Water-wind-light storage capacity planning method considering high-dimensional uncertainty

Abstract

The invention belongs to the field of storage capacity planning, and provides a water-wind-light storage capacity planning method considering high-dimensional uncertainty. Comprising the steps of time correlation model construction, spatial correlation model construction, high-dimensional uncertainty scene set generation, water-wind-light storage capacity planning model construction, income matrix construction, multi-objective optimization function construction and Pareto leading edge solution. According to the method, a scene set covering high-dimensional uncertainty is generated through a transition probability matrix C-VineCopula method, and spatial-temporal correlation and multi-dimensional random characteristics of natural conditions such as illumination intensity and wind power can be comprehensively captured; by introducing the Utoban line optimization strategy, ideal tradeoff is realized between economical efficiency and safety targets, the efficiency and quality of multi-target optimization solution are improved, and diversified tradeoff decision support is provided for the system.

Description

Technical Field

[0001] The present invention relates to the field of storage capacity planning, and in particular to a method for planning water, wind and solar storage capacity considering high-dimensional uncertainty. Background Art

[0002] With the continuous growth of global energy demand and the increasing prominence of environmental problems, clean and renewable energy has gradually become the mainstream direction of energy development, among which hydropower, wind power and photovoltaic power have received widespread attention. However, the output characteristics of renewable energy show strong uncertainty, which will lead to fluctuations in grid frequency and instability in voltage, thus affecting the safety of the power system. Therefore, it is crucial to rationally plan the system with a high proportion of renewable energy access to improve the economy and reliability of the system. At present, a lot of research has been carried out on the system planning with a high proportion of renewable energy access. The existing technology uses cluster indicators and voltage fluctuation indicators to formulate the installation plan of the storage system in the photovoltaic storage system, and uses cooperative games to determine the optimal storage capacity; by establishing an energy storage optimization model, the impact of the time variation of photovoltaics and loads on the power grid is analyzed, and a capacity configuration and optimization method for the storage system under high photovoltaic penetration is proposed; an integrated demand response program is designed, the potential interactive capabilities of electricity, gas, heat, cold and electric vehicle flexible loads are explored, and an optimization scheduling method for the wind power-photovoltaic-gas-electric vehicle community integrated energy system is proposed; by constructing a digital twin model of the integrated energy system, the prediction ability of the digital twin is used to improve the coordination between various energy converters, thereby improving energy efficiency, reducing costs and reducing carbon emissions; a system planning method based on game theory is established, so that the system not only meets the needs of cold, heat and electricity loads, but also achieves the optimal system economy. Although the above research has made some progress in the system planning of high-proportion renewable energy access, it has failed to fully consider the uncertainty of renewable energy. The existing technology describes the uncertainty of wind and solar power output through the maximum uncertainty set, and proposes an optimization scheduling method that takes into account uncertainty and equipment coupling; considering wind and solar power output and uncertainty, the interval linear programming method is used to deal with the uncertainty problem, and a two-layer optimization configuration method is proposed; considering the uncertainty of wind and solar power output and electrical coupling equipment, considering economic costs and environmental protection and emission reduction, an economic scheduling model for the integrated energy system is established; considering the impact of photovoltaic output uncertainty on the system economy, it is proposed to use the vector machine method to predict photovoltaic output and propose an economic operation strategy.

[0003] However, the above studies have not fully considered the inherent temporal correlation of various meteorological conditions with seasonal changes in systems with a high proportion of renewable energy access, as well as the high-dimensional uncertainty caused by the spatial correlation of meteorological conditions in the same area. The traditional empirical planning method estimates the equipment capacity based on past experience. It is simple to operate and has low initial cost, but lacks consideration of uncertainty, which can easily lead to unreasonable capacity configuration and high long-term operating costs. The deterministic optimization planning method is based on deterministic data and does not consider uncertainty. The calculation is relatively simple and can be implemented when data is limited, but the actual adaptability is poor. It cannot cope with the fluctuation of wind and solar resources and cannot balance economy and safety. The planning method considering partial uncertainty only considers some uncertainty factors. The computational complexity and data requirements are between the two. Although it has certain adaptability, it does not fully consider high-dimensional uncertainty, the planning is not accurate enough, and the system synergy cannot be fully utilized. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a water, wind, solar and storage capacity planning method that takes into account high-dimensional uncertainty, so as to solve the problems that traditional methods do not deeply deal with high-dimensional dependencies, lack of depth in characterizing complex dependencies and dynamic characteristics, lack of systematic scheduling optimization strategies, and ignore the multi-objective balance in optimal scheduling.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for planning water, wind and solar storage capacity considering high-dimensional uncertainty, including:

[0007] According to the monthly data fluctuations, the historical data of water, wind and light are divided into several state intervals, and a transition probability matrix is ​​constructed according to the transition frequency of the variables in different state intervals in each month. All the transition probability matrices are integrated to obtain a time correlation model;

[0008] Using non-parametric kernel density estimation method to fit the marginal distribution of each variable in the water, wind and light historical data, to obtain the cumulative distribution function;

[0009] Calculate the Kendall correlation coefficient of each variable in the water, scenery and historical data, and select the variable with the largest Kendall correlation coefficient as the root node of the vine structure, describe the correlation between the root node and the remaining variables in the water, scenery and historical data through a binary Copula function to construct the first layer of the vine structure, and recursively describe the interdependence between the remaining variables layer by layer to obtain a multidimensional joint probability density function;

[0010] The cumulative distribution function, the multidimensional joint probability density function and the mixed Copula function are integrated to obtain a spatial correlation model;

[0011] Generate an initial scene set that considers time correlation according to the time correlation model and the space correlation model, generate a first random variable that satisfies uniform distribution through cumulative probability from the initial scene corresponding to the variable with the largest absolute value of the sum of the Kendall correlation coefficients in the initial scene set, and use the first random variable as the first sample point;

[0012] Generate a second random variable and a third random variable that satisfy a uniform distribution by cumulative probability, calculate a second sample point according to the first sample point and the second random variable, and calculate a third sample point according to the second sample point and the third random variable;

[0013] Mapping the first sample point, the second sample point, and the third sample point to an actual distribution using an inverse transform sampling method to obtain a high-dimensional uncertainty scene set;

[0014] Constructing a water, wind, solar, and storage capacity planning model; the water, wind, solar, and storage capacity planning model includes: an upper capacity configuration model and a lower capacity configuration model; the upper capacity configuration model includes: an upper objective function, a device sub-model, and a capacity constraint; the lower capacity configuration model includes: a lower objective function, a constraint condition, and a power balance constraint;

[0015] Converting the upper layer objective function and the lower layer objective function into a standard form, constructing a profit matrix according to the converted upper layer objective function and the lower layer objective function, and normalizing the profit matrix;

[0016] Extracting the Utopia line according to the normalized benefit matrix, calculating the distance between the target point in the standardized space and the Utopia line to obtain the Utopia line distance, and constructing a multi-objective optimization function according to the Utopia line distance;

[0017] The multidimensional joint probability density function is optimally solved according to the high-dimensional uncertainty scenario set and the water-wind-solar-storage capacity planning model to obtain the Pareto frontier, and the Pareto frontier is updated to the target system control strategy.

[0018] Preferably, the transition probability matrix is:

[0019]

[0020] in, P w is the transition probability matrix; s t+1 and t They represent the lighting conditions at month t+1 and month t respectively; f(i,j) represents the frequency of transition from state i to state j; θ is the total number of state intervals.

[0021] Preferably, the calculation formula of the Kendall correlation coefficient is:

[0022]

[0023] in, τ l,m is the Kendall correlation coefficient; P(·) represents the probability density function; L represents the total number of variables in the historical data of water, wind and light; X l and X m They respectively represent the lth and mth variables in the water, scenery and light historical data.

[0024] Preferably, the calculation formula of the second sample point is:

[0025]

[0026] Among them, Y2 is the second random variable; Z1 is the first sample point; Z2 is the second sample point.

[0027] Preferably, the upper layer objective function includes:

[0028]

[0029] as well as

[0030]

[0031] Where F is the minimized total annual cost; is the initial investment cost of the equipment; The subsequent operation and maintenance costs of the equipment; is the revenue generated by selling electricity to the upper grid; k is the scenario index; w k is the probability of scene k; E wi 、E pv and E eh are the annual value coefficients of wind power, photovoltaic power and storage equipment respectively; R wi , R pv and R eh are the unit capacity investment costs of each equipment respectively; and are the total investment capacity of each device in scenario k; C op,wa , C op,pv and C op,eh are the unit operation and maintenance costs of each device; and are the power generation power of each device respectively; and They are the charging and discharging power of the storage system; is the unit electricity price at time t; The system sells electricity to the upper grid.

[0032] Preferably, the device sub-model includes:

[0033]

[0034] as well as

[0035]

[0036] in, is the photovoltaic power generation; is the power generation of hydropower; η pv is the efficiency of the photovoltaic panel; k pv is the photovoltaic panel conversion coefficient; is the solar radiation intensity; k pv Photovoltaic panel temperature coefficient; T t is the ambient temperature at time t; T ref is the reference temperature of the photovoltaic panel; k wa is the power generation coefficient of the hydropower station; Water flow for power generation; is the power generation head height; is the runoff; S i,t-1 is the amount of water discarded; SOC t+1 and SOC t Corresponding to the energy states at time t+1 and time t respectively; are charging power and discharging power respectively; η charge , η discharge are the charging coefficient and the discharging coefficient respectively; They are charging sign and discharging sign respectively.

[0037] Preferably, the profit matrix is:

[0038]

[0039] Wherein, φ is the profit matrix; f1(x) and f2(x) represent the transformed upper objective function and the lower objective function respectively; is the decision variable corresponding to the optimal solution when only optimizing f1(x); is the decision variable corresponding to the optimal solution when only optimizing f2(x).

[0040] Preferably, the multi-objective optimization function is:

[0041]

[0042] Where β=[β1,β2] T;0≤β1,β2≤1;β1+β2=1; β represents the coordinates of the target point; A normal unit vector representing the utopia line; is the normalized objective function vector; g(x) and h(x) are the constraints of the optimization problem.

[0043] Preferably, the lower layer objective function includes:

[0044] and

[0045]

[0046] in, is the expected value of electricity sold to the superior power grid.

[0047] Preferably, the power balance constraint is:

[0048]

[0049] The present invention discloses the following technical effects:

[0050] The present invention provides a method for planning water, wind, and solar storage capacity considering high-dimensional uncertainty. A scenario set covering high-dimensional uncertainty is generated through a transfer probability matrix C-VineCopula method, which solves the problem that traditional methods do not deeply deal with high-dimensional dependencies and lack the depth of characterization of complex dependencies and dynamic characteristics, and realizes the capture of the spatiotemporal correlation and multi-dimensional random characteristics of natural conditions such as light intensity and wind power. By introducing a Utopia line optimization strategy, the problem that traditional methods lack a systematic scheduling optimization strategy and ignore the multi-objective balance in optimization scheduling is solved, and an ideal trade-off between economic and safety goals and diversified trade-off decision support are achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] 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 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 based on these drawings without paying creative labor.

[0052] Figure 1 A schematic diagram of a water, wind, solar, and storage capacity planning process considering high-dimensional uncertainty provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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.

[0054] The purpose of the present invention is to provide a water, wind, and solar storage capacity planning method that takes high-dimensional uncertainty into account, so as to solve the problems that traditional methods do not deeply deal with high-dimensional dependencies, lack the depth of characterization of complex dependencies and dynamic characteristics, lack of systematic scheduling optimization strategies, and ignore the multi-objective balance in optimal scheduling.

[0055] 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.

[0056] Figure 1 A schematic diagram of a water, wind, solar, and storage capacity planning process considering high-dimensional uncertainty provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the present invention provides a method for planning water, wind and solar storage capacity considering high-dimensional uncertainty, including:

[0057] Step 100: According to the monthly data fluctuations, the historical data of water, wind and light are divided into several state intervals, and a transition probability matrix is ​​constructed according to the transition frequency of the variables in different state intervals in each month. All transition probability matrices are integrated to obtain a time correlation model;

[0058] Step 200: Use a non-parametric kernel density estimation method to fit the marginal distribution of each variable in the historical data of water, wind and light to obtain a cumulative distribution function;

[0059] Step 300: Calculate the Kendall correlation coefficient of each variable in the historical data of water, scenery and light, and select the variable with the largest Kendall correlation coefficient as the root node of the vine structure. Use the binary Copula function to describe the correlation between the root node and the remaining variables in the historical data of water, scenery and light to construct the first layer of the vine structure. Recursively describe the interdependence between the remaining variables layer by layer to obtain a multidimensional joint probability density function.

[0060] Step 400: Fusing the cumulative distribution function, the multidimensional joint probability density function and the mixed Copula function to obtain a spatial correlation model;

[0061] Step 500: Generate an initial scene set that takes time correlation into consideration according to the time correlation model and the space correlation model, generate a first random variable that satisfies uniform distribution through cumulative probability from the initial scene corresponding to the variable with the largest absolute value of the sum of Kendall correlation coefficients in the initial scene set, and use the first random variable as the first sample point;

[0062] Step 600: Generate a second random variable and a third random variable that satisfy a uniform distribution by cumulative probability, calculate a second sample point according to the first sample point and the second random variable, and calculate a third sample point according to the second sample point and the third random variable;

[0063] Step 700: Mapping the first sample point, the second sample point, and the third sample point to the actual distribution using an inverse transformation sampling method to obtain a high-dimensional uncertainty scene set;

[0064] Step 800: construct a water, wind, solar, and storage capacity planning model; the water, wind, solar, and storage capacity planning model includes: an upper capacity configuration model and a lower capacity configuration model; the upper capacity configuration model includes: an upper objective function, a device sub-model, and a capacity constraint; the lower capacity configuration model includes: a lower objective function, a constraint condition, and a power balance constraint;

[0065] Step 900: converting the upper layer objective function and the lower layer objective function into a standard form, constructing a profit matrix according to the converted upper layer objective function and the lower layer objective function, and normalizing the profit matrix;

[0066] Step 1000: extracting the utopia line according to the normalized benefit matrix, calculating the distance between the target point and the utopia line in the standardized space, obtaining the utopia line distance, and constructing a multi-objective optimization function according to the utopia line distance;

[0067] Step 1100: Optimally solve the multidimensional joint probability density function according to the high-dimensional uncertainty scenario set and the water, wind, solar and storage capacity planning model to obtain the Pareto frontier, and update the Pareto frontier to the target system control strategy.

[0068] Specifically, the transition probability matrix is:

[0069]

[0070] in, P w is the transition probability matrix; s t+1 and t They represent the lighting conditions at month t+1 and month t respectively; f(i,j) represents the frequency of transition from state i to state j; θ is the total number of state intervals.

[0071] Furthermore, the calculation formula of Kendall correlation coefficient is:

[0072]

[0073] in, τ l,m is the Kendall correlation coefficient; P(·) represents the probability density function; L represents the total number of variables in the historical data of water, wind and light; X l and X m They represent the lth and mth variables in the historical data of water and scenery respectively.

[0074] Specifically, the calculation formula for the second sample point is:

[0075]

[0076] Among them, Y2 is the second random variable; Z1 is the first sample point; Z2 is the second sample point.

[0077] Furthermore, the upper objective function includes:

[0078]

[0079] as well as

[0080]

[0081] Where F is the minimized total annual cost; is the initial investment cost of the equipment; The subsequent operation and maintenance costs of the equipment; is the revenue generated by selling electricity to the upper grid; k is the scenario index; w k is the probability of scene k; E wi 、E pv and E eh are the annual value coefficients of wind power, photovoltaic power and storage equipment respectively; R wi , R pv and R eh are the unit capacity investment costs of each equipment respectively; and are the total investment capacity of each device in scenario k; C op,wa , C op,pv and C op,eh are the unit operation and maintenance costs of each device; and are the power generation power of each device respectively; and They are the charging and discharging power of the storage system; is the unit electricity price at time t; The system sells electricity to the upper grid.

[0082] Specifically, the device sub-model includes:

[0083]

[0084] as well as

[0085]

[0086] in, is the photovoltaic power generation; is the power generation of hydropower; η pv is the efficiency of the photovoltaic panel; k pv is the photovoltaic panel conversion coefficient; is the solar radiation intensity; k pv Photovoltaic panel temperature coefficient; T t is the ambient temperature at time t; T ref is the reference temperature of the photovoltaic panel; k wa is the power generation coefficient of the hydropower station; Water flow for power generation; is the power generation head height; is the runoff; S i,t-1 is the amount of water discarded; SOC t+1 and SOC t Corresponding to the energy states at time t+1 and time t respectively; are charging power and discharging power respectively; η charge , η discharge are the charging coefficient and the discharging coefficient respectively; They are charging sign and discharging sign respectively.

[0087] Furthermore, the profit matrix is:

[0088]

[0089] Among them, φ is the profit matrix; f1(x) and f2(x) represent the upper objective function and the lower objective function after transformation respectively; is the decision variable corresponding to the optimal solution when only optimizing f1(x); is the decision variable corresponding to the optimal solution when only optimizing f2(x).

[0090] Specifically, the multi-objective optimization function is:

[0091]

[0092] Where β=[β1,β2] T ;0≤β1,β2≤1;β1+β2=1; β represents the coordinates of the target point; represents the normal unit vector of the utopia line; is the normalized objective function vector; g(x) and h(x) are the constraints of the optimization problem.

[0093] Furthermore, the lower layer objective function includes:

[0094] and

[0095]

[0096] in, is the expected value of electricity sold to the superior power grid.

[0097] Preferably, the power balance constraint is:

[0098]

[0099] Specifically, in terms of temporal correlation modeling, the computer device takes light intensity as an example (other natural conditions are similar), divides the state interval according to the data fluctuations in different months, and constructs a transition probability matrix by counting the transition frequencies of variables in different states in each month, and then calculates the transition frequency matrix. Next, when modeling spatial correlation, the computer uses a model based on C-Vine Copula to first fit the marginal distribution of each variable to obtain a cumulative distribution function, calculate the Kendall correlation coefficient between variables, select the variable with the largest correlation as the root node of the vine structure, and construct the first layer of the vine structure through the binary Copula function, and then recursively describe the dependency of the remaining variables layer by layer, use the mixed Copula function and estimate its weight and dependent parameters. Then, in the scene generation stage, the computer first generates an initial scene set that takes into account temporal correlation, determines the state and light intensity of each month through random number judgment, and loops multiple times to obtain the initial scene set, then selects the variable with the largest absolute value of the sum of correlation coefficients as the starting point, uses cumulative probability, random numbers and related calculations to obtain uniformly distributed random variables, obtains sample points through specific calculations, and then uses the inverse transformation sampling method to generate the final scene set. In the water, wind, solar and storage capacity planning model, the computer optimization goal of the upper capacity configuration model is to minimize the total annual cost, which involves the calculation of various equipment costs and benefits, and is subject to the equipment model (including the power generation and charging and discharging power-related rules of photovoltaic, wind power, hydropower, and storage systems) and capacity constraints (the capacity of each device is limited by site factors); the lower collaborative optimization model takes typical scenarios as input. Based on the capacity determined by the upper layer, the computer adjusts the reservoir water level to ensure the optimal system economy and safety, subject to equipment power generation constraints (each device has an upper limit on power generation) and power balance constraints (total power generation meets power consumption). Finally, when solving the multi-objective optimization problem, the computer first converts the dual-objective optimization problem into a standard form, constructs a benefit matrix to describe the optimal solution, normalizes the objective function and benefit matrix, connects the anchor points to form a utopian line, introduces the multi-objective problem of distance conversion between the point and the utopian line, and obtains the non-dominated solution through optimization to form the Pareto frontier to balance the system economy and safety.

[0100] Furthermore, time correlation modeling is performed. Since the modeling methods of light intensity, wind power, temperature, and runoff are the same, light intensity is used as an example. According to the data fluctuations in different months, the historical data of photovoltaics are divided into several state intervals. The variables of each month can be divided into θ states, with a total of θ×12 state intervals. The transition frequency of variables in each month between different states is counted to construct the transition probability matrix P w :

[0101]

[0102] Among them, each row sums to 1. The transition probability matrix can be used to calculate the transition frequency matrix Q w .Qw is an s×s matrix, where the elements

[0103] Specifically, spatial correlation modeling. In order to accurately capture the spatial correlation in the water, wind, solar and storage systems, a model building method based on C-Vine Copula is used. The C-Vine Copula model can effectively handle the nonlinear dependencies between multidimensional variables. For each variable X l First, we need to use the nonparametric kernel density estimation method to fit the marginal distribution of each variable to obtain the cumulative distribution function F(X l ), where l∈{1,2,...,L}, can provide input for the subsequent Copula model. Secondly, in order to determine the correlation between different variables, the Kendall correlation coefficient between the variables needs to be calculated. For each pair of variables X l and X m , its Kendall correlation coefficient τ l,m The calculation formula is:

[0104]

[0105] According to the calculated Kendall correlation coefficient, the variable with the greatest correlation with other variables is selected as the root node of the vine structure, denoted as X1. Then, the first layer of the vine structure is constructed by describing the correlation between the root node and other variables through the binary Copula function. After the first layer is constructed, the interdependence between the remaining variables is recursively described layer by layer. Through this hierarchical method, the complex dependency between multidimensional variables can be recursively described. The multidimensional joint probability density function f(X1,X2,...,X L ) can be decomposed by the following formula:

[0106]

[0107] Among them, cm,m+1|1,...,m-1 is the variable X under the condition that the first m-1 variables are known. m and X m+1 The Copula function between m |X1,...,X m-1 ) is a known X1,...,X m-1 Next, the variable X m The distribution function of is:

[0108]

[0109] Among them, Y m is the mth element of vector Y; represents vector Y minus Ym Since a single Copula function is not sufficient to accurately describe the dependency between variables, a mixed Copula function is used:

[0110]

[0111] in, is the oth Copula function, λ o is the corresponding weight, and is the dependent parameter of the Copula function. Use the expectation maximization algorithm to calculate the weight parameter λ o and dependent parameters Make an estimate.

[0112] Preferably, scene generation. To generate a high-dimensional uncertainty scene that takes into account spatiotemporal correlation, it is first necessary to generate an initial scene set that takes into account temporal correlation. Assume that the light state of the current month is The light intensity is P t v . Generate a random number a that follows a uniform distribution t ∈[0,1], determine the cumulative frequency matrix of the current month The data of the row is the same as a t The size of the light condition a in the next month is determined t+1 After determining the state, generate a random number b t ∈[0,1], and assume that state a t+1 The corresponding light intensity interval is [β min ,β max ], then β t+1 =β min +b t (β max -β min ). Repeat the above steps until t = 12, that is, the generation of a scene sequence is completed. Loop B times to obtain the initial scene set S containing B scenes:

[0113]

[0114] Where N is the length of the light intensity sequence.

[0115] Furthermore, the high-dimensional uncertainty scenario generation process considering spatiotemporal correlation is as follows: First, calculate the Kendall correlation coefficient of each variable with other variables, and select the variable with the largest absolute value of the sum of the correlation coefficients as the starting point for scenario generation. Generate a uniformly distributed random variable Y1, Y1∈[0,1] from the corresponding initial scenario through cumulative probability. Let the three sets of variables to be determined be Z1, Z2, and Z3, and let Z1=Y1. Generate two random numbers Y2 and Y3 that satisfy uniform distribution:

[0116]

[0117] Among them, C2 and Z1 are known quantities, and the sample points of Z2 are obtained from them. Similarly, Z3 is obtained. After obtaining the three groups of sample points Z1, Z2 and Z3, the inverse transform sampling method is used to map these uniformly distributed sample points to the actual distribution of each variable to generate the final scene set containing each natural condition.

[0118] Specifically, the hydropower, wind, photovoltaic and storage capacity planning model. Considering the uncertainty of renewable energy, the existing hydropower stations are planned to be equipped with wind power, photovoltaic and storage systems, and a two-layer optimization model for hydropower, wind, photovoltaic and storage capacity planning is proposed. The upper model takes economic optimization as the objective function and determines the optimal capacity of wind power, photovoltaic and storage systems based on a typical scenario set. The lower model is based on the optimal capacity determined by the upper layer, with the goal of optimal economy and safety, and performs collaborative optimization scheduling.

[0119] Furthermore, the upper capacity configuration model, the objective function: The upper optimization goal is to minimize the total annual cost F, which includes the initial investment cost of the equipment Later operation and maintenance costs At the same time, the income generated by the system selling electricity to the upper power grid must also be deducted The objective function is as follows:

[0120]

[0121] Constraints:

[0122] 1) Equipment model:

[0123] Photovoltaic power generation It is mainly affected by solar radiation intensity and ambient temperature. The specific formula is as follows:

[0124]

[0125] Wind power generation power P k,t wi Mainly affected by wind speed, the specific formula is:

[0126]

[0127] Among them, v k,t is the wind speed; V c is the cut-in wind speed; V f is the cut-out wind speed; V r is the rated wind speed; C wi is the performance coefficient of the fan; ρ is the air density; A wi is the area swept by the rotor;

[0128] The power generation of hydropower mainly depends on the water flow and head height. The specific formula is:

[0129]

[0130] The charging and discharging behavior of the storage system needs to follow the law of conservation of energy. The energy state of the storage is a dynamic evolution process, which is affected by the charging power and the discharging power. The formula is as follows:

[0131]

[0132] When charging, When discharging,

[0133] 2) Capacity constraints:

[0134] The decision variables of the upper optimization are the capacity of each device, which is limited by factors such as the site and should satisfy the following inequality:

[0135]

[0136] in, and They are the upper limits of the configuration capacity of wind power, photovoltaic power and storage respectively.

[0137] Specifically, the lower-level collaborative optimization model has the following objective function:

[0138] The lower-level model takes typical scenarios of water, wind and light as input, and adjusts the reservoir water level based on the optimal capacity of each device determined by the upper-level capacity configuration model to ensure optimal system economy and safety.

[0139]

[0140] Constraints: To ensure the safety of equipment operation, the power generation of each device must meet the following constraints:

[0141]

[0142] in, and They are the upper limits of wind power, hydropower, photovoltaic power, and storage charging and discharging power respectively.

[0143] Power balance constraint: The total power generation of the system at any time must meet the power consumption at the current time. The specific constraints are as follows:

[0144]

[0145] Furthermore, a multi-objective optimization problem solving method is used. A multi-objective optimization model is used to balance the economy and safety of the water, wind, and solar storage system. To achieve this, the dual-objective optimization problem is first transformed into a standard form:

[0146] min{f1(x),f2(x)},

[0147] st.g(x)≤0,h(x)=0,x∈R

[0148] Among them, f1(x) and f2(x) represent economic goals and safety goals respectively, which need to be optimized simultaneously. The constraints g(x)≤0 and h(x)=0 define the solution space R.

[0149] In order to analyze the trade-offs between objective functions, a profit matrix needs to be constructed to describe the optimal solution under different optimization objectives. The profit matrix φ is defined as follows:

[0150]

[0151] In order to eliminate the scale differences between different objectives, the objective function and the benefit matrix need to be normalized. The normalization formula is as follows:

[0152]

[0153] The normalized profit matrix can be expressed as:

[0154]

[0155] in, and is the anchor point obtained in the profit matrix. By connecting these two anchor points, a utopia line is formed. This line represents the ideal trade-off between the two objective functions. In order to measure the quality of the solution, the distance D between the point P (β1, β2) in the standardized space and the utopia line is introduced, that is, the normal vector perpendicular to the utopia line through the point P is drawn. Therefore, the multi-objective problem can be converted into the following form:

[0156] max D

[0157]

[0158] {x∈R|g(x)≤0,h(x)=0}

[0159] Through different The optimization is performed under the combination to obtain multiple different non-dominated solutions, that is, these solutions are used to form an approximate Pareto frontier.

[0160] The beneficial effects of the present invention are as follows:

[0161] The present invention generates a scene set covering high-dimensional uncertainty through the transfer probability matrix C-VineCopula method, which can comprehensively capture the spatiotemporal correlation and multi-dimensional random characteristics of natural conditions such as light intensity and wind power; by introducing the Utopia line optimization strategy, an ideal trade-off between economic and safety goals is achieved, the efficiency and quality of multi-objective optimization solutions are improved, and diversified trade-off decision support is provided for the system.

[0162] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0163] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for planning water, wind and solar storage capacity considering high-dimensional uncertainty, characterized in that: include: According to the monthly data fluctuations, the historical data of water, wind and light are divided into several state intervals, and a transition probability matrix is ​​constructed according to the transition frequency of the variables in different state intervals in each month. All the transition probability matrices are integrated to obtain a time correlation model; Using non-parametric kernel density estimation method to fit the marginal distribution of each variable in the water, wind and light historical data, to obtain the cumulative distribution function; Calculate the Kendall correlation coefficient of each variable in the water, scenery and historical data, and select the variable with the largest Kendall correlation coefficient as the root node of the vine structure, describe the correlation between the root node and the remaining variables in the water, scenery and historical data through a binary Copula function to construct the first layer of the vine structure, and recursively describe the interdependence between the remaining variables layer by layer to obtain a multidimensional joint probability density function; The cumulative distribution function, the multidimensional joint probability density function and the mixed Copula function are integrated to obtain a spatial correlation model; Generate an initial scene set that considers time correlation according to the time correlation model and the space correlation model, generate a first random variable that satisfies uniform distribution through cumulative probability from the initial scene corresponding to the variable with the largest absolute value of the sum of the Kendall correlation coefficients in the initial scene set, and use the first random variable as the first sample point; Generate a second random variable and a third random variable that satisfy a uniform distribution by cumulative probability, calculate a second sample point according to the first sample point and the second random variable, and calculate a third sample point according to the second sample point and the third random variable; Mapping the first sample point, the second sample point, and the third sample point to an actual distribution using an inverse transform sampling method to obtain a high-dimensional uncertainty scene set; Construct a water, wind and solar storage capacity planning model; The water, wind, and solar storage capacity planning model includes: an upper capacity configuration model and a lower capacity configuration model; the upper capacity configuration model includes: an upper objective function, a device sub-model, and a capacity constraint; the lower capacity configuration model includes: a lower objective function, a constraint condition, and a power balance constraint; Converting the upper layer objective function and the lower layer objective function into a standard form, constructing a profit matrix according to the converted upper layer objective function and the lower layer objective function, and normalizing the profit matrix; Extracting the Utopia line according to the normalized benefit matrix, calculating the distance between the target point in the standardized space and the Utopia line to obtain the Utopia line distance, and constructing a multi-objective optimization function according to the Utopia line distance; The multidimensional joint probability density function is optimally solved according to the high-dimensional uncertainty scenario set and the water-wind-solar-storage capacity planning model to obtain the Pareto frontier, and the Pareto frontier is updated to the target system control strategy.

2. According to claim 1, a method for planning water, wind and solar storage capacity considering high-dimensional uncertainty is characterized in that: The transition probability matrix is: in, P w is the transition probability matrix; s t+1 and t They represent the lighting conditions at month t+1 and month t respectively; f(i,j) represents the frequency of transition from state i to state j; θ is the total number of state intervals.

3. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 1 is characterized in that: The calculation formula of the Kendall correlation coefficient is: in, τ l,m is the Kendall correlation coefficient; P(·) represents the probability density function; L represents the total number of variables in the historical data of water, wind and light; X l and X m They respectively represent the lth and mth variables in the water, scenery and light historical data.

4. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 1 is characterized in that: The calculation formula of the second sample point is: Among them, Y2 is the second random variable; Z1 is the first sample point; Z2 is the second sample point.

5. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 1 is characterized in that: The upper layer objective function includes: as well as Where F is the minimized total annual cost; is the initial investment cost of the equipment; The subsequent operation and maintenance costs of the equipment; is the revenue generated by selling electricity to the upper grid; k is the scenario index; w k is the probability of scene k; E wi 、E pv and E eh are the annual value coefficients of wind power, photovoltaic power and storage equipment respectively; R wi , R pv and R eh are the unit capacity investment costs of each equipment respectively; and are the total investment capacity of each device in scenario k; C op,wa , C op,pv and C op,eh are the unit operation and maintenance costs of each device; and are the power generation power of each device respectively; and They are the charging and discharging power of the storage system; is the unit electricity price at time t; The system sells electricity to the upper grid.

6. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 1 is characterized in that: The device sub-model includes: as well as in, is the photovoltaic power generation power; is the power generation of hydropower; η pv is the efficiency of the photovoltaic panel; k pv is the photovoltaic panel conversion coefficient; is the solar radiation intensity; k pv Photovoltaic panel temperature coefficient; T t is the ambient temperature at time t; T ref is the reference temperature of the photovoltaic panel; k wa is the power generation coefficient of the hydropower station; Water flow for power generation; is the power generation head height; is the runoff; S i,t-1 is the amount of water discarded; SOC t+1 and SOC t Corresponding to the energy states at time t+1 and time t respectively; are charging power and discharging power respectively; η charge , η discharge are the charging coefficient and the discharging coefficient respectively; They are charging sign and discharging sign respectively.

7. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 1 is characterized in that: The profit matrix is: Wherein, φ is the profit matrix; f1(x) and f2(x) represent the transformed upper objective function and the lower objective function respectively; is the decision variable corresponding to the optimal solution when only optimizing f1(x); is the decision variable corresponding to the optimal solution when only optimizing f2(x).

8. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 1 is characterized in that: The multi-objective optimization function is: Where β=[β1,β2] T ;0≤β1,β2≤1;β1+β2=1; β represents the coordinates of the target point; A normal unit vector representing the utopia line; is the normalized objective function vector; g(x) and h(x) are the constraints of the optimization problem.

9. The method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to claim 5 is characterized in that: The lower layer objective function includes: and in, is the expected value of electricity sold to the superior power grid.

10. A method for planning water, wind and solar storage capacity considering high-dimensional uncertainty according to any one of claims 5 and 6, characterized in that: The power balance constraint is:

Images