Reservoir group multi-objective optimization scheduling method coupled with ISMAA-VIKOR decision model

Abstract

The invention discloses a reservoir group multi-objective optimization scheduling method coupled with an ISMAA-VIKOR decision model, and belongs to the technical field of reservoir scheduling, and the method comprises the steps: obtaining basic data and parameters of reservoir group optimization scheduling operation, and carrying out the reasonable generalization of a reservoir group system; a specific scheduling task of the reservoir group is analyzed, an objective function and constraint conditions of an optimization scheduling model are determined, and a multi-objective optimization scheduling model is constructed; an NSGA-III algorithm is adopted to solve the multi-target optimization scheduling model, and a reservoir group multi-target optimization scheduling scheme set is obtained; and establishing an evaluation index system for optimizing a scheduling scheme, and determining an optimal scheduling operation scheme of the reservoir group in combination with an ISMAA-VIKOR multi-attribute decision model. By the adoption of the method, global optimization of the multi-target operation scheme of the reservoir group can be achieved, and the actual requirement of a decision maker for reasonably and efficiently selecting the optimal scheduling scheme of the reservoir group is effectively met.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir operation, and in particular to a multi-objective optimal operation method for reservoir groups coupling an ISMAA-VIKOR decision-making model. Background Art

[0002] With the deepening of the concept of integrated river basin management, multi-objective collaborative operation of reservoirs has become a key technical means to balance multiple objectives such as water supply guarantee, flood control safety, ecological maintenance, and power generation benefits. Existing technologies mainly focus on constructing an optimal operation plan set with Pareto frontier characteristics, and then screening the optimal solution set of comprehensive benefits through multi-attribute decision-making theory. The traditional multi-attribute decision-making framework includes three core elements: (1) an evaluation index system, that is, a standardized measurement system composed of mutually exclusive and complete decision criteria; (2) a preference information representation system, mainly reflected in the quantitative expression of the criterion weight vector; (3) a decision aggregation operator, which converts the multi-dimensional criterion value range into a single-dimensional comprehensive evaluation value through mathematical mapping. This decision-making process is essentially a value judgment process based on the cognitive structure of the decision-making subject, and its technical route usually follows the standardized process of "scheme generation - criterion quantification - weight assignment - comprehensive ranking". However, the existing method system has two significant defects: First, the criterion performance parameters and weight coefficients are fixed as deterministic values, ignoring the parameter uncertainty commonly existing in the actual decision-making environment; Second, a linear additive utility function is used for value aggregation, making it difficult to effectively handle the non-linear compensation effect between incommensurable criteria.

[0003] As a typical paradigm of inverse weight space analysis, the Stochastic Multi-criteria Acceptability Analysis (SMAA-2) method realizes the global sensitivity evaluation of schemes under uncertain environments by introducing the criterion value probability distribution and weight space uniform sampling technology. Its core algorithm calculates key indicators such as the acceptability index, central weight vector, and confidence factor of each scheme through Monte Carlo simulation. However, this method still has significant technical bottlenecks in engineering practice: (1) When using a linear additive utility function for value aggregation, no standardization preprocessing is performed on incommensurable criteria, resulting in a systematic deviation of high-dimensional criteria from the comprehensive evaluation value; (2) The weight space uses a uniform distribution to simulate parameter uncertainty. Although it can ensure ergodicity, it cannot effectively integrate prior preference information, resulting in weakened decision-making orientation; (3) It lacks a compatibility design for non-linear utility functions and is difficult to characterize the diminishing marginal utility characteristics commonly existing in actual decision-making. Summary of the Invention

[0004] The purpose of the present invention is to provide a multi-objective optimal operation method for reservoir groups coupling an ISMAA-VIKOR decision-making model, which significantly improves the analytical ability of the decision-making system for the multi-criteria interaction under complex constraint conditions of the reservoir multi-objective optimal operation problem through the innovation of the weight allocation mechanism and the reconstruction of the non-linear utility function.

[0005] To achieve the above object, the present invention provides a multi-objective optimal operation method for a reservoir group coupled with an ISMAA-VIKOR decision model, and the steps include:

[0006] S1. Obtain the basic operation data and relevant parameters of the reservoir group, and conduct systematic generalization based on the topological structure of the reservoir group;

[0007] S2. Analyze the specific operation tasks of the reservoir group, and construct a multi-objective optimal operation model for the reservoir group according to the basic operation principles of the reservoir group;

[0008] S3. Use the NSGA-Ⅲ algorithm to solve the model and obtain the Pareto non-dominated solution set of the multi-objective optimal operation model of the reservoir group;

[0009] S4. Establish an evaluation index system for the optimal operation plan, and combine it with the ISMAA-VIKOR multi-attribute decision model to determine the optimal operation plan of the reservoir group.

[0010] Preferably, the basic operation data and relevant parameters of the reservoir group include: the inflow runoff data of each reservoir in the reservoir group, the water level-storage capacity relationship curve of the reservoir, the characteristic parameters of the water conservancy project facilities, and the water use demand of the water receiving area. The characteristic parameters of the water conservancy project facilities include the water conveyance capacity limit of the river channel, the operation parameters of the pumps and gates.

[0011] Preferably, the multi-objective optimal operation model of the reservoir group takes the minimum of the water shortage rate, pumping volume, river pumping volume, water abandonment rate, and water supply uniformity coefficient as the objective function, and the constraint conditions include the reservoir regulation capacity, pumping station water transfer capacity, gate flow capacity, river channel water conveyance capacity, and water balance. The expression is:

[0012] minF(x) = {f 1 , f 2 , f 3 , f 4 , f 5};

[0013]

[0014] In the formula, F(x) represents the set of objective functions; f 1 , f 2 , f 3 , f 4 , and f 5 respectively represent the water shortage rate, pumping volume, river pumping volume, water abandonment rate, and water supply uniformity coefficient; s.t represents the water balance, reservoir regulation capacity, pumping station water transfer capacity, control gate station flow capacity, and river channel water conveyance capacity constraints; RV t+1 represents the reservoir capacity at the (t + 1)-th period; RVt represents the reservoir storage volume at the t-th time period; I t represents the inflow into the reservoir at the t-th time period; D t represents the outflow from the reservoir at the t-th time period; Δt is the calculation time period length; RV t,min and RV t,max respectively represent the minimum and maximum reservoir storage volumes at the t-th time period of the reservoir; represents the pumping volume of the water pump at the t-th time period; represents the maximum pumping volume of the water pump at the t-th time period; represents the water discharge volume of the sluice at the t-th time period; represents the maximum water discharge volume of the sluice at the t-th time period; represents the water conveyance volume of the river channel at the t-th time period; represents the maximum water conveyance capacity of the river channel.

[0015] Preferably, step S3 specifically includes:

[0016] S31. Input the basic operation data and related parameters required for the operation of the multi-objective optimal operation model of the reservoir group, and accordingly set the reservoir regulation capacity, the water transfer capacity of the pumping station, the flow capacity of the gate, the water conveyance capacity of the river channel, and the water balance constraint conditions, and give the feasible domain intervals of the reservoir water level, the water extraction volume of the pumping station, the water discharge volume of the gate, and the water conveyance volume of the river channel, and set the number of schemes S and the number of iterations Iter max ;

[0017] S32. Initialize the pumping volume of the pumping station and the water discharge volume of the gate;

[0018] S33. Simulate the operation of the reservoir group hour by hour, and calculate the water level at the end of each time period of each reservoir based on the water balance equation, which is used as the initial water level of the reservoir for the next time period;

[0019] S34. After the simulation of the complete scheduling time period is completed, count the water extraction volume of each pumping station and the actual water supply volume of each water receiving area, and calculate the total pumping volume and the total water shortage rate of the inter-basin water transfer project, and then go to step S35;

[0020] S35. Use the NSGA-Ⅲ algorithm to update the pumping volume of the pumping station and the water discharge volume of the gate in the next stage, and return to step S33 until the number of iterations reaches the preset maximum number of iterations Iter max , and obtain the Pareto non-dominated solution set of the multi-objective optimal operation model of the reservoir group.

[0021] Preferably, step S4 specifically includes:

[0022] S41. Let the candidate solution set be A = {A 1 , A 2 , …, A m}, where Am Denote the \(m\)-th scheme, where \(m = 1, 2, \ldots, M\), and each scheme \(A\) m has \(J\) evaluation attributes \(C=\{C\) 1 , C\) 2 , \ldots, C\) J \}\), and construct a decision matrix \(X = [x\) ij \]_{M\times J}\);

[0023] S42. Use \(K\) subjective and objective weighting methods for initial weighting, including the Analytic Hierarchy Process, the CRITIC method, and the Entropy method. Each method calculates an initial weight vector:

[0024]

[0025] where represents the weight value of the criterion \(C\) j calculated by the weighting method \(k\);

[0026] S43. Use a game theory optimization model to determine the final comprehensive weight. Establish an optimization game framework for multi-method weighting. Assume that each weighting method is a game participant, and its goal is to minimize the deviation between its own weight and the final comprehensive weight. The comprehensive weight \(\omega^*\) is obtained by solving the optimization problem expression:

[0027]

[0028] In the formula, \(\omega\) (k) is the weight vector calculated by the weighting method \(k\); \(\alpha\) k is the weight balance factor of the \(k\)-th weighting method, and the optimal \(\omega^*\) is obtained by solving quadratic programming or convex optimization methods;

[0029] S44. Use \(\omega^*\) as the mean of the prior distribution for random weight sampling to ensure compliance with the existing preference information;

[0030] Random weight vector:

[0031]

[0032] where \(\omega\) j (n) represents the weight value of the criterion \(C\) j in the \(n\)-th group of random weights.

[0033] Use the Dirichlet distribution for random sampling with \(\omega^*\) as the center:

[0034] \(\omega\) (n) \sim Dirichlet(\beta\omega\) * );

[0035] where \(\beta\) is an adjustment parameter that controls the randomness range;

[0036] S45, for each group of random weights ω (n) , use the VIKOR method to calculate the utility value of each candidate solution, and rank the candidate solutions according to the utility value;

[0037] S46. Through Monte Carlo simulation, the frequency of candidate solutions appearing in different rankings is counted, and the ranking acceptability index is calculated to evaluate the stability of the solution; the ranking acceptability index b is calculated based on the utility value m,p :

[0038]

[0039] In the formula, Refers to Plan A m In the nth group of random weights ω (n) Is it ranked p?

[0040] S47, calculate the central weight vector and confidence factor, conduct in-depth analysis on the stability of the scheme, and evaluate its robustness under different weight distributions;

[0041] Center weight vector:

[0042]

[0043] In the formula, Indicates A m In the nth group of random weights ω (n) Is it ranked first?

[0044] Confidence Factor:

[0045]

[0046] Among them, O m The smaller, the better. m It is not sensitive to weight changes and has high stability. m The larger the m The ranking is easily affected by weight changes and has low stability;

[0047] S48. According to the needs of decision makers, select the solution with the highest overall acceptability index as the final recommendation, or select the solution with the highest confidence factor O. m The smallest solution is the most robust solution.

[0048] Preferably, step S45 specifically includes:

[0049] S451. Calculate the positive ideal value of each indicator based on the standardized matrix and negative ideal value

[0050] S452, calculate group utility value Um (n) and the individual regret value V m (n) , U m (n) also represents the comprehensive distance between Solution A m and the ideal solution under the random weight vector ω in the nth group, V (n) m (n) also represents the distance of Solution A m from the single most unfavorable index under the random weight vector ω in the nth group: (n)

[0051]

[0052] S453. Based on U m (n) and V m (n) , calculate the VIKOR comprehensive score Q of Solution A m under the random weight vector ω in the nth group: (n) m (n) :

[0053]

[0054] where U + and U - represent the minimum and maximum utility values among all candidate solutions respectively; V + and V - represent the minimum and maximum regret values among all candidate solutions respectively;

[0055] S454. Rank the candidate solutions according to the utility values.

[0056] Therefore, the reservoir group multi-objective optimal scheduling method adopting the above-mentioned coupled ISMAA-VIKOR decision-making model has the following beneficial effects:

[0057] (1) By organically combining comprehensive weighting and multi-attribute decision-making, the requirements of multi-objective joint optimal scheduling of the reservoir group are realized;

[0058] (2) Introduce the game theory method, integrate the analytic hierarchy process, entropy weight method and CRITIC method, combine subjective weighting and objective weighting, optimize the weight allocation, reflect the decision-maker's preference and capture the internal characteristics of the data, and improve the scientificity and rationality of the weight allocation;

[0059] (3) Compress the uncertain range of the weight space through the composite weighting mechanism, reduce the decision preference representation error, and significantly improve the robustness of the solution ranking;

[0060] ​​​(4) Combine the advantages of SMAA-2 and VIKOR to construct a unified framework, improve the multi-objective comprehensive benefits of the scheduling scheme in the optimization model, and significantly enhance the independent decision-making ability of the reservoir group scheduling system;

[0061] (5) It is applicable to the multi-objective optimization problems of complex water resource systems, with strong model generality, provides a scientific decision-making basis for water resource management, and has wide application value.

[0062] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Description of the Drawings

[0063] Figure 1 is the flowchart of the method of the embodiment of the present invention;

[0064] Figure 2 is the generalized diagram of the reservoir group system of the embodiment of the present invention;

[0065] Figure 3 is the schematic diagram of the spatial distribution of the pumping water volume of the reservoir group of the embodiment of the present invention. Detailed Embodiment

[0066] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Usually, the components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0067] Embodiment

[0068] As Figure 1 shown, the present invention provides a multi-objective optimal scheduling method for a reservoir group coupling an ISMAA-VIKOR decision model, and the steps include:

[0069] S1. Obtain the basic operation data and related parameters of the reservoir group, and generalize the system based on the topological structure of the reservoir group. The basic operation data and related parameters of the reservoir group include: the inflow runoff data of each reservoir in the reservoir group, the water level-storage capacity relationship curve of the reservoir, the characteristic parameters of water conservancy engineering facilities (the water conveyance capacity limit of the river channel, the operation parameters of pumps and gates), and the water use demand of the water receiving area.

[0070] S2. Analyze the specific scheduling operation tasks of the reservoir group, and construct a multi-objective optimal scheduling model for the reservoir group according to the basic scheduling principles of the reservoir group. The multi-objective optimal scheduling model for the reservoir group takes the minimization of water shortage rate, pumping volume, river pumping volume, water abandonment rate, and water supply uniformity coefficient as the objective function, and the constraint conditions include reservoir storage capacity, pumping station water transfer capacity, gate flow capacity, river water conveyance capacity, and water balance. The expression is as follows:

[0071] minF(x) = {f 1 , f 2 , f 3 , f 4 , f 5};

[0072]

[0073] In the formula, F(x) represents the set of objective functions; f 1 , f 2 , f 3 , f 4 , and f 5 represent the water shortage rate, pumping volume, river pumping volume, water abandonment rate, and water supply uniformity coefficient respectively; s.t represents the constraints of water balance, reservoir storage capacity, pumping station water transfer capacity, control gate station flow capacity, and river water conveyance capacity; RV t+1 represents the reservoir storage at the (t + 1)-th time period; RV t represents the reservoir storage at the t-th time period; I t represents the inflow of the reservoir at the t-th time period; D t represents the outflow of the reservoir at the t-th time period; Δt is the calculation time period length; RV t,min and RV t,max represent the minimum and maximum reservoir storages at the t-th time period respectively; represents the pumping volume of the water pump at the t-th time period; represents the maximum pumping volume of the water pump at the t-th time period; represents the water discharge of the sluice at the t-th time period; represents the maximum water discharge of the sluice at the t-th time period; represents the water conveyance volume of the river at the t-th time period; represents the maximum water conveyance capacity of the river.

[0074] S3. Use the NSGA-Ⅲ algorithm to solve the model and obtain the Pareto non-dominated solution set of the multi-objective optimal scheduling model for the reservoir group.

[0075] The steps include:

[0076] S31. Input the basic operation data and related parameters required for the operation of the multi-objective optimal operation model of the reservoir group. Accordingly, set the reservoir storage capacity, pumping station water transfer capacity, gate flow capacity, river water conveyance capacity and water balance constraint conditions, and specify the feasible range intervals of reservoir water levels, pumping station water extraction volumes, gate discharge volumes and river water conveyance volumes. Set the number of schemes S and the number of iterations Iter. max .

[0077] S32. Initialize the water extraction volume of the pumping station and the discharge volume of the gate. The water extraction volume and discharge volume should meet the requirements of the feasible range, and the actual water supply rate in the water receiving area shall not exceed its maximum water demand.

[0078] S33. Under the requirements of meeting the actual control constraints of the model, simulate the operation of the reservoir group hour by hour. Calculate the water level at the end of each time period of each reservoir based on the water balance equation, which serves as the initial water level of the reservoir for the next time period.

[0079] S34. After the simulation of the complete scheduling period is over, count the water extraction volumes of each pumping station and the actual water supply volumes of each water receiving area, and calculate the total water extraction volume and total water shortage rate of the inter-basin water transfer project, and then go to step S35.

[0080] S35. Use the NSGA-Ⅲ algorithm to update the water extraction volume of the pumping station and the discharge volume of the gate in the next stage, and return to step S33 until the number of iterations reaches the preset maximum number of iterations Iter. max , and obtain the Pareto non-dominated solution set of the multi-objective optimal operation model of the reservoir group.

[0081] S4. Establish an evaluation index system for the optimal operation plan, and combine it with the ISMAA-VIKOR multi-attribute decision-making model to determine the optimal operation plan of the reservoir group. It includes:

[0082] S41. Let the candidate plan set be A = {A 1 , A 2 , …, A m}, where A m represents the m-th plan, m = 1, 2, …, M. Each plan A m has J evaluation attributes C = {C 1 , C 2 , …, C J}. Construct the decision matrix X = [x ij M×J;

[0083] S42. Use K subjective and objective weighting methods for initial weighting, including the analytic hierarchy process, CRITIC method and entropy weight method. Each method calculates the initial weight vector:

[0084]

[0085] Among them, represents the weight value of standard C calculated by the weighting method k. j of the weight.

[0086] S43. To reasonably integrate the results of different weighting methods, a game theory optimization model is used to determine the final comprehensive weight. By establishing an optimization game framework for multi-method weighting, assuming that each weighting method is a game participant, whose goal is to minimize the deviation between its own weight and the final comprehensive weight, the comprehensive weight ω* is determined by solving the following optimization problem:

[0087]

[0088] In the formula, ω (k) is the weight vector calculated by the weighting method k; α k is the weight balance factor of the k-th weighting method; the optimal ω* is obtained by solving quadratic programming or convex optimization methods.

[0089] S44. Use ω* as the mean of the prior distribution for random weight sampling to ensure compliance with the existing preference information;

[0090] Random weight vector:

[0091]

[0092] Among them, ω j (n) represents the weight value of standard C in the n-th group of random weights. j of the weight.

[0093] Use the Dirichlet distribution for random sampling, with ω* as the center:

[0094] ω (n) ~Dirichlet(βω * );

[0095] Among them, β is the adjustment parameter, controlling the randomness range.

[0096] S45. For each group of random weights ω (n) , use the VIKOR method to calculate the utility value of each candidate solution, and rank the candidate solutions according to the utility value, specifically:

[0097] S451. Calculate the positive ideal value and negative ideal value

[0098] S452. Calculate the group utility value U m (n) and individual regret value Vm (n) , U m (n) also represents Solution A m The comprehensive distance between the nth group of random weight vectors ω (n) and the ideal solution, V m (n) also represents Solution A m The distance to the single most unfavorable index under the nth group of random weight vectors ω (n) :

[0099]

[0100] S453. Based on U m (n) and V m (n) , calculate the VIKOR comprehensive score Q m of Solution A under the nth group of random weight vectors ω (n) : m (n) :

[0101]

[0102] where U + and U - represent the minimum and maximum utility values among all candidate solutions respectively; V + and V - represent the minimum and maximum regret values among all candidate solutions respectively;

[0103] S454. Rank the candidate solutions according to the utility values.

[0104] S46. Through Monte Carlo simulation, count the occurrence frequencies of candidate solutions at different rankings, calculate the ranking acceptability index, and evaluate the solution stability; calculate the ranking acceptability index b based on the utility values m,p :

[0105]

[0106] In the formula, refers to whether Solution A m is ranked pth under the nth group of random weights ω (n) :

[0107] S47. Calculate the central weight vector and confidence factor, and conduct in-depth analysis on the solution stability to evaluate its robustness under different weight distributions;

[0108] Central weight vector:

[0109]

[0110] In the formula, represents A m whether it ranks first under the random weight ω of the nth group (n) ;

[0111] Confidence factor:

[0112]

[0113] wherein, O m the smaller, the less sensitive the scheme A m is to the weight change and the higher the stability. O m the larger, the more susceptible the ranking of the scheme A m is to the weight change and the lower the stability.

[0114] S48. According to the needs of the decision maker, select the scheme with the highest overall acceptability index as the final recommended scheme, or select the scheme with the smallest confidence factor O m as the most robust scheme.

[0115] Now, taking the optimal operation of a certain section of the Eastern Route Project of the South-to-North Water Diversion Project as an example, the rationality and effectiveness of the method of the present invention are illustrated. As Figure 2 shown, a certain section of the Eastern Route Project of the South-to-North Water Diversion Project is located between 32°15′-34°30′N and 117°00′-119°45′E. The water supply scope involved in the project is 62000 km 2 , and it includes a total of 3 water sources, namely the Yangtze River, Hongze Lake and Luoma Lake, 2 water conveyance lines (Yunxi River and Canal Line) and 14 pumping stations. The dead water levels of Hongze Lake and Luoma Lake are 11.3 m and 21 m respectively, the normal storage levels are 13.5 m and 23 m respectively, and the flood control limited levels are 12.5 m and 22.5 m respectively. The parameters of the pumps and sluice gates of the Eastern Route Project of the South-to-North Water Diversion Project are shown in Table 1.

[0116] Table 1 Parameters of pumps and sluice gates in the Jiangsu section of the Eastern Route Project of the South-to-North Water Diversion Project

[0117]

[0118] The present invention takes the water supply volume of each water receiving area and the pumping volume of the pumps as decision variables, adopts NSGA-Ⅲ for optimal operation, and aims to minimize the water shortage rate, pumping volume, water pumping volume from the Yangtze River, water abandonment rate and water supply uniformity coefficient of the cross-basin water diversion project, with water volume balance, lake water level limit, pumping station working capacity, control sluice station flow capacity, river water conveyance capacity, etc. as constraint conditions. The specific parameter settings of the NSGA-Ⅲ algorithm are determined as follows: the population size is 100, the maximum number of iterations is 30000, the size of the external archive set is 100, and a typical wet year is selected for optimal operation. The spatial distribution of the operation scheme set is as Figure 3 shown. From Figure 3It can be seen that there is an obvious competitive relationship in the spatial distribution of the scheduling scheme set. The scheduling schemes are widely and evenly distributed. There is an obvious competitive relationship between the two objectives of the minimum total pumping volume and the minimum total water shortage rate. The smaller the total pumping volume, the larger the total water shortage rate. Therefore, the optimized scheduling scheme set of the reservoir group solved by the NSGA-Ⅲ algorithm is reasonable and effective.

[0119] Based on the Pareto non-dominated solution set, the ISMAA-VIKOR method is used to determine the optimal scheduling scheme of the reservoir group. In this project, meeting the water supply demand of the water-receiving areas along the line is the core objective. Therefore, 20 schemes with the minimum total water shortage rate are used as alternative schemes. In addition, the optimized operation scheme not only requires the pumping volume and water shortage rate to be as small as possible, but also needs to reduce the total system water abandonment volume, ensure the uniformity of water supply along the line, and minimize the pumping volume from the river. These 5 key indicators are selected as evaluation indicators to construct a comprehensive evaluation index system for scheme optimization (see Table 2).

[0120] Table 2 Comprehensive evaluation index system

[0121]

[0122]

[0123] The subjective and objective weights calculated by the analytic hierarchy process, entropy weight method and CRITIC method are shown in Table 3. Table 3 also shows the comprehensive weight values calculated based on the game theory method. The optimal scheme number obtained based on the ISMAA-VIKOR multi-attribute decision-making model is 1. The target values of the corresponding total water transfer volume, total water shortage rate, water supply uniformity coefficient, total water abandonment volume, and pumping volume from the river are 30.923 billion m 3 ³, 1.17%, 0.0611, 18.459 billion m 3 ³ and 72.67184.59 billion m 3 .

[0124] Table 3 Weights of different multi-attribute decision-making methods

[0125]

[0126]

[0127] Therefore, the present invention adopts the above-mentioned multi-objective optimization scheduling method for reservoir group coupling the ISMAA-VIKOR decision-making model, combines the advantages of subjective weighting and objective weighting through the game theory method, and realizes the scientificity and rationality of weight distribution; at the same time, combines the advantages of SMAA-2 and VIKOR, constructs a unified framework, improves the multi-objective comprehensive benefit of the scheduling scheme in the optimization model, significantly enhances the independent decision-making ability of the reservoir group scheduling system, and improves the comprehensive operation benefit of the reservoir group scheduling.

[0128] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-objective optimization scheduling method for a reservoir group coupled with the ISMAA-VIKOR decision model, characterized in that the steps include: S1. Obtain the basic operation data and related parameters of the reservoir group, and systematically generalize them based on the topological structure of the reservoir group; S2. Analyze the specific dispatching and operation tasks of the reservoir group, and build a multi-objective optimization dispatching model for the reservoir group based on the basic dispatching principles of the reservoir group; S3. Use NSGA-Ⅲ algorithm to solve the model and obtain the Pareto non-inferior solution set of the multi-objective optimization scheduling model of the reservoir group; S4. Establish an evaluation index system for optimizing the dispatching scheme, and combine it with the ISMAA-VIKOR multi-attribute decision-making model to determine the optimal dispatching operation scheme for the reservoir group.

2. According to claim 1, a multi-objective optimization scheduling method for a reservoir group coupled with the ISMAA-VIKOR decision model is characterized in that: The basic operation data and related parameters of the reservoir group in step S1 include: the inflow runoff data of each reservoir in the reservoir group, the water level-storage capacity relationship curve of the reservoir, the characteristic parameters of the water conservancy project facilities and the water demand of the receiving area. The characteristic parameters of the water conservancy project facilities include the water delivery capacity limit of the river channel and the operation parameters of the water pumps and gates.

3. The multi-objective optimization scheduling method for a reservoir group coupled with the ISMAA-VIKOR decision model according to claim 1 is characterized by: The multi-objective optimization scheduling model of the reservoir group takes the minimization of water shortage rate, pumping volume, river water pumping volume, water abandonment rate and water supply uniformity coefficient as the objective function, and the constraints include the reservoir storage capacity, pumping station water transfer capacity, gate flow capacity, river water transfer capacity and water balance.

4. The multi-objective optimization scheduling method for a reservoir group coupled with the ISMAA-VIKOR decision model according to claim 1 is characterized in that: Step S3 specifically includes: S31. Input the basic operation data and related parameters required for the operation of the multi-objective optimization scheduling model of the reservoir group, and set the reservoir storage capacity, pump station water transfer capacity, gate flow capacity, river water transfer capacity and water balance constraints accordingly. And give the feasible range of reservoir water level, pump station water lifting capacity, gate water discharge and river water transfer capacity, set the number of solutions S and the number of iterations Iter max ; S32, initializing the pumping volume of the pump station and the discharge volume of the gate; S33, simulating the operation of the reservoir group in each period, and calculating the water level at the end of each period of each reservoir based on the water balance equation as the initial water level of the reservoir in the next period; S34, after the simulation of the complete scheduling period is completed, the water extraction volume of each pumping station and the actual water supply volume of each water receiving area are counted, and the total water extraction volume and total water shortage rate of the inter-basin water diversion project are calculated, and then go to step S35; S35, using NSGA-Ⅲ algorithm to update the pumping capacity of the pump station and the discharge capacity of the gate in the next stage, and return to step S33 until the number of iterations reaches the preset maximum number of iterations Iter max , and obtain the Pareto non-inferior solution set of the multi-objective optimization scheduling model of reservoir groups.

5. The multi-objective optimization scheduling method for a reservoir group coupled with the ISMAA-VIKOR decision model according to claim 1 is characterized in that: Step S4 specifically includes: S41, let the candidate solution set be A = {A1, A2, ..., A m }, where A m represents the mth solution, m = 1, 2, .., M, each solution A m With J evaluation attributes C = {C1, C2, ..., C J }, construct the decision matrix X = [x mj ] M×J ; S42. Use K subjective and objective weighting methods to perform initial weighting, including analytic hierarchy process, CRITIC method and entropy weight method. Each method calculates the initial weight vector: in, Denotes the standard C for calculating the weighting method k j The weight value of S43. The game theory optimization model is used to determine the final comprehensive weight, and an optimization game framework for multi-method weighting is established. It is assumed that each weighting method is a game participant, and its goal is to minimize the deviation between its own weight and the final comprehensive weight. The comprehensive weight ω* is expressed by solving the optimization problem: In the formula, ω (k) The weight vector calculated for the weighting method k; α k is the weight balance factor of the kth weighting method; the optimal ω* is obtained by solving the quadratic programming or convex optimization method; S44, using ω* as the mean of the prior distribution for random weight sampling to ensure compliance with existing preference information; Random weight vector: Among them, ω j (n) Represents the random weights of the nth group, standard C j The weight value of Random sampling is performed using a Dirichlet distribution with ω* as the center: oh (n) ~Dirichlet(b) * ); Among them, β is an adjustment parameter that controls the range of randomness; S45, for each group of random weights ω (n) , use the VIKOR method to calculate the utility value of each candidate solution, and rank the candidate solutions according to the utility value; S46. Through Monte Carlo simulation, the frequency of candidate solutions appearing in different rankings is counted, and the ranking acceptability index is calculated to evaluate the stability of the solution; the ranking acceptability index b is calculated based on the utility value m,p : In the formula, Refers to Plan A m In the nth group of random weights ω (n) Is it ranked p? S47, calculate the central weight vector and confidence factor, conduct in-depth analysis on the stability of the scheme, and evaluate its robustness under different weight distributions; Center weight vector: In the formula, Indicates A m In the nth group of random weights ω (n) Is it ranked first? Confidence Factor: Among them, O m The smaller the size, the better the solution A m It is not sensitive to weight changes and has high stability. m The larger the m The ranking is easily affected by weight changes and has low stability; S48. According to the needs of decision makers, select the solution with the highest overall acceptability index as the final recommendation, or select the solution with the highest confidence factor O. m The smallest solution is the most robust solution.

6. A multi-objective optimization scheduling method for a reservoir group coupled with the ISMAA-VIKOR decision model according to claim 5, characterized in that: Step S45 specifically includes: S451. Calculate the positive ideal value of each indicator based on the standardized matrix and negative ideal value S452, calculate group utility value U m (n) and individual regret value V m (n) , U m (n) Also means plan A m In the nth group of random weight vectors ω (n) The comprehensive distance between the ideal solution and the m (n) Also means plan A m In the nth group of random weight vectors ω (n) The distance to the single most unfavorable indicator: S453, based on U m (n) and V m (n) , calculation scheme A m In the nth group of random weight vectors ω (n) VIKOR comprehensive score Q m (n) : Among them, U + and U - Represent the minimum and maximum utility values ​​of all candidate solutions respectively; V + and V - Represent the minimum and maximum regret values ​​among all candidate solutions respectively; S454. Rank the candidate solutions according to their utility values.

Images