Trans-basin water transfer project adaptability regulation and control method under uncertainty

Abstract

The invention discloses a cross-basin water transfer project adaptability regulation and control method under uncertainty, and belongs to the technical field of reservoir regulation and control. Comprising the following steps: step 1, generating multiple uncertain scenes as an input side of an interbasin water transfer project simulation scheduling model; step 2, constructing an interbasin water transfer project simulation scheduling model, and analyzing project operation conditions in an uncertain scene; 3, constructing an adaptive scheduling framework based on a hedging principle, and performing multi-objective optimization by using an NSGA-II algorithm; according to the method, an AOMU model for coordinating water transfer, power generation and energy consumption targets under adverse conditions is established, the model can be used in a cross-basin water transfer project under uncertainty, the toughness of the project to cope with the uncertainty and adverse water transfer conditions can be effectively improved, and economic benefits are improved while water transfer is guaranteed; the problem that supply and demand imbalance is easily caused by improper scheduling of traditional uncertainty factors is solved.

Description

Technical Field

[0001] The invention relates to the technical field of reservoir regulation, and in particular to an adaptive regulation method for a cross-basin water diversion project under uncertainty. Background Art

[0002] With the increasing demand for water and the uneven distribution of available water resources in time and space, we are facing the risk of water shortage. To deal with this problem, the construction of inter-basin water transfer projects is one of the main means to balance water resources and meet regional water demand. Inter-basin water transfer is artificially regulated to adjust water resources from water-rich areas to water-deficient areas within and between basins, and to meet the needs of different receiving areas through resource allocation. The multi-reservoir system of the water transfer project delivers water and energy to multiple target demand sides through joint scheduling, and effectively takes into account water supply benefits, ecological benefits and power generation benefits through water diversion in the water source area and water diversion in the water receiving area. It is an effective measure to alleviate the pressure of water supply and demand and promote social development.

[0003] The risks brought by the uncertainty of supply and demand input, scheduling model, and operating parameters are one of the main factors affecting the reliability of reservoir operation. The randomness of water inflow and the hydrological connection and hydraulic coupling between reservoirs bring challenges to the optimal scheduling of reservoir groups. Traditional reservoir scheduling rules mainly determine the reservoir water transfer strategy based on the current period of water inflow and predicted values, which is difficult to cope with the decision-making needs under potential uncertainties. In addition, due to the existence of uncertainty factors, the risk of supply and demand imbalance caused by improper scheduling always exists. Therefore, how to reduce or alleviate the impact of uncertainty has become a hot and difficult issue in current research.

[0004] Based on this, the present invention proposes an adaptive control method for inter-basin water diversion projects under uncertainty to solve the problems existing in the above-mentioned prior art. Summary of the invention

[0005] In view of this, the main purpose of the present invention is to provide an adaptive control method for inter-basin water diversion projects under uncertainty, so as to solve the problem that traditional uncertain factors are prone to cause imbalance between supply and demand due to improper scheduling.

[0006] The technical solution of the present invention is achieved in this way:

[0007] An adaptive control method for inter-basin water diversion projects under uncertainty, including:

[0008] Step 1: Generate multiple uncertainty scenarios

[0009] Select uncertainty factors, capture the temporal dependence of uncertainty factors through Markov Monte Carlo and three-dimensional Copula functions, and generate multiple uncertainty scenarios as the input side of the inter-basin water diversion project simulation scheduling model;

[0010] Step 2: Construct a simulation scheduling model for inter-basin water diversion projects based on SOP

[0011] According to the principle of water balance and the actual constraints of the project, a water supply sequence is selected to formulate a standard operation strategy to build a simulation scheduling model for inter-basin water diversion projects based on SOP, and analyze the project operation under uncertain scenarios;

[0012] Step 3: Build an adaptive scheduling framework based on the hedging principle

[0013] Based on the constructed multiple uncertainty scenarios, the storage capacity hedging strategy is introduced, the energy index is added, an adaptive scheduling framework based on the hedging principle is constructed, and the NSGA-II algorithm is used for multi-objective optimization.

[0014] In a preferred embodiment, the process of generating multiple uncertainty scenarios in step 1 includes:

[0015] Step 1-1: Identify uncertainty factors

[0016] During the operation of the reservoir, the errors generated during the simulation of each uncertainty factor are the risk source; the prediction uncertainty is defined as the error between the measured value and the simulated value; the error probability distribution method generates a probabilistic forecast by assuming the distribution of the errors of each factor, and the prediction sequence under the scenario with simulation errors is obtained as follows:

[0017] q t =x t +ε t =f(X t )+ε t (1)

[0018] In formula (1), q t is the output value of the uncertainty factor; x t and f(X t ) is the predicted value and prediction function of the uncertainty factor; ε t Indicates the corresponding error;

[0019] Step 1-2: Generate multiple uncertainty scenarios based on the uncertainty factors identified in step 1-1;

[0020] Wherein: the multiple uncertainty scenarios include uncertainty scenarios of independent variables and multi-source uncertainty scenarios with related dependent variables.

[0021] In a preferred embodiment, the uncertainty scenarios of the independent variables are generated based on MCMC, including:

[0022] Step 1-2-1-1: Assume that the conditional probability distribution of the current observation is only related to the most recent observation and is independent of all previous observations. The first-order Markov chain under the measured conditions is:

[0023]

[0024] In formula (2), Q(q t ) is the measured runoff at time t, T is the sequence length;

[0025] Step 1-2-1-2: Expand the first-order Markov chain and introduce the implicit state variable z t ∈{1,…,M},z t It follows a first-order Markov process, which is at z t The probability of t-1 Determined by; runoff observation model Q(q t |z t ) depends on the state variables; z t The probability of changing from one state to another is called the state transition probability, and the combination of all state transition probabilities is represented by the state transition matrix A;

[0026] According to the distribution characteristics of runoff, the "wet and dry" state obeys the Markov random process, and the runoff sequence is affected by the "wet and dry" state, which constitutes the implicit Markov model of runoff; the implicit state variable z of runoff t It is divided into flood season, normal season and low season, so its state transfer matrix Α is:

[0027]

[0028] In formula (3), A ij =q(z t =j|z t-1 =i) represents the probability of the state variable changing from i to j, and Α satisfies 0≤A ij ≤1 and During the flood season, t =1; at normal water level, z t =2; during low water period, z t =3;

[0029] Step 1-2-1-3: Integrate the discretized scenarios in the sample set into runoff uncertainty scenarios that are consistent with actual laws based on the Markov chain.

[0030] In a preferred embodiment, the multi-source uncertainty scenario with related dependent variables is generated by relying on a three-dimensional Copula function, including:

[0031] Step 1-2-2-1: Use the three-dimensional Copula function to establish multiple uncertainty scenarios based on inflow prediction; the joint density function and the corresponding random variables are represented as:

[0032] p(x 1 ,…,x m )=c(u 1 ,…,u m )·p(x 1 )…p(x m ) (4)

[0033]

[0034] P(x i )=u i (i=1,…,m) (6)

[0035] In formula (4) to formula (6), p(x 1 ),…,p(x m ) and P(x 1 ),…,P(x m ) are the related random variables X 1 ,…,X m The probability density function and cumulative distribution of C(u 1 ,…,u m ) is the Copula function;

[0036] Step 1-2-2-2: Based on the inflow, adjustable water volume and water demand of historical observations and forecast samples, the Kendall and Spearman rank correlation coefficients are used to evaluate the correlation between characteristic variables.

[0037] In a preferred embodiment, the inter-basin water diversion project simulation scheduling model in step 2 includes an input side, a scheduling side and an output side; wherein:

[0038] The input side is used to reflect the initial storage conditions, final boundary conditions, engineering constraints and generated multiple uncertainty scenarios of actual observation results;

[0039] The dispatching side focuses on the actual operation of the inter-basin water diversion project, uses the SOP strategy to target the multiple reservoirs in the water source area, and selects the timing for starting water diversion of each reservoir based on the project's storage capacity and natural water supply conditions;

[0040] The output side is used to output the outflow, water storage, energy consumption and power generation of each reservoir.

[0041] In a preferred embodiment, the construction process of the adaptive scheduling framework based on the hedging principle in step 3 includes:

[0042] Step 3-1: Construct an AOMU model based on storage capacity hedging strategy;

[0043] Step 3-2: Construct the objective and constraint functions of the AOMU model;

[0044] Step 3-3: Introduce the NSGA-Ⅱ algorithm to perform multi-objective optimization on the AOMU model.

[0045] In a preferred embodiment, the objective function of the AOMU model is:

[0046] Goal 1: Maximize water transfer

[0047]

[0048] In formula (7), t = 1, ..., T and n = 1, ..., N represent the operation period and reservoir respectively, T is the number of operation periods, and N is the number of reservoirs; Q n,t is the total water transfer volume of a reservoir during a certain operation period; Δt is the dispatching period;

[0049] Goal 2: Minimize water waste

[0050]

[0051] In formula (8), The water discharge of a reservoir during a certain operation period;

[0052] Objective 3: Minimize the water shortage index during water transfer

[0053]

[0054] In formula (9), is the total water transfer volume of the project during a certain operation period, It is the total water demand of the receiving area during a certain operation period of the project;

[0055] Goal 4: Maximize power generation

[0056]

[0057] In formula (10), It is the discharge flow of a reservoir used for power generation during a certain operation period, and the rest is the same as above;

[0058] Goal 5: Minimize energy consumption

[0059]

[0060] In formula (11), H n,t is the average hydraulic head of a reservoir during a certain operation period; It is the water pumping flow of a pumping station of a reservoir during a certain operation period.

[0061] In a preferred embodiment, the constraint function of the AOMU model is:

[0062] Constraint 1: Reservoir water balance

[0063]

[0064] In formula (12), V i,j,s and V i+1,j,s For the s The initial and final storage capacity of the i-th period in the j-th year under the scenario; I i,j,s and Q i,j,s Respectively s The reservoir inflow and outflow in the i-th period in the j-th year under the scenario; V 1,j+1,s For the s The initial storage capacity in the j+1th year under the scenario; V N+1,j,s No. s The storage capacity at the end of the jth year under the scenario, N is the total number of time periods;

[0065] Constraint 2: Reservoir capacity

[0066]

[0067] In formula (13), VU i and VL i are the upper and lower limits of the reservoir capacity respectively;

[0068] Constraint 3: Maximum flow rate

[0069]

[0070] In formula (14), QU i and QL i are the upper and lower limits of the reservoir flow rate respectively;

[0071] Constraint 4: Maximum output

[0072]

[0073] In formula (15), NU i and NL i They are the upper and lower limits of power generation output respectively.

[0074] Compared with the prior art, the present invention provides an adaptive control method for inter-basin water diversion projects under uncertainty, which has the following beneficial effects:

[0075] 1. This method establishes an AOMU model that coordinates water transfer, power generation and energy consumption targets under adverse conditions by generating multiple uncertainties, constructing a simulation scheduling model and constructing an adaptive optimization model. This model can be used in inter-basin water transfer projects under uncertainty, which can effectively improve the resilience of the project to deal with uncertainty and adverse water transfer conditions, and improve economic benefits while ensuring water transfer.

[0076] 2. The uncertainty factors of inflow runoff, water demand and adjustable water volume selected by this method reflect the key to improving the resilience of the Han-Wei inter-basin water diversion project. At the same time, the three are highly correlated, and the marginal distribution fitting satisfies Clayton Copula. The constructed joint scenario can accurately describe the multidimensional dependency relationship between different factors.

[0077] 3. The simulation results of this method under the multi-source uncertainty scenario show that the total water transfer volume is maintained at 1.347 to 1.517 billion m 3 , an average increase of 9.2%, which can adapt to the different characteristics of water inflow and regulation conditions in Huangjinxia and Sanhekou.

[0078] 4. After AOMU optimization, the annual water transfer volume is guaranteed to be higher than 1.3 billion m3 under unfavorable conditions. 3 Under the premise of reducing the total energy consumption of the project by 14%, increasing the power generation by 2.76%, and significantly improving the economic benefits, it solves the problem that the traditional uncertainty factors are prone to cause imbalance between supply and demand due to improper scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained like these drawings without paying creative work.

[0080] Figure 1 It is a flow chart of the adaptive control method of the inter-basin water diversion project under uncertainty of the present invention;

[0081] Figure 2 This is an overview of the research area of ​​the present invention;

[0082] Figure 3 It is a framework diagram of the adaptive control method of the inter-basin water diversion project under uncertainty of the present invention;

[0083] Figure 4 A block diagram generated for multiple uncertainty scenarios of the present invention;

[0084] Figure 5 It is a simulation scheduling model framework based on SOP of the present invention;

[0085] Figure 6 It is the flow chart of NSGA-II algorithm of the present invention;

[0086] Figure 7 This is a runoff uncertainty scene diagram of Example 2 of the present invention;

[0087] Figure 8 This is a kernel density estimation diagram of runoff-adjustable water volume-water demand in Example 2 of the present invention;

[0088] Fig. 9 This is a diagram of water regulation, energy consumption and power generation for simulating scheduling in Example 2 of the present invention;

[0089] Fig.10 This is a diagram showing the interannual variation of the amount of water transferred from Huangjinxia and Sanhekou to the receiving area according to Example 2 of the present invention;

[0090] Fig.11 This is a distribution diagram of water transfer from Huangjinxia and Sanhekou to the receiving area in Example 2 of the present invention;

[0091] Fig.12 This is a diagram of energy index scheduling results in Example 2 of the present invention;

[0092] Fig.13 This is a comparison chart of scheduling results under uncertainty scenario input and historical observation data input in Example 2 of the present invention;

[0093] Fig.14 This is a multi-objective optimization result diagram of Example 2 of the present invention;

[0094] Fig.15 This is a parallel coordinate axis diagram of the water volume, energy consumption and power generation of the simulation and optimization model of Example 2 of the present invention.

[0095] Among them, Figure 7 In the figure, Figure (a) is the monthly runoff scenario of Huangjinxia; Figure (b) is the monthly runoff scenario of Sanhekou; Figure (c) is the annual runoff scenario of Huangjinxia; Figure (d) is the annual runoff scenario of Sanhekou;

[0096] exist Figure 8 In the figure, Figure (a) is the kernel density estimation diagram of runoff-water demand; Figure (b) is the kernel density estimation diagram of runoff-adjustable water volume;

[0097] exist Fig. 9 In the figure, Figure (a) is the water volume diagram under simulated scheduling; Figure (b) is the energy consumption diagram under simulated scheduling; Figure (c) is the power generation diagram under simulated scheduling;

[0098] exist Fig.10Figure (a) shows the interannual variation of water transfer from Huangjinxia and Sanhekou to the receiving area in the low-water series; Figure (b) shows the interannual variation of water transfer from Huangjinxia and Sanhekou to the receiving area in the normal-water series;

[0099] exist Fig.11 In the figure, Figure (a) is the distribution map of water transfer volume from Huangjinxia to the receiving area; Figure (b) is the distribution map of water transfer volume from Sanhekou to the receiving area; Figure (c) is the distribution map of water transfer density from Huangjinxia to the receiving area; Figure (d) is the distribution map of water transfer density from Sanhekou to the receiving area;

[0100] exist Fig.12 Figure (a) shows the monthly power generation distribution of Huangjinxia Figure 1 Figure (b) shows the monthly power generation distribution of Huangjinxia Figure 2 ; Figure (c) is the monthly energy consumption distribution map of Huangjinxia; Figure (d) is the monthly energy consumption distribution map of Sanhekou;

[0101] exist Fig.13 In , a value greater than zero indicates that the uncertainty scheduling result is greater than the design year scheduling result, and vice versa;

[0102] exist Fig.14 In the figure, Figure (a) is the multi-objective optimization result diagram of the low-water series; Figure (b) is the multi-objective optimization result diagram of the normal-water series;

[0103] exist Fig.15 In the figure, Figure (a) is a parallel coordinate axis diagram of water volume, energy consumption, and power generation in the low-water series simulation and optimization model; Figure (b) is a parallel coordinate axis diagram of water volume, energy consumption, and power generation in the normal-water series simulation and optimization model. DETAILED DESCRIPTION

[0104] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0105] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0106] Embodiment 1:

[0107] The following is attached to the instruction manual Figure 1-Figure 6The present invention describes the adaptive control method of inter-basin water diversion projects under uncertainty. This method is mainly used to solve the adaptive optimization scheduling and multi-objective operation problems of inter-basin water diversion projects under multiple uncertainties. This method takes the reservoir group in the water source area of ​​the Han-Wei inter-basin water diversion project as the research object, and uses Markov Monte Carlo and three-dimensional Copula functions to construct multiple uncertainty scenarios; and applies the joint scenarios to quantify the impact of uncertainty on indicators such as water diversion volume and power generation; based on the reservoir capacity hedging theory, constructs an AOMU model that coordinates water diversion, power generation and energy consumption targets under adverse conditions. Specifically:

[0108] Overview of research subjects:

[0109] 1. The inter-basin water diversion project of Han River to Wei River. The Wei River occupies an important position in the management and development of water resources in the Yellow River Basin. However, since the Wei River Basin is in the transition zone between drought and humidity, the temporal and spatial distribution of runoff is uneven, making the Wei River Basin, especially the middle and lower reaches, vulnerable to drought. The Guanzhong region is adjacent to the Wei River. It is difficult to meet the needs of economic growth in the Guanzhong region by relying solely on the water resources of the Wei River Basin. Therefore, the inter-basin water diversion project of Han River to Wei River is planned to alleviate the prominent water use contradiction in the Guanzhong region. The project diverts water from the Han River Basin in southern Shaanxi to the Guanzhong region in the Wei River Basin. The water receiving area has a total of 21 water demand units, including four large cities, an industrial park and 16 small and medium-sized towns along both sides of the Wei River. The project area is about 163km long from east to west and about 84km wide from north to south, with a total water receiving area of ​​about 14,000km 2 The project mainly includes Huangjinxia Water Conservancy Project, Sanhekou Water Conservancy Project, Qinling Water Conveyance Tunnel and Water Conveyance and Distribution Pipeline Network, etc. ( Figure 2 ), among which Huangjinxia and Sanhekou are two important water sources for the project.

[0110] Huangjinxia is located in the mainstream of Hanjiang River, with abundant natural water supply. The average annual runoff into the reservoir is 6.64 billion cubic meters, and the average annual planned water transfer is about 1 billion cubic meters, but the reservoir has no regulation capacity. In addition, the Huangjinxia pump station is located behind the dam and is arranged together with the power station. Its function is to lift the water resources of Huangjinxia to the water inlet of the water transfer tunnel, so Huangjinxia requires a lot of energy when transferring water.

[0111] Sanhekou is located in the canyon section of the middle reaches of the Ziwu River. The average annual runoff into the reservoir is 860 million cubic meters, and the average annual planned water transfer is about 550 million cubic meters. The reservoir has a multi-year regulation capacity of 649 million cubic meters. The water transfer from Sanhekou to the receiving area is gravity diversion, and no pump station energy consumption is required. The main task of its pump station is to replenish the excess water after the water transfer from Sanhekou into the remaining storage capacity of Sanhekou, and the energy consumption is far less than that of the Huangjinxia Pump Station.

[0112] In summary, the problems that still exist in the Han-Wei inter-basin water diversion project are:

[0113] (1) The water source area is affected by the temperate continental climate, and the runoff is unevenly distributed throughout the year. The water volume difference between the flood and dry seasons is obvious, and the runoff uncertainty has a great impact on the operation of the project;

[0114] (2) The natural water inflow from Huangjinxia and Sanhekou does not match the reservoir storage capacity. As the main water source, Huangjinxia consumes a lot of energy when transferring water, which affects the economic benefits of the project operation.

[0115] 2. Data source. This method uses the monthly runoff data of Huangjinxia and Sanhekou from 1956 to 2009, the water demand in Guanzhong area and the maximum adjustable water volume as the input side of the multiple uncertainty scenario generation framework; the parameters of reservoirs, power stations and pumping stations in the water source area are used as the constraints of simulation-optimization. Among them, the reservoir inflow data and the parameters of the Han-Wei Diversion Project are provided by the Han-Wei Diversion Project Construction Co., Ltd. The main engineering characteristic parameters used in this method are shown in Table 1.

[0116] Table 1: Main operating parameters of Huangjinxia and Sanhekou

[0117]

[0118]

[0119] Based on the above data and problems, the main process of the adaptive regulation method of inter-basin water diversion project under uncertainty designed by the present invention includes the following steps:

[0120] Step 1: Generate multiple uncertainty scenarios; select uncertainty factors, capture the temporal dependence of uncertainty factors through Markov Monte Carlo (MCMC) and three-dimensional Copula function, and generate multiple uncertainty scenarios as the input side of the inter-basin water diversion project simulation scheduling model

[0121] Step 1-1: Identify uncertainty factors

[0122] During the operation of the reservoir, the error generated in the simulation process of each uncertainty factor is the risk source; the prediction uncertainty is defined as the error between the measured value and the simulated value; the error probability distribution method generates a probabilistic forecast by assuming the distribution of the error of each factor, and obtains the prediction sequence under the scenario of simulation error, which is expressed as:

[0123] q t =x t +ε t =f(X t )+ε t (1)

[0124] In formula (1), q t is the output value of the uncertainty factor; x t and f(X t) is the predicted value and prediction function of the uncertainty factor; ε t Indicates the corresponding error;

[0125] In inter-basin water transfer projects whose main goal is to meet the water demand of the receiving areas, the inflow volume, water demand and maximum adjustable water volume are the main uncertain factors.

[0126] Step 1-2: Generate multiple uncertainty scenarios based on the uncertainty factors identified in step 1-1

[0127] Step 1-2-1: Generate uncertain scenarios for independent variables

[0128] The runoff series is highly dependent on time, with obvious trends of wet and dry changes within the year and between years, and the regularity between the inter-annual "wet and dry encounters" is more prominent. The Markov Monte Carlo method (MCMC) can describe the pre-existing regularities within the measured results, and extract the likelihood function of the generated change scenarios, thereby converting the data set into a reasonable sample set with predictive properties.

[0129] This method uses MCMC to generate runoff uncertainty scenarios, which is mainly divided into three steps:

[0130] (1) For each monthly runoff under measured conditions, the optimal distribution (normal distribution, gamma distribution, lognormal distribution, GEV distribution) is selected from the distribution types commonly used in hydrological analysis, and its parameter information is calculated;

[0131] (2) Monte Carlo simulation combined with Latin hypercube sampling was used to generate discrete runoff simulation scenarios with equal likelihood;

[0132] (3) By calculating the probability of "flood and drought" in the measured runoff sequence between years, the probability transfer matrix is ​​obtained, and the discretized scenarios in the sample set are integrated into runoff uncertainty scenarios that are more consistent with the actual laws according to the Markov chain;

[0133] Step 1-2-1-1: Markov process, also known as random process without aftereffect, satisfies the translation invariance assumption. By assuming that the conditional probability distribution of the current observation is only related to the most recent observation and is independent of all previous observations, we can obtain a first-order Markov chain under measured conditions, expressed as:

[0134]

[0135] In formula (2), Q(q t ) is the measured runoff at time t, T is the sequence length;

[0136] Step 1-2-1-2: Expand the first-order Markov chain and introduce the implicit state variable z t ∈{1,…,M},zt It follows a first-order Markov process, which is at z t The probability of t-1 Determined by; runoff observation model Q(q t |z t ) depends on the state variables; z t The probability of changing from one state to another is called the state transition probability. The combination of all state transition probabilities can be represented by the state transition matrix A.

[0137] According to the distribution characteristics of runoff, its "flood-dry encounter" state obeys the Markov random process, and the runoff sequence is affected by the "flood-dry encounter" state, thus forming a hidden Markov model of runoff. The implicit state variable z of runoff t It can be divided into Fengshui (z t =1), flat water (z t =2) and low water (z t =3), so its state transfer matrix A can be expressed as:

[0138]

[0139] In formula (3), A ij =q(z t =j|z t-1 =i) represents the probability of the state variable changing from i to j, and Α satisfies 0≤A ij ≤1 and ∑A ij =1.

[0140] Step 1-2-1-3: Integrate the discretized scenarios in the sample set into runoff uncertainty scenarios that are more consistent with the actual laws based on the Markov chain;

[0141] This method uses Monte Carlo to generate 1000 sets of discrete runoff prediction sequences for Huangjinxia and Sanhekou each month, and uses Latin hypercube sampling to obtain 1000 sets of discrete annual runoff sequences. According to the state transfer matrix Α, the discrete scenarios can be combined to obtain 20 sets of 15-year runoff prediction scenarios.

[0142] Step 1-2-2: Generate multi-source uncertainty scenarios with relevant dependent variables

[0143] Each uncertainty factor often has a certain degree of autocorrelation and conditional heteroscedasticity in the time series, which makes it difficult to meet the assumption of independent and identically distributed variables in statistical analysis. In the inter-basin water transfer system, the adjustable water volume is highly dependent on the total water inflow. Therefore, the prediction error of the adjustable water volume is related to the flow error. Overestimated inflow usually corresponds to overestimated adjustable water volume, and vice versa. Water demand is also affected by adjustable water volume and inflow runoff. Due to the dependence between inflow and water demand and adjustable water volume in time continuity, there is a dominant dependency in the generation of multiple uncertainty scenarios.

[0144] Step 1-2-2-1: Relying on the advantage of multidimensional Copula function in expressing the marginal distribution and dependency relationship of any dependency structure, this method uses three-dimensional Copula function to establish multiple uncertainty scenarios based on inflow prediction; the joint density function and the corresponding random variable are characterized as follows:

[0145] p(x 1 ,…,x m )=c(u 1 ,…,u m )·p(x 1 )…p(x m ) (4)

[0146]

[0147] P(x i )=u i (i=1,…,m) (6)

[0148] In formula (4) to formula (6), p(x 1 ),…,p(x m ) and P(x 1 ),…,P(x m ) are the related random variables X 1 ,…,X m The probability density function and cumulative distribution of C(u 1 ,…,u m ) is the Copula function;

[0149] Step 1-2-2-2: Based on the inflow, adjustable water volume and water demand of historical observations and forecast samples, the Kendall and Spearman rank correlation coefficients are used to evaluate the correlation between characteristic variables to determine whether the variables are suitable for constructing a joint distribution function;

[0150] In order to more accurately capture the dependence of the change process of uncertainty factors, five single variables (LOGN, Gamma, GEV, P-III, Gaussian) are selected as candidate distributions; the Copula functions used in multivariate analysis mainly include elliptic family, extreme value family and Archimedean family, and the Archimedean family Copula function can capture a wide range of dependency structures, so Gumbel copula, Clayton copula and Frank copula are selected to construct the joint distribution, and the maximum likelihood method is used to estimate its parameters. The Akaike Information Criterion (AIC) and root mean square error (RMSE) are used to select the optimal Copula function that characterizes the dependence between variables.

[0151] This step assumes that the inflow runoff is a random process that depends on the density distribution function, so its statistical characteristics can be obtained from historical samples. In steps 1-2-1 and 1-2-2, MCMC simulation is used to generate discretized runoff uncertainty scenarios, and the three-dimensional Copula function is used to obtain the dependencies between various factors, thereby generating uncertainty scenarios for water demand and adjustable water volume based on inflow runoff. The simulation results of various uncertainty factors are superimposed point by point to obtain multiple uncertainty scenarios of runoff-water demand-water diversion. The process of generating multiple uncertainty scenarios is as follows: Figure 4 shown.

[0152] Figure 4 The uncertainty scenario generation process demonstrated here realizes the extension from single-source uncertainty to multi-source uncertainty, taking into account the various prediction uncertainties contained in incomplete information, so that the operation and dispatching strategy of the inter-basin water diversion system can meet future needs as much as possible. The above uncertainty scenario combination not only takes into account the distribution characteristics of each uncertainty factor itself, but also takes into account the actual laws of mutual connection and mutual influence between the units within the reservoir group system under actual circumstances. At the same time, considering the impact of adverse conditions on the project, continuous low water sequence and normal water sequence are added in the scenario combination for subsequent analysis.

[0153] Step 2: Construct a simulation scheduling model for inter-basin water diversion projects based on SOP; according to the water balance principle and the actual constraints of the project, select an appropriate water supply sequence to formulate a standard operating strategy (SOP) to analyze the project operation under uncertain scenarios

[0154] The simulation scheduling model of inter-basin water transfer projects under multiple uncertainties consists of the input side, the scheduling side and the output side. Among them: the input side mainly includes the initial storage conditions, final boundary conditions, engineering constraints and generated multiple uncertainty scenarios that reflect the actual observation results. In the scheduling side, according to the situation of multiple reservoirs in the water source area, the appropriate water transfer start time of each reservoir is selected according to the project storage capacity and natural water supply. After the scheduling operation, the outflow, water storage, energy consumption and power generation of each reservoir can be output on the output side. In the scheduling side, focusing on the actual operation of the inter-basin water transfer project, the SOP strategy is selected to give priority to the water transfer demand of the receiving area, while taking into account ecology and power generation. The simulation scheduling model based on the SOP strategy uses the water balance equation to calculate the adjustable water volume, while considering the inflow runoff of each reservoir in each month and the water demand of the receiving area. Under the SOP strategy, the water transfer reservoir transfers water to the receiving area to the maximum extent within the allowable range of the adjustable water volume. When there is still surplus water after the water transfer, the surplus water will be stored for future use. The simulation scheduling model set in this embodiment is as follows Figure 5 shown.

[0155] Figure 5 The simulation scheduling model presented is based on SOP. In order to solve the problem of mismatch between the natural water inflow and the storage capacity of the two reservoirs in the water source area of ​​the inter-basin water diversion project, a scheduling model was established in which Huangjinxia prioritized water diversion to the water receiving area and Sanhekou replenished water at an appropriate time. This was also the rule and operation mode at the beginning of the design of the project. (1) When the natural water inflow is abundant, Huangjinxia prioritized water diversion to the water receiving area, and the excess water was stored in Sanhekou for future storage; (2) When the natural water inflow is insufficient to support Huangjinxia's water diversion alone, Sanhekou jointly replenished water.

[0156] Huangjinxia has abundant water and only has daily regulation capacity and high lift water, while Sanhekou is the opposite. Therefore, under the SOP mode, Huangjinxia, ​​as the main water source, may result in high energy consumption under the premise of meeting the water diversion demand. By calculating the results of the simulation scheduling model under multiple uncertainties, we focus on analyzing the changes in reservoir water diversion, power generation, energy consumption and other indicators under adverse scenarios such as runoff normal water sequence and low water sequence, analyze the impact of uncertainty factors on project operation, and find room for optimization.

[0157] Step 3: Based on the SOP-based inter-basin water diversion project simulation scheduling model, an adaptive scheduling framework (AOMU) based on the hedging principle is constructed; based on the constructed multiple uncertainty scenarios, a reservoir capacity hedging strategy is introduced, energy indicators are added, and the NSGA-II algorithm is used for multi-objective optimization.

[0158] In the operation of inter-basin water transfer projects, especially under extremely unfavorable conditions of continuous low water flow, there are complexities in decision diversity and objective functions. Under multiple uncertainties, the SOP strategy is the operation and scheduling strategy with the smallest total water shortage. Although it can meet the basic needs of the receiving area, considering only the benefits of the current period may increase the water transfer damage depth of the model. Therefore, in this step, a multi-objective scheduling AOMU framework is constructed based on the reservoir capacity hedging strategy to improve the project operation indicators under adverse scenarios in order to explore the optimal mode of project operation. In the AOMU model, two issues are considered: (1) Under adverse water transfer conditions, how to reduce the damage depth while maximizing the water transfer requirements; (2) Under the low water sequence, how to reduce the energy consumption index while keeping the water transfer volume basically unchanged.

[0159] Step 3-1: Build AOMU model

[0160] The AOMU model applied to the operation of the inter-basin water transfer system is based on the reservoir capacity hedging strategy, retaining the existing water storage for future use. By rationing the water supply, the time and magnitude of the current water transfer reduction are determined to make the system sustainable under the condition of marginal supply reduction. By analyzing and weighing the current and future water demand and project benefits, the water transfer volume under the uncertainty conditions of the current stage and the future stage is formulated to adapt to different hedging strategies corresponding to different goals.

[0161] When the water inflow in the future dispatching period is uncertain and it is known through analysis that there will be a risk of water shortage, in order to ensure the benefits, the high risks that may occur in the future are reduced by accepting a certain loss of benefits in the current period during the dispatching operation of the reservoir group. The AOMU model sets the random water transfer coefficient κ (0<κ<1) of Huangjinxia to reduce the total water transfer volume of Huangjinxia, ​​and the remaining water transfer demand is met by Sanhekou. Some water is reserved for supply when there is a serious risk of water shortage, so as to ensure the efficiency of the project operation under the condition of future uncertainty.

[0162] Step 3-2: Construct the objective and constraint functions of the AOMU model

[0163] The primary task of inter-basin water transfer projects is to ensure the water demand of the receiving areas, so the primary goal of the AOMU model is that the water transfer process must meet actual needs. In addition, in the case of a mismatch between the amount of water coming in and the storage capacity of the project, different water transfer schemes will cause huge changes in energy consumption. In studying the operation of inter-basin water transfer projects involving reservoirs in multiple water source areas, the AOMU model not only considers the water transfer needs of the project, but also minimizes the energy consumption caused by pumping water from pumping stations during water transfer to reduce economic losses, while ensuring the power generation benefits of the project. This is a typical multi-objective hedging problem. When performing adaptive optimization scheduling, the AOMU model makes trade-offs between various objectives and continuously adjusts the corresponding hedging strategies based on new scenario inputs.

[0164] The multi-objective optimal operation model library established by this method can select the corresponding objective function according to the actual needs of the project, mainly including the following objectives:

[0165] Goal 1: Maximize water transfer

[0166]

[0167] In formula (7), t = 1, ..., T and n = 1, ..., N represent the operation period and reservoir respectively, T is the number of operation periods, and N is the number of reservoirs; Q n,t is the total water transfer volume of a reservoir during a certain operation period; Δt is the scheduling period.

[0168] Goal 2: Minimize water waste

[0169]

[0170] In formula (8), It is the water discharge flow of a certain reservoir during a certain operation period, and the rest are the same as above.

[0171] Objective 3: Minimize the water shortage index during water transfer

[0172]

[0173] In formula (9), is the total water transfer volume of the project during a certain operation period, It is the total water demand of the receiving area during a certain operation period of the project, and the rest are the same as above.

[0174] Goal 4: Maximize power generation

[0175]

[0176] In formula (10), It is the discharge flow of a reservoir used for power generation during a certain operation period, and the rest are the same as above.

[0177] Goal 5: Minimize energy consumption

[0178]

[0179] In formula (11), H n,t is the average hydraulic head of a reservoir during a certain operation period; It is the water pumping flow of a pumping station in a certain reservoir during a certain operation period, and the rest are the same as above.

[0180] In the AOMU model, the following equality and inequality constraints are considered:

[0181] Constraint 1: Reservoir water balance

[0182]

[0183] In formula (12), V i,j,s and V i+1,j,s For the s The initial and final storage capacity of the i-th period in the j-th year under the scenario; I i,j,s and Q i,j,s Respectively s The reservoir inflow and outflow in the i-th period in the j-th year under the scenario; V 1,j+1,s For the s The initial storage capacity in the j+1th year under the scenario; V N+1,j,s No. s The storage capacity at the end of the jth year under the scenario, N is the total number of time periods.

[0184] Constraint 2: Reservoir capacity

[0185]

[0186] In formula (13), VU i and VL i They are the upper and lower limits of the reservoir capacity respectively.

[0187] Constraint 3: Maximum flow rate

[0188]

[0189] In formula (14), QU i and QL i They are the upper and lower limits of the reservoir flow rate respectively.

[0190] Constraint 4: Maximum output

[0191]

[0192] In formula (15), NU i and NL i They are the upper and lower limits of power generation output respectively.

[0193] Step 3-3: Introduce NSGA-Ⅱ algorithm to perform multi-objective optimization on the AOMU model

[0194] The NSGA-II algorithm (Non-dominated Sorting Genetic Algorithm-II) is used to solve the multi-objective optimization problem of reservoir group operation under the hedging strategy. Compared with the NSGA algorithm, the main improvements of NSGA-II are:

[0195] (1) A fast non-dominated sort is proposed, which reduces the total time complexity from ο(MN 3 ) dropped to ο(MN 2 ), the space complexity is ο(N2 ), which improves the solution efficiency;

[0196] (2) An elite retention strategy is proposed, which mixes all individuals of the parent generation and the offspring generation and performs non-dominated sorting, and selects the next generation from the doubled population size, thereby preventing the loss of excellent individuals and improving the optimization accuracy;

[0197] (3) A parameterless niching operator is proposed, which uses a crowding comparison operator instead of a sharing function to ensure the diversity of the population. The NSGA-II algorithm flow chart is shown in the figure below: Figure 6 shown.

[0198] The AOMU model proposed in this method is a model based on the combination of simulation and optimization. It can minimize the energy consumption generated during water transfer and maximize the water transfer capacity and power generation. By combining the hedging strategy with the NSGA-II multi-objective optimization algorithm, the water transfer coefficient is set to determine the optimal hedging variable. According to the multiple uncertainty scenarios generated in the early stage, after 100,000 iterations, the optimal water transfer volume, energy consumption and power generation under the hedging level and hedging coefficient in different months were obtained. Finally, the model operation results were compared with the results of the simulated dispatch under the SOP to analyze the changes in reservoir performance under the hedging strategy.

[0199] Embodiment 2:

[0200] Different from the above-mentioned Example 1, after introducing data based on the research object of the above-mentioned Example 1, the experimental data of this Example 1 is used to verify the effectiveness of the adaptive control method of the inter-basin water diversion project under uncertainty described in Example 1. The verification results are as follows: Figure 7-Figure 15 shown.

[0201] Step S1: Uncertainty scenario combination

[0202] This example uses the KS test, RMSE criterion, and AIC information criterion to evaluate the goodness of fit to determine the optimal distribution of the measured inflow runoff in each month, and uses maximum likelihood estimation to fit the corresponding parameters. The specific results of the marginal distribution construction are shown in Table 2;

[0203] Table 2: Runoff distribution fitting results and related parameters

[0204]

[0205]

[0206] The upper and lower limits of the simulation error were controlled to be ±5%, and the monthly runoff scenarios of Huangjinxia and Sanhekou under uncertainty conditions were obtained as follows: Figure 7 (a) and 7(b); the simulated 15-year runoff sequence is shown in Figure 7 (c) and 7(d).

[0207] Depend on Figure 7 (a) and (b) show that the monthly runoff data obtained using MC and LHS are significantly neutral and have obvious seasonal changes; and the runoff data generated in each month contain a small number of extreme values. Figure 7 (c) and 7(d), it can be seen that the 15-year runoff series of Huangjinxia and Sanhekou obtained by MCMC simulation has an obvious trend of change from wet to dry.

[0208] The above rules are highly consistent with the climate conditions and natural water supply conditions in the project water source area, and can be used as simulation results of reservoir runoff under uncertain conditions. In addition, due to the existence of extremely abundant and extremely dry water supply conditions under measured conditions, the sensitivity of the simulated values ​​to extreme water supply conditions is enhanced, especially the dry season dry water conditions are more obvious than the measured values.

[0209] On the basis of establishing marginal distribution, the multidimensional Copula function can be used to construct a joint distribution under uncertainty conditions with multidimensional time dependence. The kernel density estimation between actual inflow and water demand and adjustable water volume is as follows: Figure 8 As shown. Figure 8 (a) and (b) show that under the measured conditions, there is a significant synchronous change trend between the water inflow, the adjustable water volume and the water demand, and there is an obvious positive correlation.

[0210] After fitting, under the measured conditions, runoff, adjustable water volume and water demand all optimally satisfy the generalized extreme value distribution (GEV). According to the interdependence of the three, Clayton-Copula is selected in the Archimedean family for fitting. Runoff, water demand and adjustable water volume are divided into three types according to their ranges, and the encounter frequency between each group is analyzed. The runoff-water demand-adjustable water volume triple uncertainty scenario is generated based on the uncertainty of the inflow runoff.

[0211] Step S2: Simulating scheduling under multiple uncertainties based on SOP

[0212] The multiple uncertainty scenarios generated in step S1 above were used as the input of the simulation scheduling model, and 200 sets of storage and discharge processes of Huangjinxia and Sanhekou in 15 years were obtained ( Figures 9 to 11 ), energy index change process ( Fig. 9 , Fig.12 )wait.

[0213] Huangjinxia is an important part of the total water transfer, power generation and energy consumption of the Han-Wei River Diversion Project. According to calculations, under multiple uncertain scenarios, the project will transfer an average of 1.347 to 1.517 billion cubic meters of water to the receiving area over the years, of which the Huangjinxia water transfer accounts for about 74% of the total water transfer, and the Sanhekou water transfer accounts for about 26% of the total water transfer; in terms of energy indicators, the project's average total energy consumption over the years is 496 million kW, and the average total power generation over the years is 521 million kW.

[0214] Depend on Fig. 9 It can be seen that: (1) the change in the total water transfer volume is mainly affected by the water transfer volume of Huangjinxia, ​​and when the water transfer volume of Huangjinxia increases, the water transfer volume of Sanhekou decreases. This is because under the SOP strategy, Huangjinxia prioritizes water transfer according to the available water volume, and the completion of Huangjinxia water transfer determines the operating status of Sanhekou; (2) The energy consumption of the two reservoirs is basically changing synchronously, and in terms of values, about 93% comes from Huangjinxia, ​​and only about 7% comes from Sanhekou. This is related to the fact that Huangjinxia accounts for a large proportion of the water transfer volume and Huangjinxia provides less water to Sanhekou; (3) The power generation of the two reservoirs is also basically increasing and decreasing at the same time, and Huangjinxia accounts for a large proportion. When the water transfer volume of Huangjinxia increases, the water volume used for power generation decreases, and the power generation decreases accordingly. At this time, Sanhekou simultaneously bears less water transfer volume and the power generation is relatively small.

[0215] The above characteristics demonstrate the feasibility of changing the operating status of Sanhekou by adjusting the water diversion ratio of Huangjinxia in subsequent hedging strategy research, thereby changing the water diversion and energy indicators of the two reservoirs and the entire project, and optimizing the water diversion process by taking advantage of the abundant hydropower resources in the upper reaches of the Han River and the many years of regulation capacity of Sanhekou.

[0216] Unfavorable water diversion scenarios often mean greater risks, and the geographical location of the project determines that the low-water risk faced by Huangjinxia and Sanhekou is much higher than the flood season risk. In the runoff uncertainty scenario, the low-water series and the normal water series were selected for intensive analysis to explore the operation of the inter-basin water diversion project under unfavorable water diversion conditions. Figure 10 to Figure 13 Displayed in.

[0217] Fig.10 The water transfer situation of the two reservoirs from 2025 to 2039 is shown. It can be seen that under multiple uncertain scenarios, the water transfer to the receiving area mainly depends on Huangjinxia, ​​and the water transfer volume of Huangjinxia is significantly higher than that of Sanhekou. Under the premise that the water demand is basically maintained at the benchmark requirements, the water transfer volume of the two reservoirs to the receiving area does not change much under the low water series and the normal water series, and the total water transfer volume can basically meet the needs of the receiving area, indicating that under the SOP strategy, the scheduling model can provide a smooth operation process, guarantee water transfer to the greatest extent, and reduce the large fluctuations between years caused by unfavorable water inflow.

[0218] Figure 101 shows the water transfer situation of the two reservoirs in each month from 2025 to 2039. It can be seen that: (1) Huangjinxia's water transfer is more concentrated during the period of abundant natural water and the flood season, and the water transfer volume fluctuates greatly during the dry season and non-flood season; (2) Sanhekou's water transfer is mainly concentrated in the non-flood season, and water transfer is even stopped in some periods of abundant water; (3) In the representative month of the non-flood season (March), the concentrated areas of water transfer volume of the two reservoirs show an opposite trend. This is because under the SOP strategy, the water transfer of the project is dominated by Huangjinxia. When Huangjinxia transfers water at a high water level, Sanhekou's joint water transfer reduces or even stops water transfer and turns to water storage, thereby improving the efficiency of water resource utilization.

[0219] As a multi-year regulating reservoir, Sanhekou is mainly used to meet water diversion and ecological water supply. In the simulation scheduling model, only the upper and lower limits of the water level in each period are considered for Sanhekou, and its water level changes can well reflect the water supply and storage process of the project. Fig.12 The monthly water level distribution of Sanhekou is shown. It can be seen that: (1) the monthly operating water level of Sanhekou is between the dead water level and the normal water level, and there is no situation where the constrained water level is exceeded, indicating that the simulation model operation is reasonable; (2) the water level of Sanhekou changes with the inflow flow. When the water inflow is abundant, the water transfer is mainly completed by Huangjinxia, ​​and the surplus water is stored in Sanhekou, so Sanhekou can maintain a high water level; when the water inflow is less, Sanhekou starts joint water transfer and maintains a lower water level (characteristic water level).

[0220] Fig.12 The distribution of power generation and energy consumption in Huangjinxia and Sanhekou in each month is shown. Fig.12 (a) and (b) show that: (1) the monthly power generation in Huangjinxia is basically in the same frequency as the water inflow. The power generation in the flood season is significantly higher than that in the non-flood season, but the difference in power generation between the dry season and the normal water series is not significant; (2) the power generation in the normal water series of Sanhekou is significantly higher than that in the dry season, but the power generation in each month is relatively stable. This shows that although high water level operation has brought higher power generation, it will still cause certain water losses due to the increase in water abandonment and the limitation of project scale. In the dry year group and dry season, the rapid decline in water storage in the reservoir is the main reason for the significant reduction in power generation. Fig.12 (c) and (d) show that: (1) the energy consumption of Huangjinxia during water transfer is much higher than that of Sanhekou; (2) the energy consumption of Huangjinxia varies with the change of inflow runoff; (3) Sanhekou basically consumes energy only during the flood season, and it is basically zero during the non-flood season. This is because the water inflow increases during the flood season, the water transfer volume of Huangjinxia increases, and the energy consumption during water transfer will also increase. Sanhekou mainly stores the surplus water from Huangjinxia during the flood season, and jointly transfers water with Huangjinxia during the non-flood season. Therefore, the energy consumption of Sanhekou is mainly concentrated in the stage when Huangjinxia has surplus water for replenishment. Based on the above characteristics, the energy consumption of the two reservoirs is closely related to their water transfer conditions.

[0221] Step S3: Model optimization space analysis under SOP strategy

[0222] Under the operation scheduling strategy with the minimum total water shortage, the generated three-source uncertainty scenario of runoff-water demand-adjustable water volume is used as the input side of the scheduling model to obtain the 15-year project operation process from 2025 to 2039 (step S2). It can be seen from the water supply and power generation guarantee rate under SOP in Table 3 that the power generation damage changes little under different water inflow conditions, but under the low water sequence, only considering the benefits of the current period will bring a greater depth of water supply damage, which cannot fully cope with the uncertainty scenario and bring potential risks to the project operation.

[0223] Table 3: Water supply and power generation guarantee rate

[0224]

[0225] Fig.13 The comparative relationship between the water transfer volume, energy consumption and power generation output of the simulation scheduling model under the two conditions of using the measured design year data as the model input and using the multi-source uncertainty scenario as the model input is described. It can be seen that although the total water transfer volume has increased after the simulation scheduling under uncertainty conditions, the energy consumption has increased significantly and the power generation has decreased significantly. This is because: (1) Under the condition of SOP strategy to ensure water transfer, the priority water transfer in Huangjinxia will increase its total water transfer volume, and the water transfer in Huangjinxia will consume a lot of energy, thus increasing energy consumption; (2) The priority water transfer in Huangjinxia will lead to water abandonment in Sanhekou, resulting in waste of water resources and reduced power generation.

[0226] In summary, it is difficult to find a feasible solution that satisfies water use, power generation and economic benefits under all constraint scenarios by relying solely on the SOP strategy. In order to enhance the resilience to multiple uncertain scenarios and adverse engineering conditions, the reservoir capacity hedging concept in the AOMU model is introduced to optimize the engineering operation mode, and explore how to adjust the water transfer ratio of the two reservoirs on the basis of ensuring the total water transfer volume of the project, reduce energy consumption, and improve the economic benefits of the project. The specific analysis results are shown in step S4.

[0227] Step S4: AOMU model optimization scheduling results

[0228] In order to further adapt to the actual needs of project operation, continuous low-flow sequence and normal-flow sequence were selected from the generated 200 sets of uncertainty scenarios as the input side of the AOMU optimization model, and multi-objective optimization analysis was performed for the unfavorable project situation where the inflow runoff was low and the water demand was basically unchanged. In order to increase the resilience of the project, the hedging strategy mainly selected objectives 1, 4, and 5 for multi-objective optimization, and the Pareto frontier of the three objective functions of water transfer, energy consumption, and power generation was obtained ( Fig.14 ), and compared it with the simulated scheduling operation results under the same input to verify the feasibility of the optimization model ( Fig.15 ).

[0229] Using the NSGA-II algorithm, AOMU obtained a total of 44 Pareto solution sets in the low water series and a total of 33 Pareto solution sets in the normal water series. Fig.14 It can be seen that in the Pareto frontier solution: (1) the total water transfer volume of the low-flow sequence is higher than 1.3 billion m 3 The total water transfer volume in the normal water sequence is higher than 1.5 billion m 3 , meeting the water diversion demand of the project; (2) It can be seen from the two-dimensional projection plane that when the water diversion volume increases, the energy consumption and power generation increase accordingly; (3) There is a relatively obvious negative correlation between the target solutions of energy consumption and power generation, which is most obvious in the low water series. There are complex contradictions and dependencies between the three objectives of optimal scheduling. From the Pareto solution set, there is an unavoidable conflict between economic benefits and water diversion demand. Abundant water will increase the water diversion volume, which will inevitably lead to an increase in energy consumption. This shows that the AOMU model does not change the characteristics of each objective itself, but seeks to ease the contradiction between objectives.

[0230] Depend on Fig.15 (a) It can be seen that: (1) Under the low-water series and normal-water series, after AOMU optimization, the total energy consumption of the project is significantly reduced, and the power generation is basically unchanged or slightly increased on the basis of the total water transfer volume of the project; (2) After calculation, the total energy consumption of the low-water series project after optimization is reduced by 13.95%, and the total energy consumption of the normal-water series project is reduced by 15.56%, and the power generation is increased by 2.76%. This shows that through hedging, the total energy consumption of the project can be reduced by reducing the proportion of water transfer in Huangjinxia. The water demand of the water receiving area is not fully borne by Huangjinxia, ​​but part of it is reserved for water transfer by Sanhekou, and the surplus water in Huangjinxia is used to supplement the excess storage capacity of Sanhekou.

[0231] Under this model, the offset coefficient is basically stable between 0.8 and 0.9, which is mainly restricted by Target 1. Under unfavorable water diversion conditions, the runoff of Huangjinxia is much greater than that of Sanhekou. To meet the water demand as much as possible, a large amount of water diversion from Huangjinxia is necessary. If the offset coefficient is further reduced to around 0.75 or lower, the water diversion process will be destroyed. This shows that the effect of AOMU on reducing energy consumption is limited and is mainly limited by water demand.

[0232] The above results show that the reliability of the system can be kept within the allowable range by adjusting the water transfer ratio of Huangjinxia and Sanhekou through the AOMU optimization framework and extracting the optimal variables of reservoir capacity hedging. Compared with SOP, AOMU increases the resilience of inter-basin water transfer projects to cope with multiple uncertainties, significantly reduces energy consumption while ensuring water transfer and power generation, solves practical problems in the project operation process and increases economic benefits. Especially under the adverse water transfer conditions of dry season and low water level, AOMU can avoid the reservoir water level reaching the dead water level during the entire operation period by storing part of the flow in the flood season and releasing it in the dry season, compared with the SOP strategy that only considers current benefits; it also adapts to the actual problem of mismatch between the project water supply and storage capacity, and can be used to formulate the best strategy for reservoir operation under adverse conditions. Therefore, in the adaptive dispatch under multiple uncertainties, especially for the adverse water transfer conditions of continuous dry season, the use of reservoir capacity hedging is effective, which provides a scientific and efficient operation strategy for the operation and management of inter-basin water transfer projects, and provides a reasonable solution for the adaptive optimization dispatch of reservoir groups under the interference of multi-source uncertainties.

[0233] Conclusion,Through the above verification process, it can be seen that the adaptive optimization scheduling framework under multiple uncertainty conditions, AOMU, constructed in Example 1 of the present invention, consists of three modules: scenario generation, SOP-based simulation scheduling, and HR-based optimization scheduling.

[0234] By applying AOMU to the actual inter-basin water transfer system, the following conclusions are drawn:

[0235] (1) The selected uncertainty factors of inflow, water demand, and adjustable water volume reflect the key to improving the resilience of the Han-Wei inter-basin water diversion project. The three factors are highly correlated, and the marginal distribution fittings all satisfy the Clayton Copula. The constructed joint scenario can accurately describe the multidimensional dependency relationship between different factors.

[0236] (2) Under the multi-source uncertainty scenario, the simulation results show that the total water transfer volume is maintained at 1.347 to 1.517 billion m 3 , an average increase of 9.2%, which can adapt to the different characteristics of water inflow and regulation conditions in Huangjinxia and Sanhekou to a certain extent.

[0237] (3) AOMU selected three goals for optimization: water transfer volume, power generation, and energy consumption. Through reservoir capacity hedging, the water transfer ratio of Huangjinxia and Sanhekou was optimized. Under unfavorable water transfer conditions, the total energy consumption was reduced by about 13-15% while ensuring that the total water transfer volume did not fluctuate significantly. The power generation increased by 2.76%. AOMU eased the conflict between the goals, and the economic benefits it brought were significantly higher than SOP.

[0238] The adaptive optimization scheduling framework (AOMU) of inter-basin water transfer projects proposed in this invention provides a modeling method that can effectively improve the resilience of projects to deal with uncertainties and adverse water transfer conditions, thereby improving economic benefits while ensuring water transfer.

[0239] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0240] The above-described embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A method for adaptive regulation of inter-basin water diversion projects under uncertainty, characterized in that: include: Step 1: Generate multiple uncertainty scenarios Select uncertainty factors, capture the temporal dependence of uncertainty factors through Markov Monte Carlo and three-dimensional Copula functions, and generate multiple uncertainty scenarios as the input side of the inter-basin water diversion project simulation scheduling model; Step 2: Construct a simulation scheduling model for inter-basin water diversion projects based on SOP According to the principle of water balance and the actual constraints of the project, a water supply sequence is selected to formulate a standard operation strategy to build a simulation scheduling model for inter-basin water diversion projects based on SOP, and analyze the project operation under uncertain scenarios; Step 3: Build an adaptive scheduling framework based on the hedging principle Based on the constructed multiple uncertainty scenarios, the storage capacity hedging strategy is introduced, the energy index is added, an adaptive scheduling framework based on the hedging principle is constructed, and the NSGA-II algorithm is used for multi-objective optimization.

2. The adaptive control method of inter-basin water diversion project under uncertainty according to claim 1, characterized in that: The process of generating multiple uncertainty scenarios in step 1 includes: Step 1-1: Identify uncertainty factors During the operation of the reservoir, the errors generated during the simulation of each uncertainty factor are the risk source; the prediction uncertainty is defined as the error between the measured value and the simulated value; the error probability distribution method generates a probabilistic forecast by assuming the distribution of the errors of each factor, and the prediction sequence under the scenario with simulation errors is obtained as follows: q t =x t +ε t =f(X t )+ε t (1) In formula (1), q t is the output value of the uncertainty factor; x t and f(X t ) is the predicted value and prediction function of the uncertainty factor; ε t Indicates the corresponding error; Step 1-2: Generate multiple uncertainty scenarios based on the uncertainty factors identified in step 1-1; Wherein: the multiple uncertainty scenarios include uncertainty scenarios of independent variables and multi-source uncertainty scenarios with related dependent variables.

3. The adaptive control method of inter-basin water diversion project under uncertainty as claimed in claim 2, characterized in that: The uncertainty scenarios of the independent variables are generated based on MCMC and include: Step 1-2-1-1: Assume that the conditional probability distribution of the current observation is only related to the most recent observation and is independent of all previous observations. The first-order Markov chain under the measured conditions is: In formula (2), Q(q t ) is the measured runoff at time t, T is the sequence length; Step 1-2-1-2: Expand the first-order Markov chain and introduce the implicit state variable z t ∈{1,…,M},z t It follows a first-order Markov process, which is at z t The probability of t-1 Determined by; runoff observation model Q(q t |z t ) depends on the state variables; z t The probability of changing from one state to another is called the state transition probability, and the combination of all state transition probabilities is represented by the state transition matrix A; According to the distribution characteristics of runoff, the "wet and dry" state obeys the Markov random process, and the runoff sequence is affected by the "wet and dry" state, which constitutes the implicit Markov model of runoff; the implicit state variable z of runoff t It is divided into flood season, normal season and low season, so its state transfer matrix Α is: In formula (3), A ij =q(z t =j|z t-1 =i) represents the probability of the state variable changing from i to j, and Α satisfies 0≤A ij ≤1 and During the flood season, t =1; at normal water level, z t =2; during low water period, z t =3; Step 1-2-1-3: Integrate the discretized scenarios in the sample set into runoff uncertainty scenarios that are consistent with actual laws based on the Markov chain.

4. The adaptive control method of inter-basin water diversion project under uncertainty as claimed in claim 2, characterized in that: The multi-source uncertainty scenario with related dependent variables is generated based on a three-dimensional Copula function, including: Step 1-2-2-1: Use the three-dimensional Copula function to establish multiple uncertainty scenarios based on inflow prediction; the joint density function and the corresponding random variables are represented as: P(x i )=u i (i=1,…,m) (6) In formula (4) to formula (6), p(x1),…,p(x m ) and P(x1),…,P(x m ) are the related random variables X1,…,X m The probability density function and cumulative distribution of C(u1,…,u m ) is the Copula function; Step 1-2-2-2: Based on the inflow, adjustable water volume and water demand of historical observations and forecast samples, the Kendall and Spearman rank correlation coefficients are used to evaluate the correlation between characteristic variables.

5. The adaptive control method of inter-basin water diversion project under uncertainty according to claim 1, characterized in that: The inter-basin water diversion project simulation scheduling model described in step 2 includes an input side, a scheduling side and an output side; wherein: The input side is used to reflect the initial storage conditions, final boundary conditions, engineering constraints and generated multiple uncertainty scenarios of actual observation results; The dispatching side focuses on the actual operation of the inter-basin water diversion project, uses the SOP strategy to target the multiple reservoirs in the water source area, and selects the timing for starting water diversion of each reservoir based on the project's storage capacity and natural water supply conditions; The output side is used to output the outflow, water storage, energy consumption and power generation of each reservoir.

6. The adaptive control method of inter-basin water diversion project under uncertainty according to claim 1, characterized in that: The construction process of the adaptive scheduling framework based on the hedging principle described in step 3 includes: Step 3-1: Construct an AOMU model based on storage capacity hedging strategy; Step 3-2: Construct the objective and constraint functions of the AOMU model; Step 3-3: Introduce the NSGA-Ⅱ algorithm to perform multi-objective optimization on the AOMU model.

7. The adaptive control method of inter-basin water diversion project under uncertainty according to claim 6 is characterized in that: The objective function of the AOMU model is: Goal 1: Maximize water transfer In formula (7), t = 1, ..., T and n = 1, ..., N represent the operation period and reservoir respectively, T is the number of operation periods, and N is the number of reservoirs; Q n,t is the total water transfer volume of a reservoir during a certain operation period; Δt is the dispatching period; Goal 2: Minimize water waste In formula (8), It is the water discharge of a reservoir during a certain operation period; Objective 3: Minimize the water shortage index during water transfer In formula (9), is the total water transfer volume of the project during a certain operation period, It is the total water demand of the receiving area during a certain operation period of the project; Goal 4: Maximize power generation In formula (10), It is the discharge flow of a reservoir used for power generation during a certain operation period, and the rest is the same as above; Goal 5: Minimize energy consumption In formula (11), H n,t is the average hydraulic head of a reservoir during a certain operation period; It is the water pumping flow of a pumping station in a reservoir during a certain operation period.

8. The adaptive control method of inter-basin water diversion project under uncertainty according to claim 6 is characterized in that: The constraint function of the AOMU model is: Constraint 1: Reservoir water balance In formula (12), V i,j,s and V i+1,j,s For the s The initial and final storage capacity of the i-th period in the j-th year under the scenario; I i,j,s and Q i,j,s Respectively s The reservoir inflow and outflow in the i-th period in the j-th year under the scenario; V 1,j+1,s For the s The initial storage capacity in the j+1th year under the scenario; V N+1,j,s No. s The storage capacity at the end of the jth year under the scenario, N is the total number of time periods; Constraint 2: Reservoir capacity In formula (13), VU i and VL i are the upper and lower limits of the reservoir capacity respectively; Constraint 3: Maximum flow rate In formula (14), QU i and QL i are the upper and lower limits of the reservoir flow rate respectively; Constraint 4: Maximum output In formula (15), NU i and NL i They are the upper and lower limits of power generation output respectively.

Images