, until convergence.

Adopt the step of self-adapted genetic algorithm formulation polymerization optimizing scheduling of reservoir figure as follows:

1. adopt genetic algorithm to generate at random the initial schedule line of polymerization reservoir;

2. initial schedule line produces new scheduling line through individual variation, intersection and selection; Calculate the fitness of initial schedule line and new scheduling line; The object function that the annual average power generation of step reservoir is randomness Long-term Optimal Dispatch graph model is take in the present invention, and the fitness of polymerization reservoir operation line is the target function value of randomness Long-term Optimal Dispatch graph model;

3. judge whether new scheduling line meets the condition of convergence of genetic algorithm, and the object function difference that the condition of convergence is adjacent twice iteration is less than or equal to setting accuracy, if meet, restrain output scheduling line, thereby obtain Long-term Optimal Dispatch figure; Otherwise repeating step 2..

3) by Long-term Optimal Dispatch figure, polymerization reservoir is carried out to operation simulation, just can obtain Long-term Optimal Dispatch strategy:

s _*(t+1)＝Opt(u(t),s(t),t)（16）

In formula,

S _*(t+1) for the Long-term Optimal Dispatch strategy of polymerization reservoir t+1 period;

(u (t), s (t), t) dispatches t period optimal policy for polymerization reservoir to Opt for a long time.

Step 3, according to principles in coupling long-term and short term scheduling, sets up step reservoir flood based on " large system polymerization decomposition thought " Real-time dynamic control model of restricting water supply.

Step reservoir flood in the present embodiment is restricted water supply a Real-time dynamic control model with " valid time T _ythe power benefit of interior step reservoir is maximum " as optimization aim, and meet following constraints: the constraint of (1) water balance; (2) constraint of the hydraulic connection between upstream and downstream reservoir; (3) reservoir level constraint; (4) outbound flow restriction; (5) output of power station constraint.In addition, also need to meet reservoir border constraint

$\left\{\begin{array}{c}{Z}_i\left(0\right)=Z_i^0\\ Z_i(T+1)=Z_i^{T+1}\end{array}\right.---\left(17\right)$

In formula,

Z _i(0), Z _i(T+1) be respectively schedule periods just and last water level;

be respectively schedule periods just and last given water level.

According to formula (2), the scheduling end of term water level in each reservoir border constraint formula is polymerized to period Mo accumulation of energy s (T+1):

$s(T+1)=\underset{j=1}{\overset{L}{\mathrm{\Σ}}}(V_j(T+1)-V_j\left(0\right))\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{1i}(T+1)\right)---\left(18\right)$

In formula,

V _j(T+1) be j the corresponding storage capacity of reservoir operation end of term water level;

V _j(0) be the corresponding storage capacity of a j water dead water level;

L is the number of reservoir in step reservoir;

I, j are reservoir numbering in step reservoir;

H _1i(T+1) be the average water head of i reservoir operation end of term retaining state, wherein, φ _ibe the water level storage-capacity curve of i reservoir, H _1i(0) be the downstream head of i reservoir; V _i(T+1), V _i(0) be respectively i water scheduling end of term water retention capacity and minimum capacity of a reservoir;

K _iit is the power factor of i reservoir;

C _{j, i}for the concrete value of waterpower incidence matrix, its value rule is shown in formula (3)～(5).

Set up the Real time optimal dispatch model that " polymerization reservoir " combines with Short-term Optimal Operation for a long time, key is how to utilize randomness Long-term Optimal Dispatch strategy formula (16), determine the period Mo accumulation of energy s (T+1) in Model of Short-term Optimal Dispatch, long and short phase Optimized Operation is connected, to be reflected in, in short term scheduling, how to consider that the statistics variations rule of runoff is taken into account the long-term benefit that step reservoir is dispatched.

If cascaded reservoirs is put runoff leading time T flood season in storage _y=7 days can be the period discretely to take the simple Markov Chain that be dispatching cycle flood season by week by runoff process, sets up the optimisation strategy formula (16) of the randomness Long-term Optimal Dispatch model of " polymerization reservoir ".Be located at the t of t in the period _cconstantly forecast, each reservoir and the interval prediction process of becoming a mandarin are leading time is T _y, can obtain according to the meteorological hydrologic forecast model of the constructed numerical value of step 1.The period Mo accumulation of energy s (T+1) in the Model of Short-term Optimal Dispatch of " polymerization reservoir " determines by following rule:

$s(T+1)=\frac{{\mathrm{\τ}}_t}{T_y}\mathrm{Opt}(u\left(t\right),s\left(t\right),t)+\frac{T_y-{\mathrm{\τ}}_t}{T_y}\mathrm{Opt}(u(t+1),s(t+1),t+1)---\left(19\right)$

In formula,

$u\left(t\right)=\frac{1}{{\mathrm{\τ}}_t}\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{I}_j^y\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}{c}_{j,i}{K}_i{H}_{2i}\left(t\right)\mathrm{\Δt};$

$u(t+1)=\frac{1}{T_y-{\mathrm{\τ}}_t}\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{I}_j^y\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}{c}_{j,i}{K}_i{H}_{2i}\left(t\right)\mathrm{\Δt};$

τ _tfor at leading time T _yinside belong to the time span in Long-term Optimal Dispatch period t.

Due to the Z in r (t) and formula (17) _i(0) be known, just can let out principle according to Flood Control Dispatch rules and reservoir classification control, the optimal Decomposition strategy of the power benefit maximum of the step reservoir of seeking to send as an envoy to.Based on " thought is decomposed in large system polymerization ", first according to the hydraulic connection between upstream and downstream reservoir and each flood control control point flood control standard, set up the storage capacity upper limit relation of holding in advance of upstream and downstream step reservoir, until each reservoir is at valid time T _ythe storage capacity upper limit of holding in advance meet equation (19).Each initial period according to weather report information obtain one group of optimal policy, and along with forecast information roll to be constantly updated strategy, the power benefit of step reservoir is maximized.Step reservoir flood in the present embodiment restrict water supply position in real time dynamically control problem belong to multidimensional multistage Optimal Decision-making problem, useful nonlinear optimization method solves, and adopts successively optimization optimization comparatively ripe in dynamic programming problems to obtain a step reservoir flood real-time dynamic control case of restricting water supply.

Claims (8)

1. a step reservoir flood real-time dynamic control method of restricting water supply, is characterized in that, comprises the following steps:

Step 1, sets up the step reservoir basin meteorological hydrologic forecast model of numerical value in flood season, and the forecast basin peb process in 1～7 day future that rolls;

Step 2, based on " large system polymerization thought ", build the randomness Long-term Optimal Dispatch graph model of step reservoir, and adopt self-adapted genetic algorithm to obtain the Long-term Optimal Dispatch figure of step reservoir, based on Long-term Optimal Dispatch figure, obtain the Long-term Optimal Dispatch strategy of polymerization reservoir;

Step 3, according to the principles in coupling of the Long-term Optimal Dispatch of polymerization reservoir and short term scheduling and the meteorological hydrologic forecast model of numerical value, the step reservoir flood of structure based on " large system polymerization decompose a thought " Real-time dynamic control model of restricting water supply, obtains a step reservoir flood real-time dynamic control case of restricting water supply according to a step reservoir flood Real-time dynamic control model of restricting water supply.

2. a step reservoir flood according to claim 1 real-time dynamic control method of restricting water supply, is characterized in that:

The meteorological hydrologic forecast model of numerical value in step reservoir basin flood season of described step 1 is based on numerical value weather forecast and distributedly ooze ability hydrological model under variable and build.

3. a step reservoir flood according to claim 1 real-time dynamic control method of restricting water supply, is characterized in that:

The described randomness Long-term Optimal Dispatch graph model based on " large system polymerization thought " structure step reservoir further comprises sub-step:

2-1a obtains virtual polymerization reservoir based on " large system polymerization thought " polymerization step reservoir;

2-2a with the period at the beginning of accumulation of energy and be carved into while facing and can represent polymerization reservoir running status, the period Mo accumulation of energy of take is decision variable, build relate to polymerization reservoir adjacent time interval enter can correlation randomness Long-term Optimal Dispatch model, and definite constraints.

4. a step reservoir flood according to claim 1 real-time dynamic control method of restricting water supply, is characterized in that:

The Long-term Optimal Dispatch figure that described employing self-adapted genetic algorithm obtains step reservoir further comprises sub-step:

2-1b adopts genetic algorithm to generate at random the initial schedule line of polymerization reservoir;

2-2b initial schedule line produces new scheduling line through individual variation, intersection and selection, calculates the fitness of polymerization reservoir initial schedule line and new scheduling line;

Whether the fitness judgement new scheduling line of 2-3b based on scheduling line restrains, if convergence, described new scheduling line is the Long-term Optimal Dispatch figure of polymerization reservoir, otherwise repeating step 2-2b.

5. a step reservoir flood according to claim 4 real-time dynamic control method of restricting water supply, is characterized in that:

The fitness of described polymerization reservoir operation line is the target function value of randomness Long-term Optimal Dispatch graph model, the annual average power generation that described object function is step reservoir.

6. a step reservoir flood according to claim 1 real-time dynamic control method of restricting water supply, is characterized in that:

The Long-term Optimal Dispatch strategy of described polymerization reservoir is:

s _*(t+1)＝Opt(u(t),s(t),t)

In formula,

S _*(t+1) for the Long-term Optimal Dispatch strategy of polymerization reservoir t+1 period;

(u (t), s (t), t) dispatches t period optimal policy for polymerization reservoir to Opt for a long time.

7. a step reservoir flood according to claim 1 real-time dynamic control method of restricting water supply, is characterized in that: described Long-term Optimal Dispatch and the principles in coupling of short term scheduling are:

$s(T+1)=\frac{{\mathrm{\τ}}_t}{T_y}\mathrm{Opt}(u\left(t\right),s\left(t\right),t)+\frac{T_y-{\mathrm{\τ}}_t}{T_y}\mathrm{Opt}(u(t+1),s(t+1),t+1)$

Wherein,

S (T+1) is the period Mo accumulation of energy of " polymerization reservoir ";

T is last period of short term scheduling;

T _yfor valid time, its value is 1～7 day;

τ _tfor at leading time T _yinside belong to the time span in the Long-term Optimal Dispatch t period.

S (t), s (t+1) are polymerization reservoir t, accumulation of energy at the beginning of the t+1 period;

Opt (u (t), s (t), t), (u (t+1), s (t+1) t+1) are respectively polymerization reservoir and dispatch for a long time t, t+1 period optimal policy Opt;

U (t), u (t+1) enter polymerization reservoir t, t+1 period energy, $u(t+1)=\frac{1}{T_y-{\mathrm{\τ}}_t}\underset{j=1}{\overset{L}{\mathrm{\Σ}}}{I}_j^y\left(t\right)\underset{i=1}{\overset{L}{\mathrm{\Σ}}}\left(c_{j,i}{K}_i{H}_{2i}\left(t\right)\right)\mathrm{\Δt},$ be j reservoir and interval t period to forecast the process that becomes a mandarin, L is the number of reservoir in step reservoir, c _j,ifor waterpower incidence matrix value, K _ibe the power factor of i reservoir, H _2i(t) be the average productive head of i reservoir t period, Δ t is that calculation interval is long, and i, j are the numbering of each reservoir in step reservoir.

8. a step reservoir flood according to claim 1 real-time dynamic control method of restricting water supply, is characterized in that:

A described step reservoir flood optimization aim for real-time dynamic control case of restricting water supply is valid time T _ythe power benefit of interior step reservoir is maximum.