A Medium- and Long-Term Multi-Objective Stochastic Scheduling Optimization Method Considering Complementary Water, Wind and Photovoltaic

Abstract

The present invention provides a medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind, and light, including: applying a Markov chain to perform time correlation analysis on the time series of historical values of runoff inflow, wind power output, and photovoltaic power output resources to generate a set of relevant scenarios; constructing a medium- and long-term scheduling multi-objective optimization model for reservoir groups with the objectives of maximizing the expected power generation within the scheduling period and maximizing the expected minimum output of the multi-energy complementary system; by solving the medium- and long-term scheduling multi-objective optimization model for reservoir groups, obtaining the hourly outflow of hydropower stations in the scheduling period under the comprehensive scenario, and realizing the coordinated scheduling of the complementarity of water, wind, and light. The present invention comprehensively considers the time correlation among water, wind, and light, generates a set of scenarios to describe the characteristics of runoff and wind and light power output, and constructs a multi-objective optimization scheduling and optimization algorithm to provide a feasible strategy for solving the medium- and long-term multi-objective optimization scheduling of the complementarity of water, wind, and light, and provide theoretical and methodological support for the decision-making of the operation scheduling of hydropower stations.

Description

Technical Field

[0001] The present invention belongs to the technical fields of power systems and renewable energy, and particularly relates to a medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light. Background Art

[0002] At present, there are few studies on medium- and long-term multi-objective stochastic scheduling methods considering the multi-energy complementarity of water, wind and light. Common multi-energy complementary systems include water-light complementary systems, water-wind complementary systems, and water-wind-light complementary systems; for the operation and scheduling level of energy complementary systems, research is mainly carried out on the short term and medium- and long term, such as multi-objective short-term complementary operation scheduling considering water-light or wind-light or water-wind, and single-objective or multi-objective medium- and long-term complementary operation scheduling with partial energy complementary characteristics. The robustness problem of multi-objective optimization methods in different scheduling scenarios during the medium- and long-term multi-energy complementary scheduling process has not been well solved.

[0003] Existing scheduling methods, including: short-term complementary operation scheduling considering partial energy complementarity (A), single-objective medium- and long-term complementary operation scheduling considering partial energy complementarity (B), and multi-objective medium- and long-term complementary operation scheduling considering partial energy complementarity (C), cannot solve how to consider the intraday operation risk and peak shaving ability at the medium- and long-term scale, cannot solve the huge computational burden brought by the nesting of medium- and long-term and short-term models, and the robustness problem of multi-objective optimization methods in different scheduling scenarios. The specific analysis is as follows:

[0004] Technology A mainly includes short-term output prediction of energy such as wind and light, characterization of output uncertainty, construction of a short-term complementary scheduling optimization model, and design of an efficient solution algorithm. Among them, the prediction models include linear / nonlinear regression, support vector machine, frequency domain decomposition, deep learning, and combinations of various models. However, at present, the accuracy of wind and light energy prediction still cannot meet the requirements of practical applications, that is, it cannot be applied to the medium- and long-term multi-objective stochastic scheduling method considering the complementarity of water, wind and light.

[0005] Technology B establishes a medium- and long-term energy complementary model by selecting one of maximizing energy utilization efficiency, minimizing energy cost, maximizing renewable energy penetration rate, or minimizing carbon emissions as its only goal, and obtains the final scheduling plan through medium- and long-term scheduling planning using relevant optimization methods. However, the scheduling plan obtained by traditional single-objective optimization methods usually cannot guarantee relatively better performance in other goals, especially when there is no obvious conversion relationship between different goals or there is a competitive relationship between goals, that is, it cannot be applied to the medium- and long-term multi-objective stochastic scheduling method considering the complementarity of water, wind and light.

[0006] The C technology establishes a multi-objective model for the medium- and long-term scheduling problem of multi-energy complementarity. Compared with short-term scheduling, the medium- and long-term scheduling problem involves a longer time span and often needs to consider seasonal changes, annual planning, and fluctuations in energy demand during different time periods. The characteristics of hydropower, wind power, and photovoltaic power generation make their outputs have significant uncertainties. Therefore, in the medium- and long-term scheduling problem of hydropower, wind power, and photovoltaic power, multi-objective optimization and uncertainty analysis need to be closely combined, and stochastic optimization methods are used to optimize the scheduling plan. However, for the uncertainty analysis part, only the correlations between some energies are considered, such as between water and light, wind and light, water and wind, etc., and the analysis level is relatively weak, and it cannot be applied to the medium- and long-term multi-objective stochastic scheduling method considering the complementarity of hydropower, wind power, and photovoltaic power. Summary of the Invention

[0007] In view of the defects existing in the prior art, the present invention provides a medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of hydropower, wind power, and photovoltaic power, so as to overcome the defects of the medium- and long-term multi-objective stochastic scheduling method related to the complementarity of hydropower, wind power, and photovoltaic power in the background technology and solve the multi-objective stochastic scheduling problem of the complementarity of hydropower, wind power, and photovoltaic power resources.

[0008] The technical solution adopted by the present invention is as follows:

[0009] The present invention provides a medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of hydropower, wind power, and photovoltaic power, including the following steps:

[0010] Step S1, determine the resource variables E1, E2, and E3 as the runoff inflow, wind power output, and photovoltaic power output respectively; for each resource variable E b , where b = 1, 2, 3, the time series of historical values of the resource variables in the target area in historical years are obtained. Therefore, the time series of historical runoff values, the time series of historical wind power output values, and the time series of historical photovoltaic power output values are obtained respectively;

[0011] Step S2, for each resource variable E b , determine the corresponding state space; then, apply the Markov chain to perform time correlation analysis on the time series of historical values of the resource variable E b , and obtain the cumulative probability state transition matrix Q b of the resource variable E b ;

[0012] Determine that the research time scale is the length Δt of the time period t; set the number B of simulation scenarios to be initially generated and the initial value of the resource variable E b ; based on the cumulative probability state transition matrix Q b of the resource variable E b , using a random number method, generate the time series of simulated values of the resource variable E b for each time period t within the year, forming a simulation scenario; execute B times in a loop to obtain the resource variable Eb Scenario B simulation to form the resource variable E b Initial simulation scenario set;

[0013] Step S3, set the target number J of simulation scenarios; adopt the simulation scenario reduction algorithm to reduce the B simulation scenarios in the initial simulation scenario set of each resource variable E b to obtain a reduced simulation scenario set with J simulation scenarios;

[0014] Therefore, for the runoff, wind power output, and photovoltaic power output resource variables, simulation scenario sets with J simulation scenarios are obtained respectively, and each simulation scenario in the simulation scenario set of each resource variable is randomly arranged, thereby obtaining the final runoff simulation scenario set SCE1 = {S 11 , S 12 ,..., S 1J}, wind power output simulation scenario set SCE2 = {S 21 , S 22 ,..., S 2J}, and photovoltaic power output simulation scenario set SCE3 = {S 31 , S 32 ,..., S 3J};

[0015] Combine the runoff simulation scenario set SCE1, wind power output simulation scenario set SCE2, and photovoltaic power output simulation scenario set SCE3 to form an integrated scenario set SC = {S1, S2,..., S J}; where, for the integrated scenario S j in the integrated scenario set SC, j = 1, 2,..., J, there is the following relationship: S j = {S 1j , S 2j , S 3j}, and at the same time, the probability p j of the integrated scenario S j = (p 1j + p 2j + p 3j ) / 3; p 1j , p 2j , p 3j are the probabilities of the runoff simulation scenario S 1j , wind power output simulation scenario S 2j , and photovoltaic power output simulation scenario S 3j respectively;

[0016] Step S4, construct a multi-objective optimization model for medium- and long-term reservoir operation scheduling with the goal of maximizing the expected power generation within the scheduling period and maximizing the expected minimum output of the multi-energy complementary system;

[0017] Step S5. For each comprehensive scenario S j , using the simulation value time series of the runoff simulation scenario S 1j , the wind power output simulation scenario S 2j and the photovoltaic power output simulation scenario S 3j as input values, by solving the medium and long-term scheduling multi-objective optimization model of the reservoir group, the discharge flow QT j of hydropower station n at each time period t in the scheduling period under the comprehensive scenario S n,j,t is obtained to achieve the coordinated scheduling of water, wind and light.

[0018] Preferably, step S2 is specifically as follows:

[0019] Step S2.1. For each resource variable E b , determine the corresponding state space; there are β states in the state space, which are state 1, state 2,..., state β;

[0020] Step S2.2. Apply the Markov chain to perform time correlation analysis on the historical value time series of the resource variable E b to obtain the state transition probability matrix G b of the resource variable E b :

[0021]

[0022] Where: g u',v' represents the state transition probability of the resource variable E b from state u' to state v'; u' = 1, 2,..., β; v' = 1, 2,..., β;

[0023] Step S2.3. Based on the state transition probability matrix G b of the resource variable E b , obtain the cumulative probability state transition matrix Q b of the resource variable E b :

[0024]

[0025] Where: q u',v' represents the cumulative state transition probability of the resource variable E b from state u' to state v'; q u',v' = g u',1 + g u',2 +... + g u',v' ;

[0026] Step S2.4. Generate a simulation scenario of the resource variable E b :

[0027] Step S2.4.1, set the initial value a0 of the resource variable E, and obtain the initial state β0 according to the initial value a0; where β0 = 1, 2,..., β; b The initial value a0, according to the initial value a0, obtains the initial state β0; where, β0 = 1, 2,..., β;

[0028] Step S2.4.2, let the time period t = 1;

[0029] Step S2.4.3, generate random numbers e1, e1 ∈ [0, 1]; in the cumulative probability state transition matrix Q b locate the row vector corresponding to the state β t-1 i.e.: the row vector of the β t-1 th row:

[0030] Step S2.4.4, traverse the elements in the row vector in the forward direction. When encountering an element greater than or equal to the random number e1 , v' = 1, 2,..., β, the state v' after state transition corresponding to this element is the state at time period t, and is re - expressed as state β t ; the numerical range of the resource variable E t corresponding to state β b is: [a l , a h ;

[0031] Step S2.4.5, generate random numbers e2, e2 ∈ [0, 1]; through the formula a t = a t-1 + e2×(a h - a l ), obtain the simulated value a b of the resource variable E t at time period t; where, a t-1 is the simulated value of the resource variable E b at time period t - 1;

[0032] Step S2.4.6, determine whether the time period t reaches the preset total number of scheduling time periods T. If it reaches, execute Step S2.4.7; if not, let t = t + 1, and return to Step S2.4.3;

[0033] Step S2.4.7, output the simulated values of the resource variable E b at time periods 1, 2,..., T, forming a simulated value time series a1, a2,..., a T , forming a simulation scenario;

[0034] Step S2.4.8: Determine whether the B types of simulation scenarios have been obtained. If not, return to Step S2.4.2. By executing Steps S2.4.2 to S2.4.7, form another simulation scenario; continuously loop in this way until the B types of simulation scenarios are obtained, and form the resource variable E b of the initial simulation scenario set.

[0035] Preferably, in Step S3, use the simulation scenario reduction algorithm to reduce the B simulation scenarios in the initial simulation scenario set of each resource variable E b to obtain a reduced simulation scenario set with J simulation scenarios, specifically as follows:

[0036] Step S3.1: The initial simulation scenario set of the resource variable E b includes B types of simulation scenarios, expressed as: Simulation scenario set RS = {ξ1, ξ2,..., ξ B}; ξ1, ξ2,..., ξ B respectively represent B types of simulation scenarios;

[0037] Step S3.2: Set the initial probability p k of each simulation scenario ξ k to be equal, all 1 / B; where x = 1, 2,..., B; establish a set DS to store the scenarios to be eliminated;

[0038] Step S3.3: Set the set DS to be empty;

[0039] Step S3.4: Traverse each simulation scenario ξ k in the simulation scenario set RS;

[0040] For each traversed simulation scenario ξ k , determine the simulation scenario ξ k nearest to the simulation scenario ξ k" , that is, satisfy the relationship:

[0041] DT k (k") = minDT k,s , k ∈ RS, s ∈ RS, k ≠ s

[0042] where: DT k,s represents the distance between the simulation scenario ξ k and the simulation scenario ξ s ; DT k (k") represents the distance between the simulation scenario ξ k and the nearest simulation scenario ξ k" ;

[0043] Use the formula PD k = p k*DT k (k"), k ∈ RS, calculate the simulated scenario ξ k The corresponding minimum distance probability PD k ;

[0044] Step S3.5, compare the minimum distance probabilities PD k of each simulated scenario ξ k in the simulated scenario set RS, and use the formula PD d = minPD k , k ∈ S, to obtain the minimum distance probability PD k in the simulated scenario set RS, denoted as PD d ; This PD d corresponds to the simulated scenario ξ d , and the simulated scenario closest to the simulated scenario ξ d is ξ d* ;

[0045] Step S3.6, let the simulated scenario set RS = RS - {ξ d}, DS = DS + {ξ d}, and, let p d* = p d* + p d ; Determine whether the number of simulated scenarios in the simulated scenario set RS has been reduced to J. If so, execute Step S3.7; if not, return to Step S3.4;

[0046] Step S3.7, obtain the reduced simulated scenario set with J simulated scenarios, and each simulated scenario in this simulated scenario set has a probability.

[0047] Preferably, in Step S4, in the medium - and long - term scheduling multi - objective optimization model of the reservoir group, the outflow discharge QT j of hydropower station n at time t under the comprehensive scenario S n,j,t is used as the decision variable of the model; The objective function includes objective function F1 and objective function F2;

[0048] Objective function F1: Maximize the expected power generation of the multi - energy complementary system, and the expression is:

[0049]

[0050] Where: P n,j,t represents the output of hydropower station n at time t under the comprehensive scenario S j ; t = 1, 2,..., T; T is the total number of medium - and long - term scheduling time periods; n = 1, 2,..., N; N represents the number of hydropower stations; j = 1, 2,..., J; J represents the comprehensive scenario set SC = {S1, S2,..., S J; number of comprehensive scenarios; p j denotes the comprehensive scenario S j probability; ΔT represents the length of the medium- and long-term scheduling period;

[0051] denotes the output of the wind turbine at time period t under the comprehensive scenario S j ; denotes the output of the photovoltaic unit at time period t under the comprehensive scenario S j ;

[0052] Objective function F2: maximize the minimum output expectation of the multi-energy complementary system, and the expression is:

[0053]

[0054] Thus, the objective function F1 and the objective function F2 are established.

[0055] Preferably, the medium- and long-term scheduling multi-objective optimization model of the reservoir group includes the following constraint conditions:

[0056] (1) Water balance equation:

[0057]

[0058] Where: V n,h,t+1 , V n,j,t are the reservoir capacities of hydropower station n at time periods t + 1 and t respectively; I n,j,t , QT n,j,t are the runoff inflow and outflows of hydropower station n at time period t under the comprehensive scenario S j respectively; Ω(n) is the set of upstream hydropower stations of hydropower station n; QT m,j,t is the runoff inflow of the upstream hydropower station m of hydropower station n at time period t under the comprehensive scenario S j ;

[0059] (2) Outflow equation:

[0060]

[0061] Where: QP n,j,t , S n,j,t are the power generation flow and the waste water flow of hydropower station n at time period t under the comprehensive scenario S j respectively;

[0062] (3) Outflow limit:

[0063]

[0064] Where: QT n,t , are the lower and upper bounds of the total outflow of hydropower station n in period t, respectively;

[0065] (4) Generation flow limit:

[0066]

[0067] Where: QP n,t and are the lower and upper bounds of the generation flow of hydropower station n in period t, respectively;

[0068] (5) Output limit:

[0069]

[0070] Where: P n,t and are the lower and upper bounds of the output of hydropower station n in period t, respectively;

[0071] (6) Reservoir capacity limit:

[0072]

[0073] Where: V n,t and are the lower and upper bounds of the reservoir capacity of hydropower station n in period t, respectively;

[0074] (7) Initial and final water level constraints at the beginning of the dispatching period:

[0075]

[0076] Where: Z n,Ini and Z n,End are the initial water level and the final water level of hydropower station n, respectively; Z n,j,1 and Z n,j,T are the water levels of hydropower station n in period 1 and period T, respectively;

[0077] (8) Water level - reservoir capacity constraint:

[0078]

[0079] Where: is the non - linear function between the reservoir water level and the water storage volume; Z n,j,t is the water level of hydropower station n in period t under the comprehensive scenario S j ;

[0080] (9) Tail water level - discharge constraint:

[0081]

[0082] Where: is the non - linear function between the tail water level and the outflow of hydropower station n; The tail water level of hydropower station n at time period t under the comprehensive scenario S j ;

[0083] (10) Hydropower output generation function:

[0084]

[0085] Where: H n,j,t is the generating head of hydropower station n at time period t under the comprehensive scenario S j ; K n is the output coefficient of hydropower station n; Z n,j,t+1 is the water level of hydropower station n at time period t + 1 under the comprehensive scenario S j ;

[0086] (11) Transmission capacity limit constraint:

[0087]

[0088] Where: W n represents the set of wind power stations associated with hydropower station n; P wd represents the output of wind power station wd associated with hydropower station n; B n represents the set of photovoltaic power stations associated with hydropower station n; P sol represents the output of photovoltaic power station sol associated with hydropower station n; P n represents the total output of hydropower station n; is the maximum value of the integrated water, wind and light transmission capacity of hydropower station n.

[0089] Preferably, step S5 is specifically:

[0090] Step S5.1, generate individual coding X:

[0091] X = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}; where, QT n,j,t means: the runoff inflow of hydropower station n at time period t under the comprehensive scenario S j , which is a decision variable; according to the individual coding X, generate the corresponding number of individuals with initial solutions according to the set population size, so as to form an initial population;

[0092] Step S5.2, determine the population size NI and the ratio γ; where, the ratio γ is the number of individuals in the dominant population / the number of individuals in the basic population;

[0093] Step S5.3: Determine the dominant population and the basic population using the inter - individual domination algorithm;

[0094] Step S5.4: Adopt the multi - recombination operator fusion strategy to perform crossover operations on the dominant population and the basic population to generate a new population; then return to Step S5.3, and re - use the inter - individual domination algorithm to divide the new population into a dominant population and a basic population. Repeat this process continuously until the set number of iterations is reached, and then execute Step S5.5;

[0095] Step S5.5: For each individual in the new population, use the heuristic constraint handling method to perform heuristic constraint handling on the solution value of each individual, so as to obtain each individual with an updated solution value;

[0096] Step S5.6: Based on the updated solution values of each individual obtained in Step S5.5, calculate the objective function values of the objective function F1 and the objective function F2 corresponding to each individual;

[0097] Step S5.7: According to the objective function values corresponding to each individual obtained, use the restart mechanism to determine whether the iteration termination condition is met. If it is met, execute Step S5.8; if not, determine whether the restart mechanism is needed. If it is needed, start the restart mechanism, update the basic population, and return to execute Step S5.4; if not, return to execute Step S5.3;

[0098] Step S5.8: Output the solution value of each individual in the dominant population. The solution value of each individual is a scheduling scheme, and thus a multi - objective optimal scheduling scheme set is obtained, thereby completing the multi - objective optimal scheduling for the medium - and long - term of the hydropower station group.

[0099] Preferably, Step S5.3 is specifically as follows:

[0100] Step S5.3.1: Set the value of the parameter ∈;

[0101] Step S5.3.2: With the function value of the objective function F1 as the abscissa, the function value of the objective function F2 as the ordinate, and the parameter ∈ as the grid side length, establish a grid - based objective function space;

[0102] Step S5.3.3: For each current individual X c , c = 1, 2, …, NI, determine the grid in which it is located in the objective function space according to its function value of the objective function F1 and the function value of the objective function F2;

[0103] Step S5.3.4: Traverse each individual and calculate the domination number NC of each individual X c . The calculation method is as follows:

[0104] For the individual X c, its initial domination number NC is 0; traverse the other NI - 1 individuals in sequence. For the currently traversed individual X d , d = 1, 2, …, NI, and d ≠ c. If individual X d and individual X c have function values in the same grid in the objective function space, then use Algorithm 1 to determine whether individual X c dominates individual X d . If it dominates, the domination number NC is incremented by 1; if individual X d and individual X c have function values in different grids in the objective function space, then use Algorithm 2 to determine whether individual X c dominates individual X d . If it dominates, the domination number NC is incremented by 1; thus, traverse the other NI - 1 individuals to obtain the value of the domination number NC, which is the domination number NC of individual X c , representing the number of individuals dominated by individual X c ;

[0105] Where: Algorithm 1 and Algorithm 2 are as follows:

[0106] Individual X c = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}}, the function value of the objective function F0 of individual X c is F 1c ; the function value of the objective function F2 of individual X c is F 2c , and the objective function value vector of individual X c is represented as: F c = {F 1c , F 2c};

[0107] The function value of the objective function F1 of individual X d is F 1d ; the function value of the objective function F2 of individual X d is F 2d , and the objective function value vector of individual X d is represented as: F d = {F 1d , F 2d};

[0108] Algorithm 1 is: When conditions 1 and 2 are satisfied, individual X c dominates individual X d ;

[0109] Condition 1: F 1c ≥ F 1d + ∈, and F 2c ≥ F 2d + ∈;

[0110] Condition 2: There exists F xc , x = 1, 2, satisfying the relationship: F xc > F 1d + ∈, and, F xc > F 2d + ∈;

[0111] Algorithm 2 is: When and only when one of Condition 3 and Condition 4 holds, individual X c dominates individual X d :

[0112] Condition 3:

[0113] Among them: represents and satisfies the relationship of Condition 1 and Condition 2, represents rounding down;

[0114] Condition 4:

[0115] Among them: |||| is the modulo operation; & is the logical AND;

[0116] Step S5.3.5, according to the population size NI and the ratio γ, obtain the number of individuals in the dominant population archive size and the number of individuals in the basic population population size;

[0117] Sort each individual in descending order of the domination number. If the domination numbers of two individuals are the same, sort them arbitrarily; among the NI individuals after sorting, select the top archive size number of individuals to form the dominant population, and the other individuals form the basic population.

[0118] Preferably, Step S5.4 is specifically:

[0119] Step S5.4.1, set up a crossover operator pool; there are N_CO crossover operators in the crossover operator pool;

[0120] Step S5.4.2, initially, set each crossover operator CO to have the same selection probability PC;

[0121] Step S5.4.3, select a crossover operator for crossover operation and update the selection probability PC of the crossover operator:

[0122] ① Based on the selection probability PC of each current crossover operator, select crossover operator A from the crossover operator pool; ② According to the number of parents N required by crossover operator A A , randomly select a parent individual X from the dominant population A1 , and select N - 1 parent individuals from the basic population through tournament selection, denoted as: parent individuals X A , B = 2, 3, …, N AB ; A ;

[0123] ③ Use crossover operator A to perform crossover operation on parent individual X A1 and parent individual X AB to generate N A offspring individuals, and delete the N A original parent individuals from the population;

[0124] ④ According to the function values of the N A offspring individuals, determine the grid they are in the objective function space, add the offspring individuals within the range of the dominant population to the dominant population, and add other offspring individuals to the basic population; where: the number of offspring individuals added to the dominant population generated by crossover operator A this time is SN_A;

[0125] ⑤ Use the following formula to update the selection probability PC of crossover operator A A :

[0126]

[0127] where: SN_V represents the total number of offspring individuals added to the dominant population generated by crossover operator V from the first loop to the current loop, V = 1, 2, …, N_CO; is a constant,

[0128] ⑥ At the end of this loop, return to step ① and continue the next loop until the set number of loops is reached.

[0129] Preferably, step S5.5 is specifically:

[0130] For each individual X = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}, perform the adjustment of the power generation flow dynamic balance constraint in step S5.4.1 and the adjustment of the reservoir storage capacity constraint in step S5.4.2:

[0131] Step S5.4.1, Adjustment of dynamic balance constraint of discharge flow:

[0132] ① According to the water balance equation and the discharge flow equation, obtain the water volume difference ΔQT between the water consumption and the constrained water volume of reservoir n during the entire scheduling period under the comprehensive scenario S j ; n,j ;

[0133] ② Use the following formula to adjust the water volume difference ΔQT n,j to the discharge flow QT of each time period t of reservoir n under the comprehensive scenario S j to obtain the corrected discharge flow n,j,t ;

[0134]

[0135] ③ Use the following formula to further correct the corrected discharge flow by applying the discharge flow limit constraint condition to obtain the corrected discharge flow ;

[0136]

[0137] ④ Thus, obtain the discharge flow of each time period t of reservoir n under the corrected comprehensive scenario S j ; judge whether the water balance equation is satisfied during the entire scheduling period for reservoir n under the comprehensive scenario S j ; if it is satisfied, complete the adjustment of the dynamic balance constraint of the discharge flow; otherwise, return to step ① and perform corrections in a loop;

[0138] Step S5.4.2, Adjustment of reservoir storage capacity constraint:

[0139] ① Calculate the reservoir storage capacity V of hydropower station n at time period t under the comprehensive scenario S according to the discharge flow QT of each time period t of reservoir n under the current individual X j ; n,j,t ; j ; n,j,t ;

[0140] ② Judge whether the reservoir storage capacity V of hydropower station n at time period t under the comprehensive scenario S j satisfies the reservoir storage capacity limit constraint condition; if it does not satisfy, when the reservoir storage capacity V n,j,t exceeds the upper bound of the reservoir storage capacity of hydropower station n at time period t n,j,t then execute ③; when the reservoir storage capacity V is lower than the lower bound of the reservoir storage capacity of hydropower station n at time period t n,j,t then execute ④; V n,t ;

[0141] ③ The excess amount Average it to each time period t during the scheduling period, that is: obtain the average of the excess

[0142] Then, according to the average of the excess Average it to each time period t during the scheduling period, and complete the correction of the reservoir capacity V of hydropower station n at each time period t under the comprehensive scenario S j at each time period t n,j,t ;

[0143] If, after completing the correction of the reservoir capacity V of reservoir n at all time periods t under the comprehensive scenario S j at each time period t n,j,t , there are still time periods t whose reservoir capacity V exceeds the upper bound of the reservoir capacity n,j,t , continue to average the excess to the out - flow of adjacent reservoirs;

[0144] ④ Average the shortage amount V n,t -V n,j,t to each time period t during the scheduling period to obtain the average shortage amount ΔV n,j,t =( V n,t -V n,j,t ) / T; Then, reduce the reservoir capacity V n,j,t of other time periods t by ΔV n,j,t , and supplement it to the reservoir capacity V n,j,t of the current time period t; complete the correction of the reservoir capacity V of hydropower station n at time period t under the comprehensive scenario S j at each time period t n,j,t ;

[0145] If, after completing the correction of the reservoir capacity V of reservoir n at all time periods t under the comprehensive scenario S j at each time period t n,j,t , there are still time periods t whose reservoir capacity V V n,t is less than the lower bound of the reservoir capacity n,j,t , then supplement the out - flow of adjacent reservoirs to the reservoir capacity V V n,t of the time period t whose reservoir capacity is less than the lower bound of the reservoir capacity n,j,t .

[0146] Preferably, in step S5.7, determine whether to perform a restart mechanism. If so, start the restart mechanism and update the basic population, specifically:

[0147] When the algorithm search stagnates, start the restart mechanism: according to the current ratio γ and the current number of individuals in the dominant population archive size, use the formula archive size / γ to obtain the number of individuals in the basic population;

[0148] Empty the current basic population, and fill the corresponding number of individuals with the highest dominance number NC in the current dominant population into the basic population. The insufficient part is obtained by mutating the individuals in the dominant population, so as to obtain the updated basic population.

[0149] A medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light provided by the present invention has the following advantages:

[0150] The present invention combines the correlation analysis of water, wind and light with multi-objective stochastic scheduling optimization, constructs a set of decision-making processes, and this set of decisions is based on the obtained runoff and wind and light output scenario sets, and efficiently solves the multi-objective optimal scheduling of medium- and long-term water, wind and light. Compared with the prior art, the present invention comprehensively considers the time correlation of water, wind and light, generates a scenario set on this basis to describe the characteristics of runoff and wind and light output, and constructs a multi-objective optimal scheduling and optimization algorithm to provide a feasible strategy for solving the medium- and long-term multi-objective optimal scheduling of water, wind and light complementarity, and provides theoretical and method support for the decision-making of hydropower operation scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0151] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0152] Figure 1 It is a flowchart of a medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0153] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be described in detail below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope protected by the present invention.

[0154] Driven by climate change and the "dual carbon" goal, renewable energy represented by wind and photovoltaic power has developed rapidly in recent years. Making full use of the flexibility of hydropower and the natural complementarity between different resources, complementary power generation is an effective way to promote the consumption of new energy. The present invention combines the correlation analysis of water, wind and light with multi-objective stochastic scheduling optimization to construct a decision-making process. Based on the obtained runoff and wind-light output scenario sets, this set of decisions efficiently solves the multi-objective optimal scheduling of medium- and long-term water, wind and light. Compared with the prior art, the present invention comprehensively considers the time correlation among water, wind and light, and generates a scenario set on this basis to characterize the runoff and wind-light output characteristics, and constructs a multi-objective optimal scheduling and optimization algorithm to provide a feasible strategy for solving the medium- and long-term multi-objective optimal scheduling of water, wind and light complementarity, and provides theoretical and method support for the decision-making of hydropower operation scheduling.

[0155] The present invention designs a model considering the time correlation of water, wind and light complementarity and the Borg algorithm for solving multi-objective scheduling for the medium- and long-term multi-objective scheduling considering water, wind and light complementarity; the Markov chain mainly describes the situation of the values in the time series of water, wind and light changing between different states over time. Based on this, a relevant scenario set is generated; the Borg algorithm is an efficient multi-objective optimization framework, which combines the domination rule, the random restart mechanism and the adaptive multi-operator recombination strategy. Based on the initialized population, the domination rule is used to obtain the dominant population and the basic population, and then the multi-operator fusion strategy is adopted to perform a heuristic adjustment constraint processing method. According to the calculated objective function value, judge the stagnation situation in the process and trigger a random restart; in the iterative process, different optimization problems dynamically adjust the operator, so as to improve the solution efficiency and the quality of the solution.

[0156] Refer to Figure 1 , the present invention provides a medium- and long-term multi-objective stochastic scheduling optimization method considering water, wind and light complementarity, including the following steps:

[0157] Step S1, determine that the resource variables E1, E2, and E3 are the runoff inflow, wind power output, and photovoltaic power output respectively; for each resource variable E b , b = 1, 2, 3, the historical value time series of the resource variable in the target area in the historical year is obtained. Therefore, the runoff historical value time series, the wind power output historical value time series, and the photovoltaic power output historical value time series are obtained respectively;

[0158] Step S2, for each resource variable E b , determine the corresponding state space; then, apply the Markov chain to perform time correlation analysis on the historical value time series of the resource variable E b to obtain the cumulative probability state transition matrix Q b of the resource variable E b ;

[0159] Determine the length Δt of the research time scale as the time period t; set the number B of simulation scenarios to be initially generated and the initial value of the resource variable E b ; Based on this resource variable E b cumulative probability state transition matrix Q b , using a random number method, generate the time series of simulated values of this resource variable E b for each time period t within a year, forming a simulation scenario; execute the loop B times to obtain b B simulation scenarios of this resource variable E b , forming the initial simulation scenario set of this resource variable E

[0160] Step S2 is specifically as follows:

[0161] Step S2.1, for each resource variable E b , determine the corresponding state space; there are β states in the state space, namely state 1, state 2,..., state β;

[0162] Step S2.2, apply the Markov chain to perform time correlation analysis on the historical value time series of this resource variable E b to obtain the state transition probability matrix G b of this resource variable E b :

[0163]

[0164] where: g u',v' represents the state transition probability between states of the resource variable E b from state u' to state v'; u' = 1, 2,..., β; v' = 1, 2,..., β; g u',v' is expressed as follows:

[0165]

[0166] where: α t+1 and α t represent the states of the resource variable E b at time periods t and t + 1, f u',v' is the frequency of the resource variable E b transferring from state u' to state v', u', v' ∈ s, g(·|·) is the conditional transition probability; in the present invention, the time period t can be different time scales such as month, week, day, etc.

[0167] Step S2.3, based on the state transition probability matrix G b of the resource variable E b , obtain the cumulative probability state transition matrix Q b of the resource variable E b :

[0168]

[0169] where: q u',v' represents the resource variable E b the cumulative probability of state transition from state u' to state v'; q u',v' = g u',1 + g u',2 +... + g u',v' ;

[0170] Step S2.4, generate a simulation scenario of the resource variable E b :

[0171] Step S2.4.1, set the initial value a0 of the resource variable E b , and obtain the initial state β0 according to the initial value a0; where, β0 = 1, 2,..., β;

[0172] Step S2.4.2, let the time period t = 1;

[0173] Step S2.4.3, generate random numbers e1, e1 ∈ [0, 1]; in the cumulative probability state transition matrix Q b , locate the row vector corresponding to the state β t-1 , that is: the row vector of the β t-1 th row:

[0174] Step S2.4.4, traverse the elements in the row vector in the forward direction. When traversing to an element greater than or equal to the random number e1 , v' = 1, 2,..., β, and the state v' after state transition corresponding to this element is the state of the time period t, and is re - represented as state β t ; the numerical range of the resource variable E t corresponding to state β b is: [a l , a h ;

[0175] Step S2.4.5, generate random numbers e2, e2 ∈ [0, 1]; through the formula a t = a t-1 + e2 × (a h - a l ), obtain the simulated value a b of the resource variable E t in the time period t; where, a t-1 is the simulated value of the resource variable E b in the time period t - 1;

[0176] Step S2.4.6: Determine whether the time period t has reached the preset total number of scheduling time periods T. If it has reached, execute Step S2.4.7; if not, set t = t + 1 and return to Step S2.4.3.

[0177] Step S2.4.7: Output the resource variable E obtained at this time. b The simulation values in time periods 1, 2,..., T form a simulation value time series a1, a2,..., a T , forming a simulation scenario.

[0178] Step S2.4.8: Determine whether B simulation scenarios have been obtained currently. If not, return to Step S2.4.2, and form another simulation scenario by executing Step S2.4.2 to Step S2.4.7; continuously loop like this until B simulation scenarios are obtained to form the initial simulation scenario set of this resource variable E. b of the initial simulation scenario set.

[0179] Step S3: Set the target number of simulation scenarios J; adopt a simulation scenario reduction algorithm to reduce the B simulation scenarios in the initial simulation scenario set of each resource variable E b to obtain a reduced simulation scenario set with J simulation scenarios.

[0180] The specific simulation scenario reduction algorithm is as follows:

[0181] Step S3.1: The initial simulation scenario set of the resource variable E b includes B simulation scenarios, denoted as: simulation scenario set RS = {ξ1, ξ2,..., ξ B}; ξ1, ξ2,..., ξ B respectively represent B simulation scenarios.

[0182] Step S3.2: Set the initial probability p k of each simulation scenario ξ k to be equal, all 1 / B; where x = 1, 2,..., B; establish a set DS to store the scenarios to be eliminated.

[0183] Step S3.3: Set the set DS to be empty.

[0184] Step S3.4: Traverse each simulation scenario ξ in the simulation scenario set RS k ;

[0185] For each traversed simulation scenario ξ k , determine the simulation scenario ξ k nearest to the simulation scenario ξ k" , that is, satisfying the relational expression:

[0186] DT k DT(k") = minDT k,s , where k ∈ RS, s ∈ RS, k ≠ s

[0187] Where: DT k,s represents the distance between the simulation scenario ξ k and the simulation scenario ξ s ; DT k DT(k") represents the distance between the simulation scenario ξ k and the nearest simulation scenario ξ k" ;

[0188] Using the formula PD k = p k * DT k (k"), k ∈ RS, calculate the minimum distance probability PD k corresponding to the simulation scenario ξ k ;

[0189] Step S3.5, compare the minimum distance probabilities PD k of each simulation scenario ξ k in the simulation scenario set RS, using the formula PD d = minPD k , k ∈ S, to obtain the minimum value of the minimum distance probability PD k in the simulation scenario set RS, denoted as PD d ; This PD d corresponds to the simulation scenario ξ d , and the simulation scenario closest to the simulation scenario ξ d is

[0190] Step S3.6, let the simulation scenario set RS = RS - {ξ d}, DS = DS + {ξ d}, and, let Judge whether the number of simulation scenarios in the simulation scenario set RS has been reduced to J. If so, execute Step S3.7; if not, return to Step S3.4;

[0191] Step S3.7, obtain the reduced simulation scenario set with J simulation scenarios, and each simulation scenario in this simulation scenario set has a probability.

[0192] Therefore, for the runoff, wind power output, and photovoltaic power output resource variables, respectively obtain a simulation scenario set with J simulation scenarios, and randomly arrange each simulation scenario in the simulation scenario set of each resource variable. Thus, the final runoff simulation scenario set SCE1 = {S 11 , S 12 ,..., S1J}, the wind power output simulation scenario set SCE2 = {S 21 , S 22 ,..., S 2J} and the photovoltaic power output simulation scenario set SCE3 = {S 31 , S 32 ,..., S 3J};

[0193] Combine the runoff simulation scenario set SCE1, the wind power output simulation scenario set SCE2, and the photovoltaic power output simulation scenario set SCE3 to form a comprehensive scenario set SC = {S1, S2,..., S J}; where, for the comprehensive scenario S j in the comprehensive scenario set SC, j = 1, 2,..., J, the following relationship holds: S j = {S 1j , S 2j , S 3j}, and at the same time, the probability p j of the comprehensive scenario S j = (p 1j + p 2j + p 3j ) / 3; p 1j , p 2j , p 3j are the probabilities of the runoff simulation scenario S 1j , the wind power output simulation scenario S 2j , and the photovoltaic power output simulation scenario S 3j respectively;

[0194] Step S4, construct a multi-objective optimization model for medium and long-term scheduling of a reservoir group with the goal of maximizing the expected power generation within the scheduling period and maximizing the expected minimum output of the multi-energy complementary system;

[0195] Specifically, in the multi-objective optimization model for medium and long-term scheduling of the reservoir group, the outflow discharge QT j of hydropower station n at time t under the comprehensive scenario S n,j,t is used as the decision variable of the model; the objective function includes objective function F1 and objective function F2;

[0196] Objective function F1: Maximize the expected power generation of the multi-energy complementary system, and the expression is:

[0197]

[0198] Where: P n,j,t represents the hydropower station n under the comprehensive scenario S jOutput at time period t; t = 1, 2,..., T; T is the total number of medium- and long-term scheduling time periods; n = 1, 2,..., N; N represents the number of hydropower stations; j = 1, 2,..., J; J represents the number of comprehensive scenarios in the comprehensive scenario set SC = {S1, S2,..., S J}; p j represents the probability of comprehensive scenario S j ; ΔT represents the length of the medium- and long-term scheduling time period;

[0199] represents the output of the wind turbine at time period t under comprehensive scenario S j ; represents the output of the photovoltaic unit at time period t under comprehensive scenario S j ;

[0200] Objective function F2: Maximize the expected minimum output of the multi-energy complementary system. The expression is:

[0201]

[0202] Thus, the objective function F1 and the objective function F2 are established.

[0203] In the medium- and long-term scheduling multi-objective optimization model of the reservoir group, the following constraint conditions are included:

[0204] (1) Water balance equation:

[0205]

[0206] Where: V n,h,t+1 , V n,j,t are the reservoir capacities of hydropower station n at time periods t + 1 and t respectively; I n,j,t , QT n,j,t are the runoff inflow and outflows of hydropower station n at time period t under comprehensive scenario S j respectively; Ω(n) is the set of upstream hydropower stations of hydropower station n; QT m,j,t is the runoff inflow of the upstream hydropower station m of hydropower station n at time period t under comprehensive scenario S j ;

[0207] (2) Outflow equation:

[0208]

[0209] Where: QP n,j,t , S n,j,t are the power generation flow and the waste water flow of hydropower station n at time period t under comprehensive scenario S j respectively;

[0210] (3) Outflow limit:

[0211]

[0212] Wherein: QT n,t 、 are respectively the lower bound and the upper bound of the total out - flow of hydropower station n in time period t;

[0213] (4) Generation flow rate limit:

[0214]

[0215] Wherein: QP n,t 、 are respectively the lower bound and the upper bound of the generation flow rate of hydropower station n in time period t;

[0216] (5) Output limit:

[0217]

[0218] Wherein: P n,t 、 are respectively the lower bound and the upper bound of the output of hydropower station n in time period t;

[0219] (6) Reservoir capacity limit:

[0220]

[0221] Wherein: V n,t 、 are respectively the lower bound and the upper bound of the reservoir capacity of hydropower station n in time period t;

[0222] (7) Initial and final water - level constraints at the beginning of the scheduling:

[0223]

[0224]

[0225] Wherein: Z n,Ini 、Z n,End are respectively the initial water - level and the final water - level of hydropower station n; Z n,j,1 、Z n,j,T are respectively the water - levels of hydropower station n in time period 1 and time period T;

[0226] (8) Water - level - reservoir capacity constraint:

[0227]

[0228] Wherein: is a non - linear function between the reservoir water - level and the water storage volume; Z n,j,t is the water - level of hydropower station n in time period t under the comprehensive scenario S j ;

[0229] (9) Tail water level - discharge constraint:

[0230]

[0231] Where: is a non - linear function between the tail water level and the outflow of hydropower station n; is the tail water level of hydropower station n at time period t under the comprehensive scenario S j ;

[0232] (10) Hydropower generation function:

[0233]

[0234] Where: H n,j,t is the generating head of hydropower station n at time period t under the comprehensive scenario S j ; K n is the output coefficient of hydropower station n; Z n,j,t+1 is the water level of hydropower station n at time period t + 1 under the comprehensive scenario S j ;

[0235] (11) Transmission capacity limit constraint:

[0236]

[0237] Where: W n represents the set of wind power stations associated with hydropower station n; P wd represents the output of wind power station wd associated with hydropower station n; B n represents the set of photovoltaic power stations associated with hydropower station n; P sol represents the output of photovoltaic power station sol associated with hydropower station n; P n represents the total output of hydropower station n; is the maximum value of the integrated water - wind - light transmission capacity of hydropower station n.

[0238] Step S5. For each comprehensive scenario S j , using the simulated value time series of the runoff simulation scenario S 1j , wind power output simulation scenario S 2j and photovoltaic output simulation scenario S 3j as input values, by solving the medium - and long - term scheduling multi - objective optimization model of the reservoir group, the outflow QT j of hydropower station n at each time period t in the scheduling period under the comprehensive scenario S n,j,t is obtained, realizing the coordinated scheduling of water - wind - light complementary.

[0239] Step S5 is specifically as follows:

[0240] Step S5.1, generate individual coding X:

[0241] X = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}; where QT n,j,t means the runoff inflow of hydropower station n at time period t under the comprehensive scenario S j , which is a decision variable; according to the individual coding X, a corresponding number of individuals with initial solutions are generated according to the set population size, thus forming an initial population;

[0242] Step S5.2, determine the population size NI and the ratio γ; where the ratio γ is the number of individuals in the dominant population / the number of individuals in the basic population;

[0243] Step S5.3, use the individual domination algorithm to determine the dominant population and the basic population;

[0244] Step S5.3 is specifically as follows:

[0245] Step S5.3.1, set the value of the parameter ∈;

[0246] Step S5.3.2, with the function value of the objective function F1 as the abscissa, the function value of the objective function F2 as the ordinate, and the parameter ∈ as the grid side length, establish a grid-based objective function space;

[0247] Step S5.3.3, for each current individual X c , c = 1, 2, …, NI, determine the grid it is located in the objective function space according to its function value of the objective function F1 and the function value of the objective function F2;

[0248] Step S5.3.4, traverse each individual, calculate the domination number NC of each individual X c , and the calculation method is:

[0249] For the individual X c , its initial value of the domination number NC is 0; traverse the other NI - 1 individuals in turn. For the currently traversed individual X d , d = 1, 2, …, NI, and d ≠ c. If the function values of the individual X d and the individual X c are located in the same grid in the objective function space, then use Algorithm 1 to determine whether the individual X c dominates the individual X d . If it dominates, the domination number NC is incremented by 1; if the individual X d and the individual X cIf the function values of are located in different grids in the objective function space, Algorithm 2 is adopted to determine whether individual X c dominates individual X d . If it does, the domination count NC is incremented by 1. By traversing the other NI - 1 individuals in this way, the value of the domination count NC, which represents the number of individuals dominated by individual X c , is obtained. c

[0250] Among them, Algorithm 1 and Algorithm 2 are as follows:

[0251] Individual X c = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}, and the function value of the objective function F1 of individual X c is F 1c . The function value of the objective function F2 of individual X c is F 2c . The objective function value vector of individual X c is represented as: F c = {F 1c , F 2c};

[0252] The function value of the objective function F1 of individual X d is F 1d . The function value of the objective function F2 of individual X d is F 2d . The objective function value vector of individual X d is represented as: F d = {F 1d , F 2d};

[0253] Algorithm 1 is: When Conditions 1 and 2 are satisfied, individual X c dominates individual X d ;

[0254] Condition 1: F 1c ≥F 1d +∈, and F 2c ≥F 2d +∈;

[0255] Condition 2: There exists F xc , x = 1, 2, satisfying the relationship: F xc >F 1d +∈, and F xc >F 2d +∈;​

[0256] Algorithm 2 is as follows: When and only when one of Condition 3 and Condition 4 holds, individual X c dominates individual X d :

[0257] Condition 3:

[0258] Wherein: represents and satisfies the relationship between Condition 1 and Condition 2, represents rounding down;

[0259] Condition 4:

[0260] Wherein: |||| is the modulo operation; & is the logical AND;

[0261] Step S5.3.5, according to the population size NI and the ratio γ, obtain the number of individuals in the dominant population archive size and the number of individuals in the basic population population size;

[0262] Sort each individual in descending order of the domination number. If the domination numbers of two individuals are the same, sort them arbitrarily; among the NI individuals after sorting, select the top archive size number of individuals to form the dominant population, and the other individuals form the basic population.

[0263] Step S5.4, adopt a multi-recombination operator fusion strategy to perform crossover operations on the dominant population and the basic population to generate a new population; then return to Step S5.3, and re-adopt the individual domination algorithm to divide the new population into the dominant population and the basic population, and so on in a loop until the set number of loops is reached, and execute Step S5.5;

[0264] Step S5.4 is specifically as follows:

[0265] Step S5.4.1, set up a crossover operator pool; there are N_CO crossover operators in the crossover operator pool;

[0266] Step S5.4.2, initially, set each crossover operator CO to have the same selection probability PC;

[0267] Step S5.4.3, select a crossover operator for crossover operation and update the selection probability PC of the crossover operator:

[0268] ① Based on the current selection probability PC of each crossover operator, select crossover operator A from the crossover operator pool; ② According to the number of parents N A , randomly select a parent individual X from the dominant populationA1 , select N - 1 parental individuals from the basic population through tournament selection, denoted as: parental individual X A , where B = 2, 3, …, N AB ;

[0269] ③ Use crossover operator A to perform crossover operation on parental individual X A1 and parental individual X AB to generate N A offspring individuals, and delete the N A original parental individuals from the population;

[0270] ④ According to the function values of the N A offspring individuals, determine the grid where they are located in the objective function space, add the offspring individuals within the range of the dominant population to the dominant population, and add other offspring individuals to the basic population; among them: the number of offspring individuals added to the dominant population generated by crossover operator A this time is SN_A;

[0271] ⑤ Use the following formula to update the selection probability PC A of crossover operator A:

[0272]

[0273] where: SN_V represents the total number of offspring individuals added to the dominant population generated by crossover operator V from the 1st cycle to the current cycle, V = 1, 2, …, N_CO; is a constant,

[0274] In the formula, the constant prevents the operator probability from reaching 0, thus ensuring that no operator "gets lost" during the execution of the algorithm.

[0275] ⑥ When this cycle ends, return to step ① and continue the next cycle until the set number of cycles is reached.

[0276] The multi - recombination operator fusion strategy of the present invention uses a pool including multiple crossover operators and establishes a feedback mechanism to compare the performance of the operators during the search process. The operators with excellent performance will obtain higher selection weights. Specifically, initially, the selection weights of all operators are equal, and their probabilities in the next selection are updated through the feedback mechanism, and the adjustment strategy is based on the number of solutions in the dominant population generated by the operators.

[0277] Step S5.5: For each individual in the new population, use the heuristic constraint processing method to perform heuristic constraint processing on the solution value of each individual, so as to obtain each individual with the updated solution value;

[0278] Specifically, step S5.5 is as follows:

[0279] For each individual X = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}, the dynamic balance constraint adjustment of the power generation flow in step S5.4.1 and the reservoir capacity constraint adjustment in step S5.4.2 are both executed:

[0280] Step S5.4.1, dynamic balance constraint adjustment of the outflow:

[0281] ① According to the water balance equation and the outflow equation, obtain the water volume difference ΔQT j between the water consumption and the constrained water volume of reservoir n during the entire scheduling period under the comprehensive scenario S n,j ;

[0282] ② Use the following formula to adjust the water volume difference ΔQT n,j to the outflow QT j of each time period t of reservoir n under the comprehensive scenario S n,j,t , and obtain the corrected outflow

[0283]

[0284] ③ Use the following formula to further correct the corrected outflow by applying the outflow limit constraint condition, and obtain the corrected outflow

[0285]

[0286] ④ Thus, obtain the outflow of each time period t of reservoir n under the comprehensive scenario S j after correction; judge whether the water balance equation is satisfied for reservoir n during the entire scheduling period under the comprehensive scenario S j ; if it is satisfied, complete the dynamic balance constraint adjustment of the outflow; otherwise, return to step ① and perform corrections cyclically;

[0287] Step S5.4.2, reservoir capacity constraint adjustment:

[0288] ① Calculate the reservoir capacity V j of hydropower station n at time period t under the comprehensive scenario S n,j,t according to the outflow QT j of each time period t of reservoir n of the current individual X under the comprehensive scenario S n,j,t ;

[0289] ②Judge the storage capacity V of hydropower station n at time period t under the comprehensive scenario S j whether it meets the storage capacity limit constraint condition; if not, when the storage capacity V n,j,t exceeds the upper bound of the storage capacity of hydropower station n at time period t n,j,t then execute ③; when the storage capacity V is lower than the lower bound of the storage capacity of hydropower station n at time period t n,j,t V n,t then execute ④;

[0290] ③Average the excess amount to each time period t of the scheduling period, that is: obtain the average value of the excess amount

[0291] Then, average it to each time period t of the scheduling period according to the average value of the excess amount to complete the correction of the storage capacity V of hydropower station n at each time period t under the comprehensive scenario S j ; n,j,t ;

[0292] If after completing the correction of the storage capacity V of reservoir n at all time periods t under the comprehensive scenario S j there is still a storage capacity V at time period t that exceeds the upper bound of the storage capacity n,j,t then continue to average the excess amount to the out - flow of adjacent reservoirs; n,j,t

[0293] V n,t V n,t ④Average the shortage amount V n,t -V n,j,t to each time period t of the scheduling period to obtain the average shortage amount ΔV n,j,t =( V n,t -V n,j,t ) / T; then, reduce the storage capacity V of other time periods t n,j,t by ΔV n,j,t and supplement it to the storage capacity V of the current time period t n,j,t ; complete the correction of the storage capacity V of hydropower station n at time period t under the comprehensive scenario S j ; n,j,t ;

[0294] If after completing the correction of the storage capacity V of reservoir n at all time periods t under the comprehensive scenario S j there is still a time period t with a storage capacity V that is lower than the lower bound of the storage capacity n,j,t V n,t then supplement the out - flow of adjacent reservoirs to the storage capacity V of the time period t with a storage capacity lower than the lower bound n,j,t V n,t ; n,j,t

[0295] ​​​Step S5.6: Based on the updated solution values of each individual obtained in Step S5.5, calculate the objective function values of the objective function F1 and the objective function F2 corresponding to each individual.

[0296] Step S5.7: According to the obtained objective function values corresponding to each individual, determine whether the iteration termination condition is satisfied through the restart mechanism. If it is satisfied, execute Step S5.8; if not, determine whether the restart mechanism is required. If it is required, start the restart mechanism, update the basic population, and return to execute Step S5.4; if not, return to execute Step S5.3.

[0297] In this step, to determine whether the restart mechanism is required, if it is required, start the restart mechanism and update the basic population, specifically:

[0298] When the algorithm search stalls, start the restart mechanism: According to the current ratio γ and the current number of individuals in the dominant population archive size, use the formula archive size / γ to obtain the number of individuals in the basic population.

[0299] Empty the current basic population, and fill the basic population with the corresponding number of individuals with the top-ranked domination number NC in the current dominant population. The insufficient part is obtained by mutating the individuals in the dominant population, so as to obtain the updated basic population.

[0300] Therefore, in the present invention, in order to avoid the algorithm falling into local optimum, a population restart mechanism is designed to detect the algorithm search stall. If the algorithm search stalls, the restart mechanism is used to regenerate the basic population. This mechanism enables the algorithm to re-search when it discovers stalling or falling into a local optimal solution to find a better solution.

[0301] Step S5.8: Output the solution values of each individual in the dominant population. The solution value of each individual is a scheduling plan, and thus a multi-objective optimization scheduling plan set is obtained, thereby completing the multi-objective optimization scheduling for the medium and long term of the hydropower station group.

[0302] A medium and long term multi-objective stochastic scheduling optimization method considering complementary water-wind-solar power provided by the present invention can be mainly described in the following steps:

[0303] S1: Obtain the historical value time series of runoff, wind power output, and photovoltaic power output.

[0304] S2: Construct the state transition probability matrix and the cumulative probability state transition matrix in the Markov chain.

[0305] S3: Use the Monte Carlo sampling method to obtain the initialized water-wind-solar scenario set, and reduce the initial scenario set through the synchronous back substitution elimination method to obtain the final scenario set.

[0306] S4: Construct a multi-objective optimization model for medium- and long-term reservoir group scheduling with the goal of maximizing the expected power generation within the scheduling period and maximizing the expected minimum output of the multi-energy complementary system;

[0307] S5: Generate individual coding information; generate initial solutions for them respectively to obtain an initial population; use box domination to obtain a dominant population and a basic population;

[0308] S6: Pass the obtained population through a multi-operator fusion strategy and use a heuristic constraint handling method; for the specific water volume dynamic balance constraint, first calculate the amount of water exceeding the constraint of the reservoir storage capacity, and then adjust the outflow of the reservoir during this period according to the violation amount, so that the reservoir storage capacity returns to the constraint range; for the reservoir storage capacity constraint, first calculate the amount of water exceeding the constraint of the reservoir storage capacity, and then adjust the outflow of the reservoir during this period according to the violation amount, so that the reservoir storage capacity returns to the constraint range. At the same time, in order to maintain the dynamic water balance, this strategy also makes an equal and opposite adjustment to the flow of the next period, which can satisfy the constraint of the current storage capacity without changing the storage capacity of other periods, and then calculate the objective function value;

[0309] S7: Judge whether the iteration termination condition is met according to the obtained objective function value through the restart mechanism. If it is terminated, output the corresponding solution and Pareto front value. If there is no termination condition, it is necessary to judge whether the restart mechanism is needed. If so, update the existing basic population and execute step S6 until the iteration termination condition is met, output the corresponding solution and corresponding plan. Otherwise, execute steps S5 - S6, use the domination rule again to obtain a new dominant population and basic population until the iteration termination condition is met, output the corresponding solution and corresponding plan, so as to complete the multi-objective optimal scheduling of the medium- and long-term reservoir group.

[0310] The present invention proposes a medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light, and has the following beneficial effects: Driven by climate change and the "dual carbon" goal, renewable energy represented by wind and photovoltaic power has developed rapidly in recent years. Making full use of the flexibility of hydropower and the natural complementarity between different resources for complementary power generation is an effective way to promote the consumption of new energy. The present invention combines the correlation analysis of water, wind and light with multi-objective optimization, constructs a set of decision-making processes, and this set of decisions is based on the obtained runoff and wind and light output scenario sets to efficiently solve the medium- and long-term multi-objective optimal scheduling of water, wind and light. Compared with the prior art, the present invention comprehensively considers the time correlation of water, wind and light, generates a scenario set on this basis to describe the characteristics of runoff and wind and light output, and constructs a multi-objective optimal scheduling and optimization algorithm to provide a feasible strategy for solving the medium- and long-term multi-objective optimal scheduling of water, wind and light complementarity, and provide theoretical and method support for the decision-making of hydropower operation scheduling.

[0311] An embodiment of the present invention further provides a medium- and long-term multi-objective stochastic scheduling optimization system considering the complementarity of water, wind, and light, including:

[0312] A data acquisition module, configured to obtain historical time series of runoff, wind power output, and photovoltaic power output;

[0313] A scenario generation module, which first analyzes the temporal correlation of water, wind, and light using a Markov chain, then obtains an initial scenario set of runoff and wind and light power outputs using the Monte Carlo sampling method, and finally obtains a final scenario set that meets the requirements based on a scenario elimination method based on probability distance;

[0314] A model construction module, configured to construct a medium- and long-term scheduling multi-objective optimization model for a reservoir group with the objectives of maximizing the expected power generation within the scheduling period and maximizing the expected minimum output of a multi-energy complementary system;

[0315] An initialization module, configured to generate individual coding information, where each individual information corresponds to a scenario, and generate initial solutions for them to obtain an initial population; use a domination algorithm to obtain a dominant population and a basic population;

[0316] A target function value solving module, which uses a multi-operator fusion strategy for the obtained population and a heuristic constraint handling method; for the specific water volume dynamic balance constraint, first calculate the amount of water exceeding the constraint of the reservoir storage capacity, and then adjust the outflow of the reservoir during this period according to the violation amount, so that the reservoir storage capacity returns to the constraint range; for the reservoir storage capacity constraint, first calculate the amount of water exceeding the constraint of the reservoir storage capacity, and then adjust the outflow of the reservoir during this period according to the violation amount, so that the reservoir storage capacity returns to the constraint range. At the same time, in order to maintain the dynamic water volume balance, this strategy also makes an equal and opposite adjustment to the flow in the next period, which can satisfy the constraint of the current storage capacity without changing the storage capacity in other periods, and then calculate the target function value;

[0317] A loop iteration module, which determines whether the iteration termination condition is met through a restart mechanism according to the obtained target function value. If it is terminated, the corresponding solution and Pareto front value are output. If there is no termination condition, it is necessary to determine whether a restart mechanism is required. If so, update the existing basic population, execute the target function value solving module until the iteration termination condition is met, and output the corresponding solution and corresponding plan. Otherwise, obtain a new dominant population and basic population again, and then execute the target function value solving module until the iteration termination condition is met, and output the corresponding solution and corresponding plan;

[0318] A scheduling module, configured to perform stochastic scheduling of water, wind, and light according to the obtained corresponding solution and its plan.

[0319] It should be understood that the functional unit modules in various embodiments of the present invention can be concentrated in one processing unit, or each unit module exists physically alone, or two or more unit modules are integrated into one unit module, and can be implemented in the form of hardware or software.

[0320] An embodiment of the present invention further provides an electronic terminal, including:

[0321] One or more memories, on which computer programs or instructions are stored;

[0322] One or more processors, configured to load and execute the computer programs or instructions to implement the medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind, and light as described above.

[0323] An embodiment of the present invention further provides a computer-readable storage medium, on which computer programs or instructions are stored, and when the computer programs or instructions are called by a computer, the medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind, and light as described above is implemented.

[0324] The present invention discloses a medium- and long-term multi-objective scheduling optimization method, system, terminal, and medium considering the complementarity of water, wind, and light; the Markov chain is used to characterize the time correlation of each of the three variables of water, wind, and light through the state transition matrix, the Monte Carlo sampling method is used to generate the time correlation scenario sequence of water, wind, and light, and the synchronous back substitution elimination method based on probability distance is used to reduce the initial scenario set to achieve a high-quality approximation of the initial scenario with a small number of scenarios; based on the obtained water, wind, and light scenario set, a multi-objective stochastic scheduling optimization model with the maximum expected power generation and the maximum expected minimum output of water, wind, and light is constructed on a medium- and long-term scale of a reservoir group; the Borg algorithm is used and a heuristic constraint handling strategy is designed for the model constraints to finally solve the problem. The present invention describes the time series characteristics of runoff, wind power, and photovoltaic output, and the established optimization model and designed solution algorithm can efficiently solve the medium- and long-term multi-objective optimization problem of hydropower, providing a feasible strategy for the coordinated operation of multiple energy sources to improve the renewable energy consumption capacity of the power system.

[0325] It can be understood that the same or similar parts in the above embodiments can be referred to each other, and the content not detailed in some embodiments can be seen in the same or similar content of other embodiments.

[0326] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0327] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0328] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0329] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0330] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light, characterized in that It includes the following steps: Step S1, determine that the resource variables E1, E2, and E3 are the runoff inflow, wind power output, and photovoltaic output respectively; for each resource variable E b , where b = 1, 2, 3, the time series of historical values of the resource variables in the target area for historical years are obtained. Therefore, the time series of historical runoff values, the time series of historical wind power output values, and the time series of historical photovoltaic output values are obtained respectively; Step S2. For each resource variable E b , determine the corresponding state space; then, apply a Markov chain to perform a time correlation analysis on the historical value time series of this resource variable E b to obtain the cumulative probability state transition matrix Q b of this resource variable E b ; Determine that the research time scale is the length Δt of time period t; set the number B of simulation scenarios to be initially generated and the initial value of the resource variable E b of; Based on the resource variable E b the cumulative probability state transition matrix Q b , using a random number method, generate the simulated value time series of the resource variable E at each time period t within a year b , forming a simulation scenario; execute B times in a loop to obtain B simulation scenarios of the resource variable E b , forming the initial simulation scenario set of the resource variable E b ; Step S3, set the number of simulation scenario targets J; adopt a simulation scenario reduction algorithm to reduce B simulation scenarios in the initial simulation scenario set of each resource variable E b to obtain a reduced simulation scenario set with J simulation scenarios; Therefore, for the resource variables of runoff, wind power output, and photovoltaic output, simulation scenario sets with J simulation scenarios are obtained respectively, and each simulation scenario in the simulation scenario set of each resource variable is randomly arranged. Thus, the final runoff simulation scenario set SCE1 = {S 11 , S 12 ,..., S 1J}, wind power output simulation scenario set SCE2 = {S 21 , S 22 ,..., S 2J}, and photovoltaic output simulation scenario set SCE3 = {S 31 , S 32 ,..., S 3J}; Combine the runoff simulation scenario set SCE1, the wind power output simulation scenario set SCE2, and the photovoltaic power output simulation scenario set SCE3 to form a comprehensive scenario set SC = {S1, S2,..., S J}; Among them, for the comprehensive scenario S in the comprehensive scenario set SC j , j = 1, 2,..., J, has the following relationship: S j = {S 1j , S 2j , S 3j}, meanwhile, the probability p j of the comprehensive scenario S j = (p 1j + p 2j + p 3j ) / 3; p 1j , p 2j , p 3j are the probabilities of the runoff simulation scenario S 1j , the wind power output simulation scenario S 2j and the photovoltaic power output simulation scenario S 3j respectively; Step S4: Construct a multi-objective optimization model for medium- and long-term reservoir group scheduling with the goals of maximizing the expected power generation within the scheduling period and maximizing the expected minimum output of the multi-energy complementary system; Step S5, for each comprehensive scenario S j , using the runoff simulation scenario S 1j , wind power output simulation scenario S 2j and photovoltaic power output simulation scenario S 3j as the input value of the time series of simulation values, by solving the multi-objective optimization model of medium and long-term scheduling of the reservoir group, the discharge flow QT j of hydropower station n at each time period t in the scheduling period under the comprehensive scenario S n,j,t is obtained to achieve the coordinated scheduling of water, wind and light 2. A medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light, characterized in that Specifically, step S2 is as follows: Step S2.1, for each resource variable E b , determine the corresponding state space; there are β states in the state space, which are state 1, state 2, …, state β respectively; Step S2.2, apply Markov chain to perform time correlation analysis on the historical value time series of the resource variable E b to obtain the state transition probability matrix G b of the resource variable E b : where: g u',v' represents the resource variable E b is the state transition probability from state u' to state v'; u' = 1, 2,..., β; v' = 1, 2,..., β; Step S2.3, based on the state transition probability matrix G b of the resource variable E b , obtain the cumulative probability state transition matrix Q b of the resource variable E b : where: q u',v' represents the resource variable E b the cumulative probability of the state transition from state u' to state v'; q u',v' = g u',1 + g u',2 +... + g u',v' ; Step S2.4, generate resource variable E b A simulation scenario of Step S2.4.1, set the initial value a0 of the resource variable E, and obtain the initial state β0 according to the initial value a0; where β0 = 1, 2,..., β; b ​ Step S2.4.2: Let the time period t = 1; Step S2.4.3, generate random numbers e1, e1 ∈ [0, 1]; in the cumulative probability state transition matrix Q b locate the row vector corresponding to the state β t-1 in it, that is, the row vector of the β t-1 -th row: Step S2.4.4, traverse the elements of the row vector in the forward direction from the front to the back. When an element greater than or equal to the random number e1 is traversed , for v' = 1, 2,..., β, the state v' after the state transition corresponding to this element is the state at time period t, and is re-expressed as state β t ; the resource variable E t corresponding to state β b has a numerical range of: [a l , a h ; Step S2.4.5, generate random numbers e2, e2 ∈ [0, 1]; through the formula a t = a t-1 + e2 × (a h - a l ), obtain the simulated value a b of the resource variable E t at time period t; where a t-1 is the simulated value of the resource variable E b at time period t - 1; Step S2.4.6: Judge whether the time period t reaches the preset total number of scheduling time periods T. If it reaches, execute step S2.4.7; if not, let t = t + 1 and return to step S2.4.3; Step S2.4.7, output the resource variable E obtained at this time b The simulated values in time periods 1, 2, …, T form a time series of simulated values a1, a2, ..., a T , forming a simulated scenario; Step S2.4.

8. Determine whether the B types of simulation scenarios have been obtained. If not, return to Step S2.4.

2. By executing Steps S2.4.2 to S2.4.7, another simulation scenario is formed. Keep looping like this until the B types of simulation scenarios are obtained to form the resource variable E. b The initial simulation scenario set.

3. A medium- and long-term multi-objective stochastic scheduling optimization method considering water-wind-solar complementary according to claim 2, characterized in that In step S3, the simulation scenario reduction algorithm is adopted to reduce B simulation scenarios in the initial simulation scenario set of each resource variable E b to obtain a reduced simulation scenario set with J simulation scenarios, specifically as follows: Step S3.1, resource variable E b The initial simulation scenario set of B includes B simulation scenarios, expressed as: simulation scenario set RS = {ξ1, ξ2,..., ξ B}; ξ1, ξ2,..., ξ Step S3.2, set the initial probability \(p\) of each simulation scenario \(\xi\) k to be equal, all being \(1 / B\); where \(x = 1, 2, \cdots, B\); establish a set \(DS\) for storing the scenarios to be eliminated; k ​ Step S3.3: Set the set DS to be empty; Step S3.4, traverse each simulation scenario ξ in the simulation scenario set RS k ; For each traversed simulation scenario ξ k , determine the simulation scenario ξ k nearest to the simulation scenario ξ k" , that is, satisfying the relation: DT k (k") = minDT k,s , k ∈ RS, s ∈ RS, k ≠ s Where: DT k,s represents the distance between the simulated scenario ξ k and the simulated scenario ξ s ; DT k (k") represents the distance between the simulated scenario ξ k and the nearest simulated scenario ξ k" ; Adopt the formula PD k = p k * DT k (k"), k ∈ RS, to calculate the minimum distance probability PD k corresponding to the simulation scenario ξ k ; Step S3.5, compare the minimum distance probabilities PD k of each simulation scenario ξ k in the simulation scenario set RS, and use the formula PD d = minPD k , k ∈ S, to obtain the minimum distance probability PD k in the simulation scenario set RS, which is denoted as PD d ; this PD d corresponds to the simulation scenario ξ d , and the simulation scenario closest to the simulation scenario ξ d is ξ d* ; Step S3.6, let the simulated scenario set RS = RS - {ξ d}, DS = DS + {ξ d}, and let p d* = p d* + p d ; Determine whether the number of simulated scenarios in the simulated scenario set RS has been reduced to J. If so, execute Step S3.7; if not, return to Step S3.4; Step S3.7: Obtain a reduced set of simulation scenarios with J simulation scenarios, and each simulation scenario in this set of simulation scenarios has a probability.

4. A medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light, characterized in that In step S4, in the medium- and long-term scheduling multi-objective optimization model of the reservoir group, the outflow discharge QT of hydropower station n at time period t under the comprehensive scenario S j is used as the decision variable of the model; n,j,t ​ The objective function includes objective function F1 and objective function F2; Objective function F1: Maximize the expected power generation of the multi-energy complementary system, and the expression is: Where: P n,j,t represents the output of hydropower station n at time period t under the comprehensive scenario S j ; t = 1, 2, ..., T; T is the total number of medium- and long-term scheduling time periods; n = 1, 2, ..., N; N represents the number of hydropower stations; j = 1, 2, ..., J; J represents the number of comprehensive scenarios in the comprehensive scenario set SC = {S1, S2, ..., S J}; p j represents the probability of the comprehensive scenario S j ; ΔT represents the length of the medium- and long-term scheduling time period represents the output of the wind turbine at time period t under the comprehensive scenario S j ; represents the output of the photovoltaic unit at time period t under the comprehensive scenario S j ; Objective function F2: Maximize the expected minimum output of the multi-energy complementary system, and the expression is: Thus, objective function F1 and objective function F2 are established.

5. A medium- and long-term multi-objective stochastic scheduling optimization method considering complementary water-wind-solar power, as claimed in claim 4, wherein In the multi-objective optimization model for medium- and long-term reservoir group scheduling, the following constraint conditions are included: (1) Water balance equation: Where: V n,j,t+1 and V n,j,t are the reservoir capacities of hydropower station n at time periods t + 1 and t respectively; I n,j,t , QT n,j,t are the runoff inflow and outflow discharge of hydropower station n at time period t under the comprehensive scenario S j respectively; Ω(n) is the set of upstream hydropower stations of hydropower station n; QT m,j,t is the runoff inflow of the upstream hydropower station m of hydropower station n at time period t under the comprehensive scenario S j respectively. (2) Outflow equation: Where: QP n,j,t , S n,j,t are respectively the generated flow rate and the discharged water flow rate of hydropower station n in time period t under the comprehensive scenario S j ; (3) Outflow limit: Wherein: QT n,t and are the lower bound and upper bound of the total outflow of hydropower station n in period t, respectively; (4) Generation flow limit: Wherein: QP n,t and are the lower bound and upper bound of the power generation flow rate of the hydropower station n at time period t, respectively; (5) Output limit: Wherein: P n,t and are respectively the lower bound and the upper bound of the output of hydropower station n in period y; (6) Storage capacity limit: Wherein: V n,t and are the lower bound and upper bound of the reservoir capacity of the hydropower station n at time period t, respectively; (7) Initial and final water level constraints for scheduling: Where: Z n,Ini and Z n,End are the initial water level and the final water level of hydropower station n respectively; Z n,j,1 and Z n,j,T are the water levels of hydropower station n at time period 1 and time period T respectively. (8) Water level-storage capacity constraint: Wherein: is a non-linear function between the reservoir water level and the water storage volume; Z n,j,t is the water level of hydropower station n at time period t under the comprehensive scenario S j ; (9) Tail water-discharge constraint: Wherein: is a non-linear function between the tail water level and the discharge of hydropower station n; is the tail water level of hydropower station n at time t under the comprehensive scenario S j ; (10) Hydroelectric output generation function: Where: H n,j,t is the generating head of hydropower station n at time period t under the comprehensive scenario S j ; K n is the output coefficient of hydropower station n; Z n,j,t+1 is the water level of hydropower station n at time period t + 1 under the comprehensive scenario S j ; (11) Transmission capacity limit constraint: Where: W n represents the set of wind power plants associated with hydropower station n; P wd represents the output of wind power plant wd associated with hydropower station n; B n represents the set of photovoltaic power plants associated with hydropower station n; P sol represents the output of photovoltaic power plant sol associated with hydropower station n; P n represents the total output of hydropower station n; is the maximum value of the integrated water-wind-light transmission capacity of hydropower station n.

6. The medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light according to claim 5, characterized in that Specifically, step S5 is as follows: Step S5.1: Generate an individual code X; X = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}; where QT n,j,t means the runoff inflow of hydropower station n at time period t under the comprehensive scenario S j , which is a decision variable; according to the individual coding X, a corresponding number of individuals with initial solutions are generated according to the set population size, thus forming an initial population; Step S5.2: Determine the population size NI and the ratio γ; where the ratio γ is the number of individuals in the dominant population / the number of individuals in the basic population; Step S5.3: Use the individual dominance algorithm to determine the dominant population and the basic population; Step S5.4: Adopt a multi-recombination operator fusion strategy to perform crossover operations on the dominant population and the basic population to generate a new population; then return to step S5.3 and re-use the individual dominance algorithm to divide the new population into a dominant population and a basic population, and so on in a loop until the set number of loops is reached, and then execute step S5.5; Step S5.5: For each individual in the new population, use the heuristic constraint handling method to perform heuristic constraint handling on the solution value of each individual, so as to obtain each individual with an updated solution value; Step S5.6: Based on the updated solution values of each individual obtained in step S5.5, calculate the objective function values of objective function F1 and objective function F2 corresponding to each individual; Step S5.7: According to the objective function values corresponding to each individual obtained, judge whether the iteration termination condition is satisfied through the restart mechanism. If it is satisfied, execute step S5.8; if not, judge whether the restart mechanism is needed. If it is needed, start the restart mechanism, update the basic population, and return to execute step S5.4; if not, return to execute step S5.3; Step S5.8: Output the solution value of each individual in the dominant population. The solution value of each individual is a scheduling plan, and thus a multi-objective optimization scheduling plan set is obtained, thereby completing the multi-objective optimization scheduling for the medium- and long-term of the hydropower station group.

7. A medium- and long-term multi-objective stochastic scheduling optimization method considering water-wind-solar complementary, characterized in that Step S5.3 specifically is as follows: Step S5.3.1: Set the value of parameter γ; Step S5.3.2: Using the function value of objective function F1 as the abscissa, the function value of objective function F2 as the ordinate, and parameter ∈ as the grid side length, establish a grid-like objective function space; Step S5.3.3, for each current individual X c , where c = 1, 2, …, NI, determine the grid in the objective function space it is located according to the function values of its objective function F1 and objective function F2; Step S5.3.4, traverse each individual, and calculate the domination number NC of each individual X c , and the calculation method is as follows: For individual X c , the initial value of its domination number NC is 0; traverse the other NI - 1 individuals in sequence. For the currently traversed individual X d , d = 1, 2, …, NI, and d ≠ c. If the function values of individual X d and individual X c are in the same grid in the objective function space, then use Algorithm 1 to determine whether individual X c dominates individual X d . If it does, then increment the domination number NC by 1; if the function values of individual X d and individual X c are in different grids in the objective function space, then use Algorithm 2 to determine whether individual X c dominates individual X d . If it does, then increment the domination number NC by 1; thus complete the traversal of the other NI - 1 individuals to obtain the value of the domination number NC, which is the domination number NC of individual X c , representing the number of individuals dominated by individual X c ; Where: Algorithm 1 and Algorithm 2 are: Individual X c ={QT 111 ,…,QT 11T ,QT 121 ,…,QT 12T ,…,QT NJ1 ,…,QT NHT}, Individual X c 's function value of objective function ∈1 is F 1c ; Individual X c 's function value of objective function F2 is F 2c , Individual X c 's objective function value vector is expressed as: F c ={F 1c ,F 2c}; Individual X d The function value of the objective function F1 of 1d is F d ; The function value of the objective function F2 of Individual X 2d is F d , and the objective function value vector of Individual X d is represented as: F 1d = {F 2d}; Algorithm 1 is as follows: When Conditions 1 and 2 are satisfied, individual X c dominates individual X d ; Condition 1: F 1c ≥ F 1d + ∈, and F 2c ≥ F 2d + ∈; Condition 2: There exists F xc , x = 1, 2, satisfying the relationship: F xc > F 1d + ∈, and, F xc > F 2d + ∈; Algorithm 2 is as follows: Individual X is dominated by c when and only when one of Condition 3 and Condition 4 holds. c dominates Individual X d : Condition 3: Wherein: represents and the relationship that satisfies Condition 1 and Condition 2, represents rounding down; Condition 4: Where: |||| is the modulo operation; & is the logical AND; Step S5.3.5: According to the population size NI and ratio γ, obtain the number of individuals in the dominant population archive size and the number of individuals in the basic population population size; Sort each individual in descending order of the domination number. If two individuals have the same domination number, sort them arbitrarily. Among the NI individuals after sorting, select the top archive size number of individuals to form the dominant population, and the other individuals form the basic population.

8. A medium- and long-term multi-objective stochastic scheduling optimization method considering the complementarity of water, wind and light, characterized in that Step S5.4 specifically is as follows: Step S5.4.1: Set up a crossover operator pool; there are N_CO crossover operators in the crossover operator pool; Step S5.4.2: Initially, set each crossover operator CO to have the same selection probability PC; Step S5.4.3: Select a crossover operator for crossover operation and update the selection probability PC of the crossover operator: ① Based on the current selection probabilities PC of each crossover operator, select crossover operator A from the crossover operator pool; ②According to the number of parental individuals N required by the crossover operator A A , randomly select a parental individual X from the dominant population A1 , and select N A - 1 parental individuals from the basic population, denoted as: parental individual X AB , B = 2, 3,..., N A ; ③ Use the crossover operator A to perform crossover operations on the parental individual X A1 and the parental individual X AB to generate N A offspring individuals, and delete the N A original parental individuals from the population; ④According to N A Determine the grid in the objective function space where the function values of the N offspring individuals are located. Add the offspring individuals within the range of the dominant population to the dominant population, and add the other offspring individuals to the basic population. Among them, the number of offspring individuals generated by the crossover operator A that are added to the dominant population this time is SN_A; ⑤ Update the selection probability \(P_C\) of the crossover operator \(A\) using the following formula A :[[]]END]] Where: SN_V represents the total number of offspring individuals added to the dominant population generated by the crossover operator V from the 1st iteration to the current iteration, V = 1, 2, …, N_CO; is a constant, ⑥ When this loop ends, return to step ① and continue the next loop until the set number of loops is reached.

9. A medium- and long-term multi-objective stochastic scheduling optimization method considering complementary water-wind-solar power, characterized in that Step S5.5 specifically is as follows: For each individual X = {QT 111 , …, QT 11T , QT 121 , …, QT 12T , …, QT NJ1 , …, QT NJT}, perform the adjustment of the dynamic balance constraint of the power generation flow in step S5.4.1 and the adjustment of the reservoir capacity constraint in step S5.4.2: Step S5.4.1: Adjust the dynamic balance constraint of the out-of-storage flow rate: ①According to the water balance equation and the outflow discharge equation, the water volume difference ΔQT between the water consumption and the constrained water volume of reservoir n during the entire scheduling period under the comprehensive scenario S j is obtained n,j ; ② Using the following formula, adjust the water volume difference ΔQT n,j to the outflow rate QT of reservoir n at each time period t under the comprehensive scenario S j to obtain the corrected outflow rate n,j,t in ③ Using the following formula, apply the constraint condition of the outbound flow limit to further correct the corrected outbound flow to obtain the further corrected outbound flow ④Thus, the modified outflows of reservoir n at each time period t under the comprehensive scenario S are obtained; it is judged whether the water balance equation is satisfied for reservoir n under the comprehensive scenario S j during the entire scheduling period; if it is satisfied, the adjustment of the dynamic balance constraint of the outflows is completed; j ​ Otherwise, return to step ① and loop for correction; Step S5.4.2: Adjust the reservoir storage capacity constraint: ①According to the discharge flow QT of reservoir n of the current individual X at each time period t under the comprehensive scenario S j the storage capacity V of hydropower station n at time period t under the comprehensive scenario S n,j,t is calculated j ; n,j,t ​ ② Determine the reservoir capacity V of hydropower station n at time period t under the comprehensive scenario S j ; if it does not meet the reservoir capacity limit constraint condition; if not, when the reservoir capacity V n,j,t exceeds the upper bound of the reservoir capacity of hydropower station n at time period t n,j,t , then execute ③; when the reservoir capacity V is lower than the lower bound of the reservoir capacity of hydropower station n at time period t n,j,t , then execute ④; V n,t ​ ③Average the excess amount over each time period t of the scheduling period, that is: obtain the average value of the excess amount Then, according to the average value of the excess it is averaged to each time period t of the scheduling period to complete the correction of the reservoir capacity V of hydropower station n at each time period t under the comprehensive scenario S j ; n,j,t ​ If, after completing the correction of the reservoir capacity V of reservoir n for all time periods t under the comprehensive scenario S j there are still time periods t where the reservoir capacity V exceeds the upper bound of the reservoir capacity n,j,t after the correction, the excess amount is continuously averaged into the outflow of adjacent reservoirs; for the reservoir capacity V of the time period t n,j,t ​ ④Average the insufficient quantity V n,t -V n,j,t over each time period t of the scheduling period to obtain the average insufficient quantity ΔV n,j,t =( V n,t -V n,j,t ) / T; then, reduce the reservoir capacity V n,j,t of other time periods t by ΔV n,j,t and supplement it to the reservoir capacity V n,j,t of the current time period t; complete the correction of the reservoir capacity V j of hydropower station n at time period t under the comprehensive scenario S n,j,t ; If, after completing the correction of the reservoir capacity V for all time periods t of reservoir n under the comprehensive scenario S j , there is still a shortage of the lower bound of the reservoir capacity n,j,t for the reservoir capacity V of time period t V n,t , then the outflow of the adjacent reservoir is supplemented to the lower bound of the shortage of the reservoir capacity n,j,t for the reservoir capacity V of time period t V n,t . n,j,t .

10. A medium- and long-term multi-objective stochastic scheduling optimization method considering complementary water, wind and light, characterized in that, In step S5.7, determine whether to perform a restart mechanism. If so, start the restart mechanism and update the basic population. Specifically: When the algorithm search stagnates, start the restart mechanism: According to the current ratio γ and the current number of individuals in the dominant population archive size, use the formula archive size / γ to obtain the number of individuals in the basic population; Empty the current basic population, and fill the corresponding number of individuals with the highest domination number NC in the current dominant population into the basic population. The insufficient part is obtained by mutating the individuals in the dominant population, thereby obtaining the updated basic population.

Images