CN105825437A - Complicated reservoir group common water supply task distribution method - Google Patents

Abstract

The present invention discloses a complicated reservoir group common water supply task distribution method. The method comprises the following steps of 1 according to the topology structure of a reservoir group and the distribution condition of a common water supply objective, hierarchically establishing a virtual aggregated reservoir, and drawing a corresponding joint scheduling graph; 2 determining a common water supply task distribution strategy aiming at the parallel and series two topology structures separately; 3 establishing an objective function of a joint water supply scheduling model; 4 adopting a multi-objective evolution algorithm to optimize the objective function of the joint water supply scheduling model, thereby obtain a relative optimal joint scheduling rule. The complicated reservoir group common water supply task distribution method of the present invention can realize the complicated reservoir group joint water supply uniform scheduling, can capture the space-time difference of the reservoir runoffs, fully plays the reservoir's capacity compensation and hydrology compensation roles of the reservoir group, enables the optimization configuration capability of a system to the water resource to be improved furthest, is suitable for being applied to the complicated reservoir group joint scheduling, and can be widely used for the complicated reservoir group common water supply task distribution.

Description

A kind of complicated multi-reservoir supplies water method for allocating tasks jointly

Technical field

The invention belongs to multi-reservoir combined dispatching technical field, be related specifically to a kind of complicated multi-reservoir and jointly supply water method for allocating tasks.

Background technology

Chinese large-sized river basin has formed or has been formed multi-reservoir and developed jointly the total arrangement utilized, joint water supply of reservoir group Optimized Operation is a Nonlinear Multiobjective optimization problem with Complex Constraints, but, along with storehouse group's scale and the continuous increase of topological structure complexity thereof, it is achieved the difficulty of its Optimized Operation is also constantly increasing.At present, by virtual for multi-reservoir for one of polymerization reservoir method best approach being considered multi-reservoir combined dispatching formulating water supply decision.Although the method can consider the water-retention state of multi-reservoir, Exact Location System water supply capacity, determine that system supplies confluent to each water user more objectively, but common water supply task is the most optimally assigned to corresponding concrete reservoir and is still difficult point place.

At present, the distribution of common water supply task mainly uses point water rule of three and compensates regulation method.When multi-reservoir topological structure is complicated or compensation relationship is inconspicuous frequently with a point water rule of three, wherein based on fixed proportion Y-factor method Y and dynamic proportion Y-factor method Y.Fixed proportion Y-factor method Y mainly determines by the way of simulative optimization that one group of fixed proportion coefficient distributes common water supply task；Dynamic proportion Y-factor method Y is mainly according to the thought of " compensate regulation, able people should do more work ", consider current storage capacity, storage coefficient, reservoir become a mandarin etc. because usually formulating rational allocation rule, thus better profit from compensation of hydrology and storage capacity Compensation Rule [Zhou Fen, Zheng Xiongwei, Ma Jun, etc. Advances In Science And Technology of Water Resources, 2011,05:11-13+22.], [Zhang Haotian. Dalian University of Technology, 2013].But existing dynamic proportion Y-factor method Y major part does not accounts for the water supply task of each member's reservoir present period self, when reservoir self water supply task is the biggest, its change procedure can largely affect and restrict the distribution of common water supply task.

When multi-reservoir exists obvious compensation relationship frequently with compensating regulation method: [the JayR.Lund such as JayR.Lund, JoelGuzman.JournalofWaterResourcesPlanningandManagement, 1999,125 (3): 143-153.] scheduling of supplying water is carried out successively by storage capacity utilization ratio value order from small to large；Li-ChiuChang etc. [Li-ChiuChang, Fi-JohnChang.JournalofHydrology, 2009,1 (2): 12-20.] utilize the mode compensating regulation that Aeschna melanictera, jade multi-reservoir are carried out scheduling of supplying water；Guo Xuning [Guo Xuning, Hu Tiesong, Zeng Xiang, etc. Central China University of Science and Technology's journal: natural science edition, 2011,39 (10): 121-124.] etc. utilize and compensate the pattern that regulation combines with two dimension scheduling graph green stream river, Yingna River basin are carried out supplying water and dispatch.But existing compensation regulation method is not owing to accounting for the water supply task of self, often occur compensating not enough or over compensation situation, it is impossible to rational management ensures to supply water.

Additionally, multiple targets such as whole system water supply target, member's reservoir water supply target, each water user's water supply target are related to due to the distribution of common water supply task, traditional index weights assignment method has certain limitation, it is simply converted to single goal multiple target, competition conspiracy relation [the Li-ChiuChang between each target cannot be embodied, Fi-JohnChang.JournalofHydrology, 2009,1 (2): 12-20.], [Ding Shengxiang, Dong Zengchuan, Wang Dezhi, etc. hydroscience is in progress, 2008,19 (5): 679-684.].

Summary of the invention

For the deficiencies in the prior art, present invention thought based on virtual aggregation reservoir, propose a kind of complicated multi-reservoir and jointly supply water method for allocating tasks.

For solving above-mentioned technical problem, the present invention adopts the following technical scheme that

A kind of complicated multi-reservoir supplies water method for allocating tasks jointly, comprises the following steps:

The first step, according to topological structure and the distribution situation of common water supply target of multi-reservoir, layering builds virtual aggregation reservoir, and works out corresponding combined dispatching figure.

Second step, is respectively directed to two kinds of topological structure in parallel and serial, determines the Task Assigned Policy that jointly supplies water；Parallel reservoir allocation strategy uses the dynamic proportion Y-factor method Y in point water rule of three；Connection reservoirs allocation strategy uses and compensates regulation method, for avoiding over compensation, arranges supply restraining line at compensation reservoir.

The computing formula of described dynamic proportion coefficient is:

$K_{i,t}=({\mathrm{VS}}_{i,t}+{\mathrm{IF}}_{i,t}-{\mathrm{DP}}_{i,t})/\underset{i}{\overset{N}{\Σ}}({\mathrm{VS}}_{i,t}+{\mathrm{IF}}_{i,t}-{\mathrm{DP}}_{i,t})---\left(1\right)$

Wherein, K_i,tFor partition coefficient, VS_i,tFor the reservoir storage of i reservoir t period, IF_i,tFor becoming a mandarin of i reservoir t period, DP_i,tSelf water supply task for the i reservoir t period.The common water supply task that each member's reservoir is responsible for is equal to the partition coefficient K of this member's reservoir_i,tIt is multiplied by common water supply task.

3rd step, builds the object function of joint water supply dispatching model, and described object function is:

The object function of 3.1 subsystem hydropenia indexes:

$f\left(x\right)=SI={\mathrm{\ω}}_{agr}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{agr,j}-W_{agr,j}}{D_{agr,j}}\right)}^2+{\mathrm{\ω}}_{ind}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{ind,j}-W_{ind,j}}{D_{ind,j}}\right)}^2---\left(2\right)$

Wherein, f (x) is the object function of subsystem hydropenia index, and SI is hydropenia index, and N is length series scheduling total year number, ω_agrFor the weight shared by agricultural water target, ω_indFor the weight shared by water for industrial use target, D_agr,j、D_ind,jIt is respectively subsystem agricultural water target, the gross water requirement of water for industrial use target, W in j_agr,j、W_ind,jFor subsystem in j to agriculture, industrial total supply.

3.2 systems always abandon the object function of the water yield:

$f^{\′}\left(x\right)=\frac{1}{N}{\mathrm{\Σ}}_{j=1}^N{\mathrm{SU}}_j+\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\α}}_i{\mathrm{Ra}}_i\·PS\_A+\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\β}}_i\·PS\_B---\left(3\right)$

Wherein, f'(x) it is the system object function of always abandoning the water yield, SU_jFor abandoning the water yield (comprise reservoir abandon water and interval finally enter magnanimity) in j, M is the total number of water user, Ra_iFor the fraction of water user i, α_i、β_iIt is respectively penalty coefficient, the α when water user i is unsatisfactory for fraction requirement_iTake 1, otherwise take 0, β during generation super collapse dept when water user i supplies water_iTaking 1, otherwise take 0, PS_A, PS_B are respectively sufficiently large punishment amount.

4th step, uses multi-objective Evolutionary Algorithm, optimizes the object function of joint water supply dispatching model, obtains relatively optimum combined dispatching rule；The described hydropenia index that combined dispatching rule is subsystems is minimum, system always to abandon the water yield minimum.

The invention have the benefit that and complicated joint water supply of reservoir group can be carried out United Dispatching, the space and time difference of two Phase flow can be caught, the storage capacity giving full play to storehouse group compensates and compensation of hydrology effect, farthest improve system and water resource is distributed rationally ability, suitably the application in complicated multi-reservoir combined dispatching.

Accompanying drawing explanation

Fig. 1 is complicated multi-reservoir schematic diagram；

Fig. 2 is reservoir water supply scheduling graph.

Detailed description of the invention

Below by embodiment, and combine accompanying drawing, technical scheme is further elaborated with.

A kind of complicated multi-reservoir supplies water method for allocating tasks jointly, comprises the following steps:

The first step, according to topological structure and the distribution situation of common water supply target of multi-reservoir, layering builds virtual aggregation reservoir, and works out corresponding combined dispatching figure.

Complicated multi-reservoir water supply schematic diagram as shown in Figure 1, its water user is numerous, way of supplying water is complex, whole multi-reservoir water system is made up of three subsystems: a reservoir and a mining under reservoir interval are subsystem A, b reservoir, b-c reservoir interval, c reservoir and c mining under reservoir interval are subsystem B, and the interval after reservoir a, b, c converge is subsystem C.

Subsystem A is supplied water by a reservoir, therefore optimizes the scheduling rule formulating a reservoir, and how determiner system A each water user supply water.Subsystem B is supplied water by b, c reservoir, and therefore two storehouse virtual aggregation become a reservoir (XN-2) determine the output of water user each to this subsystem internal.Three storehouse virtual aggregation, by a, b, c reservoir co-supplying, are therefore become a reservoir (XN-3), determine this subsystem each water user output according to its scheduling rule by subsystem C.

The scheduling rule of a reservoir, XN-2 reservoir and XN-3 reservoir reflects with scheduling graph form, according to water supply target priority and the height of fraction, formulates water supply scheduling graph and the rule that supplies water.As shown in Figure 2, reservoir water supply scheduling graph is made up of the restriction supply line of each water user, and 2 limit supply line and the utilizable capacity of reservoir is divided into 3 dispatch areas.In reservoir running, according to the dispatcher-controlled territory residing for reservoir current water-retention state, the water supply rule be given according to table 1 supplies water.

Supply water rule in table 1 reservoir water supply scheduling graph each district

Second step, is respectively directed to two kinds of topological structure in parallel and serial, determines the Task Assigned Policy that jointly supplies water；Parallel reservoir allocation strategy uses the dynamic proportion Y-factor method Y in point water rule of three；Connection reservoirs allocation strategy uses and compensates regulation method, for avoiding over compensation, arranges supply restraining line at compensation reservoir.

Application virtual aggregation reservoir carries out joint water supply of reservoir group scheduling and generally comprises two steps: (1) is according to each reservoir current water-retention state, determine that multi-reservoir is to limit to supply water or on-demand water supply to each water user, i.e. determine each water user for how much water.(2) according to certain allocation strategy common water supply task of distribution to each member's reservoir, determine and by whom supplied water.Concrete allocation strategy is as follows:

(1) parallel reservoir

To aggregate into XN-3 reservoir be that subsystem C supplies water to a, b, c tri-in storehouse, and b reservoir and c reservoir aggregate into XN-2 reservoir simultaneously, and therefore the common water supply task of XN-3 distribute between a reservoir and XN-2 reservoir, i.e. a reservoir and XN-2 reservoir are parallel reservoir.The basin at a reservoir place keeps with rich with withered state substantially with the basin at XN-2 reservoir place, there is not the richest withered complementary situation；Additionally a reservoir and b reservoir are all carry-over storages, utilizable capacity, aggregate storage capacity and storage coefficient are all more or less the same, the most there is not storage capacity and compensate situation, therefore a point water rule of three is used when XN-3 reservoir distributes common water supply task, the task that makes jointly to supply water effectively is distributed between a reservoir and XN-2 reservoir, and uses fixed proportion Y-factor method Y and two kinds of methods of dynamic proportion Y-factor method Y to be analyzed.

A fixed proportion Y-factor method Y

Use Exchanger Efficiency with Weight Coefficient Method directly to optimize and determine the allocation proportion coefficient (α of day part in scheduling year_t, β_t), if the common water supply task in this period is WD_t, then the common water supply task that a reservoir is responsible for is α_t×WD_t, the common water supply task that XN-2 reservoir is responsible for is β_t×WD_t, and meet constraint: α_t+β_t=1.

B dynamic proportion Y-factor method Y

Existing dynamic proportion Y-factor method Y is improved, considers that water, current storage capacity and present period self water supply task three are because usually formulating dynamic partition coefficient the most simultaneously.Computing formula is as follows:

$K_{i,t}=({\mathrm{VS}}_{i,t}+{\mathrm{IF}}_{i,t}-{\mathrm{DP}}_{i,t})/\underset{i}{\overset{N}{\Σ}}({\mathrm{VS}}_{i,t}+{\mathrm{IF}}_{i,t}-{\mathrm{DP}}_{i,t})---\left(1\right)$

Wherein, K_i,tFor partition coefficient, VS_i,tFor the reservoir storage of i reservoir t period, IF_i,tFor becoming a mandarin of i reservoir t period, DP_i,tSelf water supply task for the i reservoir t period.The common water supply task that each member's reservoir is responsible for is equal to the partition coefficient K of this member's reservoir_i,tIt is multiplied by common water supply task.

(2) connection reservoirs

The water supply task of water supply task and c mining under reservoir interval that XN-3 distributes to XN-2 is commonly fed by b, c reservoir, belongs to the common water supply task of XN-2 reservoir.B reservoir is in the upstream of c reservoir, and its utilizable capacity and storage coefficient are all big than c reservoir, but its reservoir inflow is less than c reservoir on the contrary.Therefore, XN-2 reservoir uses the regulation method that compensates when distributing common water supply task, and i.e. first by the responsible task that jointly supplies water of c reservoir, insufficient section is fed by b reservoir.

Also be responsible for supply by b in view of more than c reservoir interval water supply task and b reservoir self direct-furnish task, i.e. can not unconfined compensations c reservoir, so arranging one at b reservoir to feed restraining line.When b reservoir level is higher than this restraining line, b reservoir can feed c reservoir；When b reservoir level is less than this restraining line, b reservoir can not feed c reservoir, by the water supply task in c reservoir alone bear downstream.

3rd step, builds the object function of joint water supply dispatching model.

For subsystem A, the regulation goal of a reservoir is to ensure that water supply, reduces hydropenia, and therefore its hydropenia index is as one of object function (the least more excellent).Equally, using the hydropenia index of subsystem B, C as two other object function.The purpose of whole joint water supply of reservoir group Optimized Operation is to make reservoir group system total supply maximum, i.e. system always to abandon the water yield minimum, therefore always abandon one of minimum object function of the water yield with the average annual of system.To sum up, object function is:

${\mathrm{minf}}_A\left(x\right)={\mathrm{SI}}_A={\mathrm{\ω}}_{agr}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{A,agr,j}-W_{A,agr,j}}{D_{A,agr,j}}\right)}^2+{\mathrm{\ω}}_{ind}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{A,ind,j}-W_{A,ind,j}}{D_{A,ind,j}}\right)}^2---\left(2.1\right)$

${\mathrm{minf}}_B\left(x\right)={\mathrm{SI}}_B={\mathrm{\ω}}_{agr}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{B,agr,j}-W_{B,agr,j}}{D_{B,agr,j}}\right)}^2+{\mathrm{\ω}}_{ind}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{B,ind,j}-W_{B,ind,j}}{D_{B,ind,j}}\right)}^2---\left(2.2\right)$

${\mathrm{minf}}_C\left(x\right)={\mathrm{SI}}_C={\mathrm{\ω}}_{agr}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{C,agr,j}-W_{C,agr,j}}{D_{C,agr,j}}\right)}^2+{\mathrm{\ω}}_{ind}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{C,ind,j}-W_{C,ind,j}}{D_{C,ind,j}}\right)}^2---\left(2.3\right)$

${\mathrm{minf}}_S\left(x\right)=\frac{1}{N}{\mathrm{\Σ}}_{j=1}^N{\mathrm{SU}}_j+\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\α}}_i{\mathrm{Ra}}_i\·PS\_A+\underset{i=1}{\overset{M}{\Σ}}{\mathrm{\β}}_i\·PS\_B---\left(3\right)$

Wherein, N is length series scheduling total year number；SI is hydropenia index；ω_agrFor the weight shared by agricultural water target；ω_indFor the weight shared by water for industrial use target；D_A,agr,j、D_A,ind,jIt is respectively subsystem A agricultural water target, gross water requirement of water for industrial use target in j；W_A,agr,j、W_A,ind,jFor subsystem A in j to agriculture, industrial total supply；SU_jFor abandoning the water yield (comprise reservoir abandon water and interval finally enter magnanimity) in j；M is the total number of water user；Ra_iFraction for water user i；α_i、β_iIt is respectively penalty coefficient, takes 1 when water user i is unsatisfactory for fraction requirement, otherwise take 0；β during generation super collapse dept when water user i supplies water_iTake 1, otherwise take 0.PS_A, PS_B are respectively sufficiently large punishment amount.

4th step, uses multi-objective Evolutionary Algorithm, optimizes the object function of joint water supply dispatching model, obtains relatively optimum combined dispatching rule；The described hydropenia index that combined dispatching rule is subsystems is minimum, system always to abandon the water yield minimum.

Using multi-reservoir long series natural two Phase flow and local inflow as the input data of model, utilize multi-objective Evolutionary Algorithm that model is optimized and solve, obtain relatively optimum scheduling rule.

(1) decision variable

The decision variable that two kinds of common water supply task allocative decisions are identical is: the scheduling of XN-2, XN-3 and a graph of reservoir operation position of line, the position, supply restraining line of b reservoir.Fixed allocation proportionality coefficient method also needs member's reservoir water supply task allocation proportion in each period of extra variable: XN-3.

(2) constraints

The constraint of reservoir water yield Constraints of Equilibrium, characteristic water level of reservoir, water user limit water supply and do not surpass collapse dept and each water user's fraction constraint etc..

(3) optimization method

The non-dominated sorted genetic algorithm (ε-NSGAII) of the band elitism strategy of ε-domination is introducing ε-domination, the multi-objective Evolutionary Algorithm of self adaptation Population Size strategy on the basis of NSGAII.Compared with other multi-objective Evolutionary Algorithms (MOEAs), this algorithm has three aspect advantages: 1. reduces parameter and arranges；2. ε-domination reduces the probability that cannot restrain；3. self adaptation Population Size.This algorithm was widely used in numerous industry in recent years, and obtained good effect.Therefore, using ε-NSGA-II algorithm to solve multi-reservoir joint optimal operation model, its parameter is arranged as shown in table 2.Owing to genetic algorithm has a randomness, therefore each problem independent operating 6 times, each stochastic generation initial population iteration 100 all ages (it is demonstrated experimentally that behind 100 all ages iteration, it is thus achieved that Pareto disaggregation do not have essence to promote).Owing to true optimum Pareto forward position cannot obtain, therefore obtain from all Pareto forward positions solution that 12 search obtain and replace true optimum Pareto forward position with reference to disaggregation.

Table 2 ε-NSGAII algorithm parameter is arranged

By this method application area Hun River northeastward and the Dahuofang Reservoir on Taizihe River basin, Guanyinge Reservoir and the multi-reservoir of nest reservoir composition, set up 3 kinds of schemes and be analyzed, as shown in table 3.In scheme 1,2, point water rule of parallel reservoir uses fixed proportion Y-factor method Y, and in scheme 3, point water rule of parallel reservoir uses dynamic proportion Y-factor method Y, and in scheme 1, point water rule of connection reservoirs uses and compensates regulation method.In scheme 2,3, point water rule of connection reservoirs uses and considers to feed restrictive compensation regulation method.

Common water supply task method of salary distribution contrast table in 33 kinds of schemes of table

Scheme 1,2 Comparative result shows that compensation regulation method based on supply restraining line can supply water task by classifying rationally jointly, preferably plays storage capacity compensating action；In contrast, do not consider that the regulation method that compensates compensating restraining line makes the number of times of reservoir generation collapse dept increase, affect the task distribution of other reservoirs and system the most further, reduce the water supply fraction of common water supply task.

Scheme 2,3 results contrast shows that dynamic proportion Y-factor method Y has considered reservoir present period reservoir storage, become a mandarin and self water supply task, both reservoir current supplying capability had been considered, it is further contemplated that the accurate forecast information of next period, also contemplate water information in future, the compensation of hydrology between subsystem A and subsystem B and storage capacity compensating action can be better profited from, reduce competitiveness between the two, the preferably water supply task of assignment subsystem C.Compared with the Pareto disaggregation that fixed proportion Y-factor method Y obtains, the Pareto disaggregation of dynamic proportion Y-factor method Y disaggregation number under conditions of multiformity does not reduces significantly reduces, convergence is more preferable, abandon the water yield preferably disaggregation proportion to increase, and preferably equalized the hydropenia index of three subsystems, it is more nearly the position of preferable optimal solution.

Claims (2)

1. a complicated multi-reservoir supplies water method for allocating tasks jointly, it is characterised in that comprise the following steps:

The first step, according to topological structure and the distribution situation of common water supply target of multi-reservoir, layering builds virtual aggregation reservoir, and works out corresponding combined dispatching figure；

Second step, is respectively directed to two kinds of topological structure in parallel and serial, determines the Task Assigned Policy that jointly supplies water: parallel reservoir uses the dynamic proportion Y-factor method Y in point water rule of three；Connection reservoirs uses and compensates regulation method, for avoiding over compensation, arranges supply restraining line at compensation reservoir；

The computing formula of described dynamic proportion coefficient is:

$K_{i,t}=({\mathrm{VS}}_{i,t}+{\mathrm{IF}}_{i,t}-{\mathrm{DP}}_{i,t})/\underset{i}{\overset{N}{\Σ}}({\mathrm{VS}}_{i,t}+{\mathrm{IF}}_{i,t}-{\mathrm{DP}}_{i,t})---\left(1\right)$

Wherein, K_i,tFor partition coefficient, VS_i,tFor the reservoir storage of i reservoir t period, IF_i,tFor becoming a mandarin of i reservoir t period, DP_i,tSelf water supply task for the i reservoir t period；The common water supply task that each member's reservoir is responsible for is equal to the partition coefficient K of this member's reservoir_i,tIt is multiplied by common water supply task；

3rd step, build joint water supply dispatching model object function, the object function related to particularly as follows:

3.1) object function of subsystem hydropenia index:

$f\left(x\right)=SI={\mathrm{\ω}}_{agr}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{agr,j}-W_{agr,j}}{D_{agr,j}}\right)}^2+{\mathrm{\ω}}_{ind}\frac{100}{N}{\Σ}_{j=1}^N{\left(\frac{D_{ind,j}-W_{ind,j}}{D_{ind,j}}\right)}^2---\left(2\right)$

Wherein, f (x) is the object function of subsystem hydropenia index, and SI is hydropenia index, and N is length series scheduling total year number, ω_agrFor the weight shared by agricultural water target, ω_indFor the weight shared by water for industrial use target, D_agr,j、D_ind,jIt is respectively subsystem agricultural water target, the gross water requirement of water for industrial use target, W in j_agr,j、W_ind,jFor subsystem in j to agriculture, industrial total supply；

3.2) system always abandons the object function of the water yield:

$f^{\′}\left(x\right)=\frac{1}{N}{\mathrm{\Σ}}_{j=1}^N{\mathrm{SU}}_j+\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\α}}_i{\mathrm{Ra}}_i\·PS\_A+\underset{i=1}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\β}}_i\·PS\_B---\left(3\right)$

Wherein, f'(x) it is the system object function of always abandoning the water yield, SU_jFor abandoning the water yield (comprise reservoir abandon water and interval finally enter magnanimity) in j, M is the total number of water user, Ra_iFor the fraction of water user i, α_i、β_iIt is respectively penalty coefficient, the α when water user i is unsatisfactory for fraction requirement_iTake 1, otherwise take 0, β during generation super collapse dept when water user i supplies water_iTaking 1, otherwise take 0, PS_A, PS_B are respectively sufficiently large punishment amount；

4th step, uses multi-objective Evolutionary Algorithm, optimizes the object function of joint water supply dispatching model, obtains relatively optimum combined dispatching rule.

A kind of complicated multi-reservoir the most according to claim 1 supplies water method for allocating tasks jointly, it is characterised in that the hydropenia index that combined dispatching rule is subsystems relatively optimum in the 4th described step is minimum, system always to abandon the water yield minimum.