Reservoir multi-objective optimization scheduling scheme optimization method and system based on multi-party coupling

Abstract

The invention discloses a reservoir multi-target optimization scheduling scheme optimization method and system based on multi-party coupling, and belongs to the field of reservoir scheduling. The system comprises the steps that firstly, boundary conditions of a simulation system are determined, and then a reservoir multi-target optimization scheduling model is constructed; and then solving the reservoir multi-objective optimization scheduling model by using a non-dominated sorting genetic algorithm of the reference points, verifying the effectiveness of the solving method by using a standard multi-objective test function, and evaluating and optimizing a reservoir scheduling scheme by using a method of combining a grey target theory, a cumulative foreground theory and a fuzzy mathematical method. According to the system, the multi-objective optimization problem among flood control, power generation and ecology of the reservoir is coordinated, decision support is provided for formulating a reservoir dispatching scheme of multi-objective optimization dispatching of the reservoir, the contradiction among flood control, power generation and downstream river ecology can be effectively coordinated, and the flood control and power generation benefits are increased on the basis of considering the downstream ecological flow demand as much as possible.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy projects, and particularly to a method and system for optimizing and selecting a multi-objective optimal operation plan of a reservoir based on multi-party coupling. Background Art

[0002] Currently, reservoirs undertake comprehensive utilization tasks such as flood and ice prevention, water supply and irrigation, and power generation in the basin. There are competitive relationships among various objectives, and it is necessary to achieve the coordination of multiple objectives by constructing a multi-objective optimal operation of the reservoir. When solving complex high-dimensional multi-objective optimization problems, various methods such as the weight method and the constraint method can be used to transform them into single-objective optimization problems, and then traditional single-objective optimization algorithms are used for solving. This method is inefficient and cannot provide comprehensive information for decision-makers, and has strong subjectivity.

[0003] With the sudden change of the water and sediment conditions in the basin, the contradiction between the multi-objective optimal operation mode of the reservoir aiming at improving the comprehensive benefits of the reservoir and the operation mode mainly aiming at flood control and ecology has become increasingly prominent. The complexity of the multi-objective reservoir operation model has gradually increased. There are defects such as weak local search ability and low efficiency in solving the multi-objective reservoir operation model. When dealing with complex high-dimensional multi-objective optimization problems, it is difficult to handle the problem of complex irregular Pareto fronts.

[0004] The invention patent with the patent number CN202010871614.2 and the patent name "An Optimal Operation Method for a Multi-Objective Medium- and Long-Term Stochastic Operation Model Based on Reservoir Ecological Power Generation" discloses a method for constructing a multi-objective medium- and long-term stochastic operation model of a reservoir and an optimal operation method, and adopts the ideal ecological flow in the Tennant method. The establishment of the optimal operation model includes: first, establishing a reservoir power generation objective function and an ecological flow closeness objective function, and then obtaining a multi-objective function of power generation and ecological comprehensive benefits through the normalization processing of these two objective functions, and trying to maximize the power generation and ecological comprehensive benefits. Then, the constraint conditions of the reservoir ecological stochastic operation model are described, and then the runoff stochastic process and the Markov decision process of random variables are introduced into the medium- and long-term stochastic optimal operation problem of reservoir ecological power generation. This existing technology only considers the power generation and ecological flow closeness objective functions, which is not comprehensive enough and is prone to one-sidedness brought by a single objective or a few objectives; in addition, there may be large differences in the requirements for ecological flow in different rivers, different seasons, and different ecosystems. Only adopting the ideal ecological flow in the Tennant method cannot comprehensively and accurately reflect the actual complex and changeable ecological needs, resulting in insufficient accuracy and flexibility in ensuring ecological flow. Summary of the Invention

[0005] The objective of the present invention is to provide a method and system for optimizing and selecting reservoir multi-objective operation schemes based on multi-party coupling, which can coordinate the multi-objective optimization problem among reservoir flood control, power generation, and ecology, explore the relationships among the flood control benefits, power generation benefits, and ecological benefits of the reservoir during flood, normal, and dry years, and obtain the optimized reservoir operation scheme of the Pareto optimal solution set.

[0006] A method for optimizing and selecting reservoir multi-objective operation schemes based on multi-party coupling includes the following steps:

[0007] S1: Taking the outflow discharge in each time period as the decision variable, constructing the total objective function of the reservoir with the objective functions of economic benefit, ecological benefit, and flood control benefit, and using it as the reservoir multi-objective operation optimization model; the constraint conditions of the multi-objective operation optimization model include water balance constraint, water level constraint, outflow discharge constraint, power station output constraint, and non-negativity constraint;

[0008] S2: Using the non-dominated sorting genetic algorithm with reference points to solve the reservoir multi-objective operation optimization model and obtain the Pareto front of each objective;

[0009] S3: Using the standard function to verify the effectiveness of the solution algorithm of the reservoir multi-objective operation optimization model in step S2;

[0010] S4: Constructing an evaluation index system for optimizing and selecting reservoir operation schemes according to the first-level evaluation indexes, and the first-level indexes include power generation index, ecological index, and flood control index;

[0011] S5: Determining the decision matrix of the reservoir operation scheme set according to the grey target theory, determining the evaluation index weights through the cumulative prospect theory, obtaining the membership degrees of each reservoir operation scheme under each evaluation level, and finally screening out the reservoir operation scheme with the optimal comprehensive benefit according to the maximum membership degree principle.

[0012] Preferably, the objectives of the reservoir in step S1 include the maximum average power output of the power station during the single-reservoir operation period, the minimum average change degree of the suitable ecological flow in the downstream river channel of the single reservoir, and the maximum peak shaving rate during the single-reservoir operation period, which respectively correspond to representing the economic benefit objective, the ecological benefit objective, and the flood control benefit objective; determining the economic benefit objective according to the decision variable, the average water head of the reservoir in the set time period, and the comprehensive power output coefficient of the power station of the reservoir; determining the ecological benefit objective according to the suitable ecological flow of the downstream river channel of the reservoir in the set time period and the decision variable; determining the flood control benefit objective according to the decision variable and the average inflow of the reservoir in the set time period.

[0013] Preferably, the constraint conditions of the multi-objective operation optimization model described in step S1 include water balance constraint conditions, water level constraint conditions, outflow discharge constraint conditions, power station output constraint conditions, and non-negativity constraint conditions.

[0014] Preferably, the water balance constraint condition is that the change in the reservoir storage volume within a time period is equal to the net difference between the average inflow and outflow of the reservoir within that time period multiplied by the time period duration;

[0015] The water level constraint condition is that the initial water level of the reservoir in the set time period is not greater than the upper limit value of the water level of the reservoir in the set time period and not less than the lower limit value of the water level of the reservoir in the set time period;

[0016] The outflow discharge constraint condition is that the average outflow discharge value of the reservoir in the set time period is not greater than the upper limit value of the outflow discharge of the reservoir in the set time period and not less than the lower limit value of the outflow discharge of the reservoir in the set time period;

[0017] The power generation output constraint condition is that the average power generation output of the reservoir in the set time period is not greater than the upper limit value of the power generation output of the reservoir in the set time period and not less than the lower limit value of the power generation output of the reservoir in the set time period;

[0018] The non - negative constraint condition is that all variables in the water balance constraint condition, water level constraint condition, outflow discharge constraint condition, and power generation output constraint condition are non - negative values.

[0019] Preferably, the standard multi - objective test functions selected in step S3 adopt the DTLZ1 and DTLZ2 functions of the three - objective minimization test function, and the effectiveness of the solution algorithm is evaluated through test indexes, and the test indexes include: inverted generational distance, spacing index, and algorithm running time.

[0020] Preferably, the evaluation indexes in step S4 include first - level indexes: power generation index, ecological index, and flood control index; the second - level indexes of the power generation index are composed of the average power generation output indexes of the power stations corresponding to each reservoir in the reservoir operation plan during the corresponding operation period, the second - level indexes of the ecological index are composed of the average change degree indexes of the downstream ecological flow during the corresponding operation period of each reservoir, and the second - level indexes of the flood control index are composed of the maximum peak - shaving rate indexes during the corresponding operation period of each reservoir. The average power generation output index and the maximum peak - shaving rate index in the second - level indexes are defined as benefit - type indexes, and the ecological flow change degree index is defined as a cost - type index.

[0021] Preferably, in step S5, the grey target theory is used to determine the decision matrix of the reservoir operation plan set; first, the decision matrix of the plan set with respect to the index set is determined, and the transformed decision matrix is converted into a normalized decision matrix through decision matrix normalization. According to the index nature, the positive and negative target centers and the positive and negative target center standard sequences of each second - level index are determined, and the positive target center coefficient of the second - level index with respect to the positive target center and the negative target center coefficient with respect to the negative target center are obtained; the index nature includes cost - type and benefit - type.

[0022] Preferably, in step S5, the cumulative prospect theory is used to determine the index weights; the gain or loss situation of the secondary indexes is determined through the set reference target, and then different value functions and weight functions are selected. According to the prospect weight function, the comprehensive prospect value of the index set of the scheme set is calculated, the comprehensive prospect value of the primary index is obtained, and a comprehensive prospect value matrix is constructed.

[0023] Preferably, in step S5, fuzzy mathematics is used to quantify and optimize the reservoir operation scheme set; the Gaussian membership function is used to determine the fuzzy subsets of different comments for the comprehensive prospect value. According to the comprehensive prospect values of the indexes corresponding to different reservoirs, the fuzzy evaluation matrix of the reservoir operation scheme set belonging to the evaluation set is obtained. According to the correlation between the primary indexes, its weight vector is obtained. The fuzzy product operation of the M operator is performed on the fuzzy evaluation matrix and the weight vector, and the linear weighting is performed on the operation result to obtain the fuzzy evaluation result. The evaluation set is quantified to obtain the final score of the reservoir operation scheme; the value scores are sorted to obtain the ranking of the reservoir optimal operation scheme, and the reservoir operation scheme with the best comprehensive benefit, that is, the one with the top ranking, is obtained.

[0024] A reservoir multi-objective optimization system based on multi-party coupling includes: a model input module, an objective function module, a state variable module, a constraint condition module, a model solving module, and an evaluation and optimization module; where

[0025] The model input module is used to process the input long-term annual hydrological data and output typical annual hydrological data;

[0026] The objective function module is used to couple the flood control benefit target, power generation benefit target, and ecological benefit target to obtain the total objective function;

[0027] The constraint condition module determines the constraint conditions through the inflow and outflow discharge, initial water level in front of the dam, initial reservoir capacity at the set time period, and power station output, and then constrains the objective function of the reservoir multi-objective optimization system;

[0028] The model solving module uses the reference point-based non-dominated sorting genetic algorithm to solve the reservoir multi-objective optimization model and outputs the optimal reservoir operation scheme set of the reservoir multi-objective optimization system;

[0029] The evaluation and optimization module is used to optimize and evaluate multiple reservoir operation scheme sets.

[0030] The present invention adopts the above technical solutions and has the following beneficial effects:

[0031] Coupling the long-term annual runoff data with the hydrological model can improve the runoff forecast accuracy, provide accurate water regime information for reservoir operation, and formulate operation schemes based on reliable basic data, making the schemes more in line with the actual hydrological situation;

[0032] Through the multi-objective optimization model, the power generation scheduling model, flood control scheduling model and ecological scheduling model are coupled to reasonably determine the reservoir discharge process, ensure the flood control safety of the river channel and the ecological flow demand, achieve the balance between flood control and power generation benefits, meet the water use demands of life, industry and agriculture, etc., while ensuring the ecological water use of the downstream river channel, maintaining the stability of the ecosystem, optimizing the water supply plan, and realizing the coupling of water supply and ecology;

[0033] The standard function is used to verify the convergence of the model solved by the reference point-based non-dominated sorting genetic algorithm, which ensures the effectiveness and reliability of the algorithm and makes the obtained scheduling plan more reliable;

[0034] According to the primary evaluation indicators, an optimal evaluation index system for reservoir scheduling plans is constructed, providing a systematic and comprehensive basis for the evaluation of the plans; fully considering the influence of various evaluation indicators in screening the reservoir scheduling plan with the optimal comprehensive benefits, comprehensively evaluating the advantages and disadvantages of different plans, and realizing the maximization of the comprehensive benefits of the reservoir. Brief Description of the Drawings

[0035] Figure 1 It is a schematic diagram prepared for the method for optimizing and selecting the multi-objective optimization scheduling plan of the reservoir based on multi-party coupling of the present invention;

[0036] Figure 2 It is a logic diagram of the multi-objective optimization scheduling model of the reservoir based on multi-party coupling of the present invention;

[0037] Figure 3 It is the comprehensive benefit evaluation index system of the reservoir scheduling plan constructed in the present invention;

[0038] Figure 4 It is a flow chart of solving the multi-objective optimization scheduling model of the reservoir by using the reference point-based non-dominated sorting genetic algorithm in the present invention. Detailed Embodiment

[0039] The following is a detailed description of the present invention in conjunction with the drawings and embodiments:

[0040] As shown in the attach Figures 1 to 4 figure, the method for optimizing and selecting the multi-objective optimization scheduling plan of the reservoir based on multi-party coupling of the present invention includes the following steps:

[0041] S1: Taking the outflow discharge in each time period as the decision variable, constructing the total objective function of the reservoir with the economic benefit objective function, ecological benefit objective function and flood control benefit objective function, and taking the total objective function of the reservoir as the multi-objective optimization scheduling model of the reservoir; the constraint conditions of the multi-objective optimization scheduling model include water balance constraint, water level constraint, outflow discharge constraint, power station output constraint and non-negativity constraint.

[0042] In this embodiment, prototype observed hydrological and sediment data of a long sequence are collected. The long-sequence prototype observed hydrological and sediment data are used for water and sediment frequency analysis by means of a probability distribution curve to determine typical years of abundant, normal, and dry periods. Then, after classifying the incoming water and sediment into typical years of abundant, normal, and dry periods and combining with the flood hydrograph amplification technology, typical flood hydrographs with different frequency characteristics are obtained. Based on the ecological environment index data, the hydraulic connection between different reservoirs, and the different water conservancy tasks of the reservoirs, the boundary conditions of reservoir inflow and outflow and different objectives of the reservoirs are determined. The hydraulic connection between different reservoirs refers to the mutual influence of multiple reservoirs in terms of water flow and water volume. The different water conservancy tasks of the reservoirs refer to various functions assumed by the reservoirs to meet the needs of social and economic development, ecological environment protection, etc. Here, the water conservancy tasks include flood control, power generation of power stations, and ecological environment protection.

[0043] Among them, the probability distribution curve adopts the P-III type curve. According to the P-III type curve, a frequency curve of runoff is established, and the water volume entering the reservoir is classified into types of abundant, normal, and dry periods through the guarantee rate, so as to determine typical years of abundant, normal, and dry periods. The boundary conditions of reservoir inflow and outflow include inflow and outflow discharge, initial water level in front of the dam, initial reservoir capacity at a certain time period, and power generation of the power station. The different objectives of the reservoirs include economic benefit objective, ecological benefit objective, and flood control benefit objective.

[0044] In the present invention, M of the multi-objective optimal operation model of the reservoir 21 represents the economic benefit objective, that is, the maximum average power generation of the power station within the single-reservoir operation period; M 22 represents the ecological benefit objective, that is, the minimum average change degree of the suitable ecological flow in the downstream river channel of the single reservoir; M 23 represents the flood control benefit objective, that is, the maximum peak shaving rate within the single-reservoir operation period. The peak shaving rate within the single-reservoir operation period is proportional to the flood control benefit, and then the economic benefit objective M 21 , ecological benefit objective M 22 and flood control benefit objective M 23 are determined;

[0045] In this embodiment, the economic benefit objective M 21 is obtained through the following formula

[0046]

[0047] A = 9.81η(2)

[0048] Among them, T represents the total number of time periods in the operation period; A represents the comprehensive power generation coefficient of the power station of the reservoir; Q t out represents the average outflow discharge value of the reservoir at time period t (the time period t can be hours, days, dekads, or months), and the unit is m 3 / s; ΔH t represents the average water head of the reservoir at time period t, and the unit is m; η represents the efficiency coefficient of the water turbine generator of the reservoir;

[0049] Ecological benefit target M 22 is obtained through the following formula

[0050]

[0051] where Q t AEF represents the suitable ecological flow value of the downstream river course of the reservoir in the t period, with the unit of m 3 / s;

[0052] Flood control benefit target M 23 is obtained through the following formula

[0053]

[0054] where Q t in represents the average inflow value of the reservoir in the t period; M 23 ∈[0, 1];

[0055] Furthermore, the total target E2 of the reservoir is obtained and used as the multi-objective optimal operation model of the reservoir. The total target E2 of the reservoir is

[0056] E2 = F(M 21 , M 22 , M 23 )(5)

[0057] Construct a water balance model. Through the water balance model, a hydrological relationship curve is obtained, which is further coupled with the power generation model, the reservoir flood control operation model, and the ecological operation model to obtain the inflow and outflow discharges, the initial water level in front of the dam, the initial storage capacity at the beginning of a certain period, and the power station output. And through the above data, multiple constraint conditions of the multi-objective optimal operation model of the reservoir in step S1 are obtained, namely, the water balance constraint, the water level constraint in front of the dam, the reservoir outflow discharge constraint, and the power station output constraint.

[0058] In this embodiment, the water balance constraint condition is determined through the following formula

[0059] V t+1 - V t = (Q t in - Q t out )Δt (6)

[0060] where t represents the serial number of the time period; V t is the initial storage capacity of the reservoir in the t period, with the unit of 100 million m 3 ; V t+1 is the initial storage capacity of the reservoir water at the beginning of the t + 1 period, that is, the end storage capacity of the t period, with the unit of 100 million m 3; Δt represents the time interval from time period t to time period t+1;

[0061] The water level constraint condition is determined by the following formula

[0062] Z t min ≤Z t ≤Z t max (7)

[0063] where Z t represents the initial water level of the reservoir at time period t, with the unit of m; Z t max represents the upper limit value of the water level of the reservoir at time period t, with the unit of m; Z t min is the lower limit value of the water level of the reservoir at time period t, with the unit of m;

[0064] The reservoir discharge flow constraint condition is determined by the following formula

[0065] Q t min ≤Q t out ≤Q t max (8)

[0066] where Q t min represents the lower limit value of the reservoir discharge flow at time period t, with the unit of m 3 / s; Q t max represents the upper limit value of the reservoir discharge flow at time period t, with the unit of m 3 / s;

[0067] The power generation output constraint condition of the power station is determined by the following formula

[0068] N t min ≤N t ≤N t max (9)

[0069] where N t is the average power generation output of the power station of the reservoir at time period t, with the unit of kW; N t max is the upper limit value of the power generation output of the power station of the reservoir at time period t, with the unit of kW; N t min is the lower limit value of the power generation output of the power station of the reservoir at time period t, with the unit of kW;

[0070] Determine the non - negative constraint condition, that is, all kinds of variables are non - negative values.

[0071] S2: Solve the multi-objective optimal operation model of the reservoir by using the reference point-based non-dominated sorting genetic algorithm to obtain the Pareto front of each objective.

[0072] In the present invention, the key of the reference point-based non-dominated sorting algorithm lies in the selection mechanism, that is, individuals are selected based on the method associated with the reference line, and the next generation of parental population is generated based on environmental selection;

[0073] As shown in the appendix Figure 4 The reference point-based non-dominated sorting genetic algorithm includes the following steps:

[0074] S201: Generate reference points;

[0075] First, determine the number of reference points, secondly, define the basic conditions of the reference point set, then introduce the set of transition parameters, and finally generate the reference point set; that is:

[0076] First, in the objective space, a normalized hyperplane with a dimension of (M - 1) (M ≥ 2 and is a natural number) has the same inclination angle and the same axis distance for each objective axis. If a (M - 1)-dimensional hyperplane is divided p times along each objective, the number of generated reference points H is:

[0077]

[0078] where M ≥ 2 and is a natural number, p represents the number of divisions of each objective, p ≥ 1 and is a natural number;

[0079] Secondly, define the reference point set S = (s1, s2,..., s M ), then the coordinate values of the reference points are calculated as follows:

[0080]

[0081] where M represents the total number of objectives, j ∈ [1, M].

[0082] Then, define the set C belonging to the normalized hyperplane as a transition parameter for calculating the reference point coordinates, and

[0083]

[0084] Let c ij ∈ C, where i is the coding of the number of reference points; j is the coding of the number of objectives, and

[0085]

[0086] Finally, generate the reference point set S, s ij ∈ S and c ij ∈ C, s ijThe calculation method is as follows:

[0087]

[0088] S202: Standardize the objective space; First, calculate the minimum value z of each objective function in the population, j min , to form an ideal point set (z1 min , z2 min , …, z M min ), and subtract z from all objective values f j (x), then calculate the intercept a of each objective axis j min , and finally complete the standardization calculation through the ideal point and the intercept, j The calculation formula is:

[0089] where ω represents the reservoir number; f

[0090]

[0091] (x) is the standardized objective value of the normalized hyperplane reference point x on each objective axis, j ω x represents the coordinate of the reference point; f ′(x) is the objective value of the transformed population S j t ;

[0092] S203: Association operation; Construct the mapping relationship between the population individuals and the reference point. First, define a reference line in the objective space, that is, the line connecting the origin and the reference point, then calculate the perpendicular distance from each individual in the population to the reference line, and finally select the reference line closest to each individual and associate with it;

[0093] S204: Environmental selection; Perform non - dominated sorting on the population S t to obtain non - dominated levels (F1, F2, …, F L L ), set the next - generation parent population to be generated as P t+1 , and the population number is N; Calculate the number of associations between each reference point and the individuals in the population P t+1 , that is, the niche number denoted as ρ s ; Define a reference point set S min = {s: argmin s ρ s}, randomly select a reference point If there is an individual in F1 associated with this reference point then randomly select an individual from these individuals and store it in P t+1 , and the current reference point After the traversal is completed, the above operations are continued in the next non-dominated layer until the number of individuals in P t+1 reaches N, and the non-dominated solutions in the finally constructed population approximately form the Pareto front.

[0094] S3: Use standard functions to verify the effectiveness of the solution algorithm of the reservoir multi-objective optimization scheduling model in step S2;

[0095] In the present invention, in step S3, standard functions are also used for algorithm testing to verify the effectiveness of the algorithm in solving multi-objective optimization problems. The standard multi-objective test functions include DTLZ1 and DTLZ2 functions.

[0096] In this embodiment, DTLZ1 and DTLZ2 are respectively three-objective minimization test functions. The true Pareto front of the DTLZ1 function is a plane, and the true Pareto front of the DTLZ2 function is a convex surface. The expressions are as follows:

[0097]

[0098]

[0099] where x i represents the decision variable. x1 and x2 have a direct effect in the expressions of the objective functions f1(x), f2(x), and f3(x), while x3 to x 12 mainly indirectly affects the objective functions through the function g(x); f1(x), f2(x), and f3(x) are objective functions, and the goal is to minimize these three functions simultaneously; g(x) represents an auxiliary function that integrates the information of the decision variables x3 to x 12 and has an impact on the objective functions.

[0100] In this embodiment, the lengths of the decision variables of the DTLZ1 and DTLZ2 functions are set to 6 and 10 respectively; the population size popsize of the NSGA-Ⅲ algorithm and the NSGA-Ⅱ algorithm is set to 200, the number of generations gen is set to 500, the crossover distribution index is set to 20, and the mutation distribution index is set to 20.

[0101] The selected test function is a three-objective minimization function, and the effectiveness of the solution algorithm is evaluated through test indicators. The test indicators include the inverted generational distance, the spacing indicator, and the algorithm running time. The proximity and distribution uniformity of the solution set obtained by the algorithm are evaluated through the inverted generational distance in the test indicators. The smaller the value of the inverted generational distance, the closer the solution set obtained by the algorithm is to the true Pareto front, and the more uniform the distribution, which means the better the performance of the algorithm. The distribution uniformity of the non-dominated solution set obtained by the algorithm in the objective space is measured through the spacing indicator in the test indicators. If the value of the spacing indicator is small, it means that the solutions in the solution set are more evenly distributed in the objective space, and diverse solutions can be found in the entire feasible solution region. On the contrary, if the value of the spacing indicator is large, it means that the distribution of the solutions is uneven. The algorithm running time in the test indicators reflects the calculation efficiency of the algorithm. A short operation time indicates a high calculation efficiency of the algorithm.

[0102] S4: Construct an optimal evaluation index system for the reservoir operation scheme according to the first-level evaluation indicators. The first-level indicators include power generation indicators, ecological indicators, and flood control indicators.

[0103] In the present invention, in step S4, an optimal evaluation index system for the reservoir operation scheme is constructed according to the first-level indicators, where the first-level indicators include power generation indicators, ecological indicators, and flood control indicators; the first-level indicators respectively correspond to secondary indicators. The secondary indicator corresponding to the power generation indicator is the average output of the reservoir power station, the secondary indicator corresponding to the ecological indicator is the average change degree of the downstream ecological flow of the reservoir, and the secondary indicator corresponding to the flood control indicator is the maximum peak shaving rate of the reservoir.

[0104] The power generation indicator consists of the average output index of the power station during the scheduling period corresponding to each reservoir in the reservoir operation scheme. The higher the average output of the power station, the higher the beneficial utilization benefit of the corresponding reservoir. The average output index of the power station is defined as a benefit-type index;

[0105] The ecological indicator consists of the average change degree index of the downstream ecological flow during the scheduling period corresponding to each reservoir in the reservoir operation scheme. The greater the average change degree of the ecological flow, the smaller the ecological benefit of the corresponding reservoir. The ecological flow change degree index is defined as a cost-type index;

[0106] The flood control indicator consists of the maximum peak shaving rate index during the scheduling period corresponding to each reservoir in the reservoir operation scheme. The greater the maximum peak shaving rate, the greater the flood control benefit of the corresponding reservoir. The maximum peak shaving rate index is defined as a benefit-type index. For the cascade reservoir operation scheme, the maximum peak shaving rate during the scheduling period can be calculated according to the inflow and outflow of each reservoir.

[0107] S5: Determine the decision matrix of the reservoir operation plan set according to the grey target theory, determine the evaluation index weights through the cumulative prospect theory, obtain the membership degrees of each reservoir operation plan under each evaluation level, and finally screen out the reservoir operation plan with the optimal comprehensive benefit according to the maximum membership degree principle.

[0108] In the present invention, in the step S5, the reservoir operation plan is evaluated and optimized by a method combining the grey target theory, the cumulative prospect theory and the fuzzy mathematics method. First, the decision matrix of the reservoir operation plan set is determined by the grey target theory. Secondly, the index weights are determined by the cumulative prospect theory. Finally, the reservoir operation plan set is quantified and optimized by the fuzzy mathematics.

[0109] In this embodiment, the decision matrix of the reservoir operation plan set is determined by the grey target theory, and the key data indexes reflecting the advantages and disadvantages of the system or the reservoir operation plan are used to form the corresponding data patterns. The key data indexes include the positive and negative target centers, the positive and negative target center standard sequences and the positive and negative target center coefficients. The positive and negative target centers and the positive and negative target center standard sequences of the indexes are determined according to the index nature; the positive target center coefficient or the negative target center coefficient is obtained by comparing each index of the reservoir with the positive target center standard sequence or the negative target center standard sequence, so as to realize the quantification of the indexes from two directions; the index nature includes cost type and benefit type.

[0110] When there are n reservoir operation plans in the multi-index evaluation system, the reservoir operation plan set S = {s1, s2,..., s n}, and each reservoir operation plan has m evaluation indexes to form an index set O = {O1, O2,..., O m}. The effect sample value x i of the reservoir operation plan S j for the index O ij (i = 1, 2,..., n, j = 1, 2,..., m), then the decision matrix (effect sample matrix) X of the reservoir operation plan set S for the index set O is:

[0111]

[0112] Let

[0113]

[0114] where z j represents the mean value of the jth index;

[0115] For the distinction of the following indexes, the sample value x ij is normalized to obtain y ij . If the corresponding index is of the benefit type, then there is:

[0116]

[0117] If the corresponding index is of the cost type, then we have:

[0118]

[0119] The transformed matrix is denoted as:

[0120] D = (y ij ) n*m (22)

[0121] Normalize the matrix D to obtain the normalized decision matrix R, and the expression is:

[0122] R = (r ij ) n*m (23)

[0123] Perform normalization on the y ij obtained after normalization processing to get r ij , and its normalization method is:

[0124]

[0125] The above transformation restricts the element values of the decision matrix to the interval [-1, 1];

[0126] Determine the positive and negative bull's-eyes and the positive and negative bull's-eye sequences of each secondary index in the decision matrix, and its expression is:

[0127]

[0128] According to the multi-objective grey target theory, calculate the positive and negative correlation coefficients of each secondary index with the positive and negative bull's-eyes;

[0129] r j + and r j - are the positive bull's-eye and the negative bull's-eye respectively. The positive bull's-eye coefficient of the jth evaluation index of the ith reservoir operation scheme with the positive bull's-eye and the negative bull's-eye coefficient with the negative bull's-eye are obtained through the following formula:

[0130]

[0131] Among them, ρ represents the discrimination coefficient, ρ ∈ [0, 1], generally taking ρ = 0.5. The value of the positive bull's-eye coefficient is directly proportional to the size of the actual index value, the value of the negative bull's-eye coefficient is inversely proportional to the size of the actual index value, and the positive bull's-eye coefficient and the negative bull's-eye coefficient are restricted to the interval (0, 1]; r ij represents the sample value after normalization processing.

[0132] In this embodiment, the cumulative prospect theory is used to determine the evaluation index weights. The magnitude of the prospect value is jointly determined by the value function and the decision weight. The cumulative prospect theory determines the gain or loss situation of the data index by setting a reference target, and then selects different value functions and weight functions; the expression of the prospect value is:

[0133]

[0134] where V is the prospect value; n represents the total number of reservoir operation schemes; π(p) is the decision weight; and v(x) is the value function.

[0135] The expression of the prospect value function corresponding to each index is:

[0136]

[0137] where r ij represents the sample value after normalization; ξ ij + and ξ ij - represent the positive and negative bull's-eye coefficients respectively; the parameters α and β are the concavity and convexity degrees of the value power functions in the gain and loss regions respectively, and α, β < 1 indicates decreasing sensitivity; θ represents the attitude towards loss; generally, α = β = 0.88 and θ = 2.25 are taken.

[0138] Then the prospect weight functions when the index value system faces gain indexes and loss indexes are π + (w ij ) and π - (w ij ) respectively, as follows:

[0139]

[0140] where w j is the weight of each index; w j r+ and w j r- represent the prospect weight functions when the index value system faces gain and loss indexes respectively; the parameter r+ = 0.61 for measuring the beneficial part and the parameter r- = 0.69 for measuring the loss part in the prospect weight function.

[0141] Then the comprehensive prospect value of the index set O of the reservoir operation scheme S is:

[0142]

[0143] where V i is the comprehensive prospect value of the first-level index; v + (r ij) represents the value function when facing the revenue index; v - (r ij ) represents the value function when facing the loss index; calculate the comprehensive prospect value of each first-level index, realize the quantification of the first-level index, and construct the comprehensive prospect value matrix.

[0144] In this embodiment, the evaluation method based on fuzzy mathematics uses a Gaussian membership function to determine the comprehensive prospect value V ik For the fuzzy subsets of different comments, its form is as follows:

[0145]

[0146] Among them, V i represents that the input variable is the comprehensive prospect value, δ represents the width parameter in the Gaussian membership function, which determines the width of the membership function curve, and c represents the center parameter in the Gaussian membership function, which is used to determine the curve center of the Gaussian function; different comments correspond to different width parameters and center parameters.

[0147] Substitute the comprehensive prospect value V ik of the kth index of the ith reservoir into the following formula to obtain the fuzzy evaluation matrix F i of the reservoir operation scheme i belonging to the evaluation set A

[0148]

[0149] Among them, is the membership degree of the comprehensive prospect value V ik in different evaluation levels A t (t = 1, 2,..., 5).

[0150] According to the correlation between the first-level indicators, take its weight vector as λ = λ1, λ2,..., λ k , perform the M-operator fuzzy product operation on the fuzzy evaluation matrix F i and the weight vector λ, and perform linear weighting on the operation result to obtain the fuzzy evaluation result S i , as follows:

[0151]

[0152] Among them, S i (A t ) is the membership degree of the reservoir operation scheme i relative to the evaluation level A t , which represents the degree to which the reservoir operation scheme i can be described by A t , t = 1, 2,..., 5; the membership degree of each reservoir operation scheme under each evaluation level can be determined, and according to the maximum membership degree principle, the reservoir operation scheme with the optimal index can be selected.

[0153] Quantify the evaluation set A, i.e.:

[0154] A = [A1 A2 A3 A4 A5] = [95 85 75 65 55] (37)

[0155] The index value evaluation score of the reservoir operation scheme i is:

[0156]

[0157] where G i is the final score of the index value of each reservoir operation scheme. Sorting the value scores gives the ranking of the optimized reservoir operation schemes, which to a certain extent reflects the advantages and disadvantages of the comprehensive benefits of each reservoir operation scheme. Finally, a reservoir operation scheme is selected as the optimal operation scheme.

[0158] The multi-objective optimization system for reservoirs based on multi-party coupling described in the present invention includes: a model input module, an objective function module, a state variable module, a constraint condition module, a model solution module, and an evaluation and optimization module; where

[0159] The model input module is used to process the input long-term annual hydrological data and output typical annual hydrological data;

[0160] The objective function module is used to couple the flood control benefit objective, power generation benefit objective, and ecological benefit objective to obtain the total objective function, which is used as the multi-objective optimization operation model for the reservoir;

[0161] The constraint condition module is used to constrain the objective function of the multi-objective optimization operation system for the reservoir through the inflow and outflow discharge, initial water level before the dam, initial reservoir storage at a certain time period, and power station output;

[0162] The model solution module uses the reference point-based non-dominated sorting genetic algorithm to solve the multi-objective optimization operation model for the reservoir and outputs the set of optimal reservoir operation schemes for the multi-objective optimization operation system for the reservoir;

[0163] The evaluation and optimization module is used to optimize and evaluate multiple sets of reservoir operation schemes, determine the decision matrix of the reservoir operation scheme set using the grey target theory, determine the index weights using the cumulative prospect theory, and quantify and sort the reservoir operation scheme set using fuzzy mathematics, and finally select a reservoir operation scheme as the optimal operation scheme.

Claims

1. A method for optimizing the selection of multi-objective optimal operation schemes of reservoirs based on multi-party coupling, characterized in that: It includes the following steps: S1: Taking the out - flow rate in each time period as the decision variable, constructing the total objective function of the reservoir with the economic benefit, ecological benefit and flood control benefit objective functions, and using it as the multi - objective optimal operation model of the reservoir; the constraint conditions of the multi - objective optimal operation model include water balance constraint, water level constraint, out - flow rate constraint, power generation output constraint and non - negative constraint; S2: Using the non - dominated sorting genetic algorithm with reference points to solve the multi - objective optimal operation model of the reservoir, and obtaining the Pareto front of each objective; S3: Using the standard function to verify the effectiveness of the solution algorithm of the multi - objective optimal operation model of the reservoir in step S2; S4: Constructing an evaluation index system for the optimal selection of reservoir operation plans according to the first - level evaluation indexes, and the first - level indexes include power generation index, ecological index and flood control index; S5: Determining the decision matrix of the reservoir operation plan set according to the grey target theory, determining the evaluation index weights through the cumulative prospect theory, obtaining the membership degrees of each reservoir operation plan under each evaluation level, and finally screening out the reservoir operation plan with the optimal comprehensive benefit according to the maximum membership degree principle.

2. The method for optimizing and selecting the multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 1, characterized in that: In step S1, the objectives of the reservoir include the maximum average power generation output of the power station within the single - reservoir operation period, the minimum average change degree of the suitable ecological flow rate in the downstream river of the single - reservoir, and the maximum peak - cutting rate within the single - reservoir operation period, which respectively correspond to representing the economic benefit objective, ecological benefit objective and flood control benefit objective; Determining the economic benefit objective according to the decision variable, the average water head of the reservoir in the set time period and the comprehensive power generation output coefficient of the reservoir power station; determining the ecological benefit objective according to the suitable ecological flow rate of the downstream river of the reservoir in the set time period and the decision variable; Determining the flood control benefit objective according to the decision variable and the average inflow rate of the reservoir in the set time period.

3. The method for optimizing and selecting the multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 2, characterized in that: The constraint conditions of the multi - objective optimal operation model described in step S1 include water balance constraint condition, water level constraint condition, out - flow rate constraint condition, power generation output constraint condition and non - negative constraint condition.

4. The method for optimizing and selecting the multi-objective optimal operation plan of a reservoir based on multi-party coupling according to claim 3, characterized in that: The water balance constraint condition is that the change in reservoir storage volume within a time period is equal to the net difference between the average inflow and outflow rates of the reservoir within that time period multiplied by the time period duration; The water level constraint condition is that the initial water level of the reservoir in the set time period is not greater than the upper limit value of the water level of the reservoir in the set time period and not less than the lower limit value of the water level of the reservoir in the set time period; The out - flow rate constraint condition is that the average out - flow rate value of the reservoir in the set time period is not greater than the upper limit value of the out - flow rate of the reservoir in the set time period and not less than the lower limit value of the out - flow rate of the reservoir in the set time period; The power generation output constraint condition is that the average power generation output of the reservoir in the set time period is not greater than the upper limit value of the power generation output of the reservoir in the set time period and not less than the lower limit value of the power generation output of the reservoir in the set time period; The non - negative constraint condition is that all variables in the water balance constraint condition, water level constraint condition, out - flow rate constraint condition and power generation output constraint condition are non - negative values.

5. The method for optimizing and selecting the multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 1, wherein: The standard multi-objective test function selected in the step S3 adopts a three-objective minimization test function of and function, and the effectiveness of the solution algorithm is evaluated through test indicators, and the test indicators include: inversion generational distance, spacing index, and algorithm running time.

6. The method for optimizing and selecting a multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 1, wherein: The evaluation indicators in step S4 include first-level indicators: power generation indicators, ecological indicators, and flood control indicators; the secondary indicators of the power generation indicators are composed of the average power generation indicators of power stations corresponding to each reservoir during the corresponding scheduling period in the reservoir scheduling plan, the secondary indicators of the ecological indicators are composed of the average change degree indicators of downstream ecological flows corresponding to each reservoir during the corresponding scheduling period, and the secondary indicators of the flood control indicators are composed of the maximum peak shaving rate indicators corresponding to each reservoir during the corresponding scheduling period. The average power generation indicator and the maximum peak shaving rate indicator in the secondary indicators are defined as profit-type indicators, and the ecological flow change degree indicator is defined as a cost-type indicator.

7. The method for optimizing and selecting the multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 6, wherein: In step S5, the decision matrix of the reservoir scheduling plan set is determined using the grey target theory; first, the decision matrix of the plan set with respect to the indicator set is determined, and the transformed decision matrix is converted into a normalized decision matrix through decision matrix normalization. According to the nature of the indicators, the positive and negative target centers and the positive and negative target center standard sequences of each secondary indicator are determined, and the positive target center coefficient of the secondary indicator with respect to the positive target center and the negative target center coefficient with respect to the negative target center are obtained; the said nature of the indicators includes cost-type and profit-type.

8. The method for optimizing and selecting the multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 7, wherein: In step S5, the cumulative prospect theory is used to determine the indicator weights; the profit or loss situation of the secondary indicators is determined through the set reference target, and then different value functions and weight functions are selected. According to the prospect weight function, the comprehensive prospect value of the indicator set of the plan set is calculated, the comprehensive prospect value of the first-level indicator is obtained, and a comprehensive prospect value matrix is constructed.

9. The method for optimizing and selecting the multi-objective optimal operation scheme of a reservoir based on multi-party coupling according to claim 8, wherein: In step S5, fuzzy mathematics is used to quantify and optimize the reservoir scheduling plan set. Use the Gaussian membership function to determine the fuzzy subsets of different comments for the comprehensive prospect value. Obtain the fuzzy evaluation matrix of the reservoir operation plan set belonging to the evaluation set according to the comprehensive prospect values of the indicators corresponding to different reservoirs. Obtain its weight vector according to the correlation between the first-level indicators, and perform the operator fuzzy product operation on the fuzzy evaluation matrix and the weight vector. After linearly weighting the operation results, obtain the fuzzy evaluation result. Quantify the evaluation set to obtain the final score of the reservoir operation plan; sort the value scores to obtain the ranking of the reservoir optimal operation plan, and obtain the reservoir operation plan with the best comprehensive benefit, that is, the one with the top ranking.

10. A reservoir multi-objective optimization system based on multi-party coupling, characterized in that A method for optimizing and selecting a multi-objective reservoir scheduling plan based on multi-party coupling according to any one of claims 1 to 8, comprising: a model input module, a target function module, a state variable module, a constraint condition module, a model solving module, and an evaluation and optimization module; wherein, The model input module is used to process the input long-term annual hydrological data and output typical annual hydrological data. The target function module is used to couple the flood control benefit target, the power generation benefit target, and the ecological benefit target to obtain a total target function. The constraint condition module determines the constraint conditions through the inflow and outflow discharges, the initial water level in front of the dam, the initial reservoir capacity at the set time period, and the power generation of the power station, and then constrains the target function of the multi-objective reservoir scheduling system. The model solving module uses a reference point-based non-dominated sorting genetic algorithm to solve the multi-objective reservoir scheduling model and outputs the optimal reservoir scheduling plan set of the multi-objective reservoir scheduling system. The evaluation and optimization module is used to optimize and evaluate multiple reservoir scheduling plan sets.

Images