Electricity-hydrogen external characteristic high-efficiency probability characterization method for virtual power plant containing electricity to hydrogen

Abstract

The invention relates to an electricity-hydrogen external characteristic high-efficiency probability characterization method for a virtual power plant containing electricity to hydrogen, which comprises the following steps of: establishing a hydrogen production plant model considering dynamic electrolysis efficiency and a virtual power plant resource aggregation optimization model, and determining the limit exchange power of a gateway node of the virtual power plant; based on a multi-parameter planning method, analyzing and expressing the electricity-hydrogen external characteristics of the electricity-containing hydrogen conversion virtual power plant; solving an electricity-hydrogen external characteristic cumulative distribution function of the electricity-to-hydrogen virtual power plant; collecting day-ahead scheduling data and system equipment parameters, converting the obtained electricity-hydrogen external characteristic cumulative distribution function of the electricity-to-hydrogen virtual power plant into a schedulable domain constraint condition by utilizing opportunity constraint, and establishing a multi-virtual power plant-electricity-hydrogen comprehensive energy system scheduling model taking the minimum cost obtained by an operator as a target function; and solving based on the multi-virtual power plant-electricity-hydrogen comprehensive energy system scheduling model. According to the method, stable operation of a virtual power plant (VPP) and electricity-to-hydrogen (P2H) combined system and accurate and efficient characterization of an electricity-hydrogen schedulable domain are realized.

Description

Technical Field

[0001] The present invention relates to the field of optimization scheduling of electricity-to-hydrogen virtual power plants, and in particular to a method for efficiently probabilistically characterizing the external characteristics of electricity-hydrogen in electricity-to-hydrogen virtual power plants. Background Art

[0002] In order to achieve energy conservation and emission reduction and alleviate the oil crisis, low-carbon development has become an international consensus. Many countries have set goals and need to build new clean energy systems. Hydrogen energy is a recognized clean energy carrier with the characteristics of high energy density, high conversion efficiency, no pollution, and zero carbon emissions. Therefore, it will play an important role in the new energy system. With the increase in the proportion of renewable energy in the energy structure, clean electricity can produce "green hydrogen" through the electrolysis of water, which can replace "blue hydrogen" or "grey hydrogen". Therefore, clean energy hydrogen production is an important development direction for the construction of new energy systems in the future. Virtual power plant (VPP) is to effectively integrate multiple distributed power sources, energy storage devices, loads and other distributed units into a controllable "independent power plant" through advanced communication technology and optimal control, and coordinate the contradiction between smart grid and distributed energy. It effectively solves the control and dispatching problems of distributed power grids by means of scale, and is a system based on customer needs. Taking into account the use of renewable energy, the excess new energy is converted into a storable and dispatchable energy system. By applying power-to-hydrogen (P2H) technology in virtual power plants (VPPs), excess renewable energy output can be converted into hydrogen and stored through a series of operations such as water electrolysis. Therefore, the optimized scheduling of virtual power plants (VPPs) by power-to-hydrogen (P2H) technology is the key to dealing with the instability of renewable energy output in virtual power plants (VPPs).

[0003] Therefore, there are certain advantages in organically combining virtual power plants (VPP) with power-to-hydrogen (P2H) to form a VPP-P2H combined system. First, using renewable energy to produce hydrogen can promote the consumption of renewable energy, reduce the cost of producing hydrogen, and facilitate the large-scale application of hydrogen energy. Second, both renewable energy and hydrogen energy are clean energy, and using renewable energy to produce hydrogen can achieve the conversion between multiple energy sources and low-carbon alternatives.

[0004] In current research, power-to-hydrogen (P2H) is placed in the virtual power plant (VPP) subject for optimization and scheduling. However, the virtual power plant (VPP) and power-to-hydrogen (P2H) usually belong to different stakeholders, and they only have their own information. Due to information asymmetry, the optimization scheduling may be large or untimely. Some scholars have considered treating the virtual power plant (VPP) and power-to-hydrogen (P2H) as different subjects and established a multi-agent decentralized coordinated operation model of VPP-P2H, but this model is a mixed nonlinear integer programming model and is difficult to apply to complex power networks. Therefore, it is an urgent problem to coordinate the relationship between the virtual power plant (VPP) and power-to-hydrogen (P2H) to ensure the stable operation of the VPP-P2H combined system and accurately characterize and use the power-to-hydrogen dispatchable domain of the virtual power plant containing power-to-hydrogen. Summary of the invention

[0005] The present invention provides a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing power-to-hydrogen, so as to realize the stable operation of a virtual power plant (VPP) and power-to-hydrogen (P2H) combined system and the accurate and efficient characterization of the electricity-hydrogen dispatchable domain.

[0006] The technical solution adopted by the present invention is:

[0007] An efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a power-to-hydrogen virtual power plant includes the following steps:

[0008] Step 1: Establish a hydrogen production plant model and a virtual power plant resource aggregation optimization model that take into account dynamic electrolysis efficiency, and determine the limit exchange power of the virtual power plant gateway node;

[0009] Step 2: Analytical expression of the external characteristics of electricity and hydrogen of the power-to-hydrogen virtual power plant based on the multi-parameter planning method;

[0010] Step 3: Solve the cumulative distribution function of the electricity-hydrogen external characteristics of the power-to-hydrogen virtual power plant based on the full probability formula;

[0011] Step 4: Collect day-ahead dispatch data and system equipment parameters, and use opportunity constraints to convert the cumulative distribution function of the electricity-hydrogen external characteristics of the power-to-hydrogen virtual power plant obtained in step 3 into dispatchable domain constraints, and establish a multi-virtual power plant-electricity-hydrogen integrated energy system dispatch model with the operator obtaining the minimum cost as the objective function;

[0012] Step 5: Based on the MATLAB platform and the YALMIP toolbox, build the multi-virtual power plant-electricity-hydrogen integrated energy system scheduling model of step 4, and call the solver Gurobi for solving it.

[0013] In step 1, starting from the dynamic operation characteristics and mechanism of the proton exchange membrane electrolyzer, a hydrogen production plant model taking into account the dynamic electrolysis efficiency with hydrogen production rate as the target is established as follows:

[0014] 1). Electrolyzer operating conditions constraints:

[0015] The working state and working condition change of the electrolytic cell require certain logical constraints, which can be expressed as follows:

[0016] O i,t +R i,t +F i,t +D i,t =1 (1);

[0017]

[0018] Among them, O i,t , R i,t 、F i,t , D i,t are all 0-1 variables, representing the overload state, full load state, fluctuation state and shutdown state of the i-th electrolytic cell at time t respectively; Formula (1) indicates that the electrolytic cell can only be in one operating state; in order to extend the service life of the electrolytic cell as much as possible, the electrolytic cell should be operated in the rated state as much as possible; Formula (2) indicates that the number of electrolytic cells in the fluctuation or overload state is at most one;

[0019] 2). Electrolyzer power constraints:

[0020] The power of the electrolyzer under fluctuating conditions is defined as fluctuating power, and the power under overload conditions is defined as overload power, which should meet the following constraints:

[0021] 0.2F i,t P rate ≤P f,i,t <F i,t P rate (3);

[0022] O i,t P rate <P o,i,t ≤1.2O i,t P rate (4);

[0023] Where P rate Indicates the rated power of the electrolytic cell, P f,i,t represents the power consumption of the ith electrolytic cell under the fluctuating state at time t, P o,i,t Indicates the power consumption of the electrolytic cell under overload condition at time t;

[0024] Then the input power P of the hydrogen plant at time t is t elen It can be expressed as:

[0025]

[0026] Where N h Indicates the number of electrolyzers. In addition, the sum of the input power of the hydrogen plant and the external output power of the virtual power plant should be within the upper and lower limits of the exchange power of the virtual power plant gateway node, as shown below:

[0027]

[0028] Where P out,t It represents the external output power of the virtual power plant at time t; and They represent the upper and lower limits of the gateway node exchange power respectively, which can be obtained by solving the gateway node limit exchange power.

[0029] 3). Dynamic hydrogen production efficiency:

[0030] According to the working principle of the proton exchange membrane electrolyzer and the relationship between the operating current, the power consumption P of the proton exchange membrane electrolyzer is t e and hydrogen production rate As shown below:

[0031] P t e =N E U cell,t I cell,t (7);

[0032]

[0033] Where P t e I is the power consumption of the proton exchange membrane electrolyzer at time t; cell,t is the operating current of the proton exchange membrane electrolyzer at time t, U cell,t is the voltage of the proton exchange membrane electrolyzer at time t, N E is the number of proton exchange membrane electrolyzers connected in series, which is related to the maximum capacity of the proton exchange membrane electrolyzer; is the hydrogen production rate of the proton exchange membrane electrolyzer at time t; k1 represents the conversion coefficient, specifically the molar mass of hydrogen; η f,t is the Faraday efficiency at time t, which is related to the production of hydrogen and oxygen in the proton exchange membrane electrolyzer; F represents the Faraday constant;

[0034] Further curve fitting is performed using the relevant variables of formula (7) and formula (8), and the following is obtained: Figure 2 As shown in the trend, the pseudo-relationship between the hydrogen production rate of the electrolyzer in a fluctuating state and the input power is nonlinear, but since it is a monotonically increasing relationship, in order to simplify the solution, the input power is divided into N k Segments, each of which is linearized and superimposed, as shown in the following formula:

[0035]

[0036] In the formula, The minimum input power of the ith proton exchange membrane electrolyzer is the power of the kth segment of the ith electrolyzer at time t; s i,k represents the slope of the i-th electrolytic cell in the k-th segment interval, represents the maximum hydrogen production rate of the i-th electrolyzer in the k-th segment interval, represents the minimum hydrogen production rate of the i-th electrolyzer in the k-th segment interval, △P i k represents the segmented power of the i-th electrolytic cell; represents the minimum hydrogen production rate of the ith electrolyzer, H f,i,t represents the hydrogen production rate of the ith electrolyzer under the fluctuating state at time t; N k Indicates the number of segments.

[0037] The hydrogen production rate can be regarded as the superposition of the hydrogen production rates of all electrolyzers under overload state, rated state and fluctuation state. The total hydrogen production rate of the electrolyzer is shown as follows:

[0038]

[0039] In the formula, H total，t Indicates the total hydrogen production rate of the hydrogen production plant at time t; indicates the hydrogen production rate of the electrolyzer under rated conditions; n h Indicates the hydrogen production rate of the electrolyzer under overload condition;

[0040] In step 1, the virtual power plant resource aggregation optimization model specifically includes the following contents:

[0041] In the virtual power plant resource aggregation optimization model, the gateway node limit interactive power is used as the objective function as shown in formula (13):

[0042]

[0043] Where P in,t is the interactive power of the virtual power plant at time t; among which, the branch flow constraints, node power balance constraints, and upper and lower limit constraints of the flow part variables in the optimization model are specifically as follows;

[0044] a). Branch power flow constraint: Line active power P L,t It should vary within the allowable transmission upper and lower limits, namely:

[0045]

[0046] P Lmin ≤P L,t ≤PLmax (15);

[0047] Where: P L,t and Q L,t are respectively the active and reactive power vectors of the internal branches of the virtual power plant at time t; U t and θ t are the node voltage amplitude and phase angle at time t respectively; G V , G θ , B V and B θ is the linearized power flow equation coefficient matrix; P Lmax and P Lmin They are the upper and lower limit vectors of branch active power transmission respectively.

[0048] b) Node power balance constraint: The sum of the inflow power of each node is equal to the sum of the outflow power, that is:

[0049]

[0050] Where: P N,t and Q N,t are the active and reactive power vectors injected into the node at time t; P G,t and Q G,t are the active and reactive power injected by the adjustable unit at time t; P R,t and Q R,t are the active and reactive power vectors of distributed renewable energy at time t; P D,t and Q D,t are the active and reactive power demand vectors of the load at time t; P in,t is the interactive power of the virtual power plant at time t; e G 、e R 、e D and e in are the association matrices of adjustable units, new energy units, loads and virtual power plant boundaries and nodes respectively; T L is the node-branch association matrix; G0 and B0 are the ground conductance matrix and ground susceptance matrix respectively.

[0051] c) Upper and lower limit constraints of the flow variables: variable U t ,θ t , P G,t and Q G,t The change of should not exceed its upper and lower limits, namely:

[0052]

[0053] Where: U max and U min are the upper and lower limit vectors of node voltage amplitude respectively; θ max and θmin are the upper and lower limit vectors of node voltage phase angle respectively; P Gmax and P Gmin are the upper and lower limits of the maximum active power adjustment range of the adjustable unit; Q Gmax and Q Gmin They are respectively the upper and lower limits of the maximum reactive power adjustment range of the adjustable units.

[0054] In step 2, firstly, taking the maximum scheduling boundary of a scheduling period as an example, the scheduling boundary optimization problem of a period is expressed in a general form:

[0055] maxH total (20);

[0056] In the formula, H total represents the total hydrogen production rate of the gateway hydrogen plant during the characterization period; the constraint can be expressed by the following formula (21):

[0057] M(H total ,x,y)= Ax+By+CH total +DP out +Ew n -I≤0 (21);

[0058] In the formula, M(H total ,x,y) is the general form of system constraints for the optimization model of the gateway hydrogen plant, where x = [O i R i F i D i ] are binary optimization variables, which respectively represent the overload state, full load state, fluctuation state and shutdown state of the i-th electrolytic cell in the specified characterization period;

[0059] y=[P f,i P o,i P elen H f,i ] and H total is a continuous optimization variable; P f,i represents the fluctuating power of the i-th electrolytic cell within the specified characterization period; P o,i represents the overload power of the i-th electrolytic cell in the specified characterization period; P elen Represents the total input power of the hydrogen production plant during the specified characterization period; H f,i represents the total hydrogen production rate of the hydrogen plant during the specified characterization period; A, B, C, D, E and I represent constant coefficient matrices.

[0060] Assume that the optimal solution of the optimization model (20) and (21) is x * ,y * and H total *, then at the optimal solution, the system constraints (21) can be divided into the effective constraints G a (H total ,x,y)=0 and the constraint G is invalid b (H total ,x,y)<0, and then the KKT optimality condition of the original problem is expressed as follows:

[0061]

[0062] Where: A a , B a , C a , D a 、E a and I a A is the constant coefficient matrix representing the constraints in effect; b , B b , C b , D b 、E b and I b The constant coefficient matrix representing the ineffective constraints; λ a represents the dual variable vector of the constraints in equation (21); b represents the dual variable vector of the ineffective constraints in equation (21); * represents the value of variable x when seeking the optimal solution to the optimization problem; y * Indicates the value of variable y when seeking the optimal solution to the optimization problem; The variable H represents the optimal solution of the optimization problem. total The value of P out represents the external output power of the virtual power plant; w n represents the random variable of renewable energy output;

[0063] Subscripts a and b represent active and inactive constraints, respectively, and λ is the dual multiplier vector of the time period scheduling boundary optimization problem expressed in general form. The optimal solution can be obtained by Gaussian elimination method. and external output power P out and new energy power w n The linear analytical relationship is:

[0064]

[0065] In the formula, F w,i , L w,i and G w,i is the corresponding coefficient matrix. Substituting the above formula into the inequality constraint of formula (22), we can obtain the random variable [w n ,P out ] A partition of the variation space is called the critical region CR i :

[0066]

[0067] In the formula, It means that the left side of the symbol is defined as the right side of the symbol, that is, formula (24) defines the critical region on the left side of the symbol as the inequality form on the right side of the symbol;

[0068] At the same time, the critical region CR of the random variable can be obtained, and the corresponding objective function virtual power plant dispatch boundary and external output power P out and new energy power w n The linear analytical relationship is:

[0069]

[0070] Based on the multi-parameter programming method, by enumerating all the constraints that work and do not work, we can obtain The hydrogen production rate boundary z * The piecewise linear analytical function of the random variable is denoted as:

[0071]

[0072] Where: CR i is the i-th critical region, CR j is the jth critical region; and the intersection of two critical regions is an empty set n CR Indicates that the number of critical regions is nth; represents the empty set;

[0073] Respectively represent the hydrogen production rates of the gateway nodes in the 1st, 2nd, and nth critical regions;

[0074] F1, L1, G1 represent the relationship matrix between the hydrogen production rate at the gateway node and the random variable obtained according to the linear analytical relationship in the first critical domain; F2, L2, G2 represent the relationship matrix between the hydrogen production rate at the gateway node and the random variable obtained according to the linear analytical relationship in the second critical domain; Represents the relationship matrix between the hydrogen production rate at the gateway node and the random variable obtained according to the linear analytical relationship in the nth critical region;

[0075] (w n ,P out )∈CR1,(w n ,P out )∈CR2, They represent random variables (w n ,P out ) belongs to the 1st, 2nd, and nth critical regions.

[0076] In step 3, the process of solving the cumulative distribution function of the electricity-hydrogen external characteristics of the power-to-hydrogen virtual power plant based on the full probability formula is as follows:

[0077] The distribution function of the electricity-to-hydrogen virtual power plant’s external output electricity-hydrogen boundary is the dispatch boundary z * The probability that the value Z is less than or equal to a given value Z can be obtained from formula (26). n When z is a fixed value, * and P out The electricity-hydrogen dispatch curve is formed, so only z * About w n The probability distribution of the electricity-to-hydrogen external characteristics of the electricity-to-hydrogen virtual power plant can be obtained.

[0078] In order to simplify the probability derivation process, the present invention sets P out =0, so the total probability formula gives z as shown in formula (27): * About w n The probability distribution of

[0079]

[0080] Where: Pr{z * ≤Z|w n ∈CR i} is to obtain the random variable w n Falling in the critical region CR i On, satisfy z * The conditional probability of being less than or equal to a given value Z; Pr(w n ∈CR i ) indicates that the random variable falls in the critical region CR i probability.

[0081] From formula (27), we can know that the cumulative distribution function Pr(z * (w n )≤Z), we need to obtain:

[0082] ① New energy power w derived from the internal resources of the virtual power plant n The probability distribution of

[0083] ②The conditional probability Pr{z * ≤Z|w n ∈CR i};

[0084] ③Probability Pr(w n ∈CR i ).

[0085] (1) Assuming that the probability information of renewable energy output in the virtual power plant distribution network follows a normal distribution, the renewable energy power w aggregated by the internal resources of the virtual power plant n The probability distribution of can be derived by the following process. The active power exchanged by the virtual power plant gateway node is affected by the sum of the active powers of the connected branches. The branch active power P L The distribution factor matrix F is transferred through the internal lines of the virtual power plant and is affected by the node injection power, that is:

[0086]

[0087] Where: F P is the submatrix related to branch active power in the virtual power plant internal line transfer distribution factor matrix F; e in is the branch-node association matrix connected to the gateway node, and It is the node-branch correlation matrix after removing the active injection power related sub-matrix of the virtual power plant gateway node.

[0088] Furthermore, according to formula (28), the distribution factor corresponding to the random variable w can be extracted, and the distributed new energy random variable w can be constructed. d On the transfer distribution factor vector F of active power exchange at the virtual power plant gateway P-w :

[0089]

[0090] Virtual power plant gateway exchange active power w n The branch active power e of the branch connected to it in P L Influence, branch active power e in P L Affected by the distributed new energy random variable w d Therefore, through formula (29) and the probability information of random variables in the virtual power plant, the gateway node limit exchange power P can be derived in The probability information is:

[0091]

[0092] Where P re and P sv are the expectation vector and variance vector composed of the expectation and variance of the random variables corresponding to the transfer distribution factor vector of the gateway exchange active power; F p-w Represents the new energy random variable w d Regarding the transfer distribution factor vector of active power exchanged at the virtual power plant gateway; Indicates P re The transpose of Indicates Psv The transpose of

[0093] (2) Conditional probability Pr{z * ≤Z|w n ∈CR i} is the probability To solve the probability Need to obtain The cumulative distribution function curve that satisfies; w n It obeys normal distribution; by the critical region CR i Linear function on Know It also obeys the normal distribution, and its expectation and variance can be obtained from formula (30). After the expectation and variance of the normal distribution, the probability can be obtained from the cumulative distribution function of the normal distribution. Then we get Pr{z * ≤Z|(w n )∈CR i}.

[0094]

[0095] Where: N w is the number of random variables in the random variable vector w; μ j and are the expectation and variance of the jth random variable respectively; L j is the row vector L i The element corresponding to the jth column in represents the linear function The coefficient of the j-th random variable in .

[0096] (3) The present invention is based on sampling calculation Pr(w n ∈CR i ): The core idea is to sample a known probability distribution and statistically satisfy the subdomain CR i The number of sample points to calculate Pr(w n )∈CR i ). The main computational burden of this method is to determine whether the sample falls into the subdomain CR i ; Since the sample judgment only needs to check inequality (24), it has less computational burden.

[0097] In step 4, a multi-virtual power plant-electric-hydrogen integrated energy system scheduling model is established with the operator obtaining the minimum cost as the objective function, and the cumulative distribution function of the electric-hydrogen external characteristics obtained in step 3 is converted into a dispatchable domain constraint condition method using opportunity constraints, as shown below:

[0098] The multi-virtual power plant-electricity-hydrogen integrated energy system scheduling model takes the lowest total cost as the optimization goal, which includes the power generation costs of conventional units and virtual power plants in the electric-hydrogen integrated energy system, the hydrogen production costs of the gateway hydrogen production plant, and the hydrogen transportation costs. It can be expressed as follows:

[0099]

[0100] Where: represents the external output power of the i-th virtual power plant at time t; It represents the external output power of the i-th conventional unit at time t; represents the output power of the i-th virtual power plant The cost function of represents the output power of the i-th conventional unit The cost function of P2H represents the hydrogen production cost of the Guankou hydrogen production plant; C RT represents the hydrogen transportation cost; T represents the total scheduling period;

[0101] Among them, the power generation cost of conventional units and virtual power plants is a quadratic function of the power generated by the units, which can be expressed as follows:

[0102]

[0103] In the formula, according to the existing market quotation requirements, the present invention assumes that the virtual power plant operator declares its operating cost as a quadratic function, They are the quadratic coefficient, linear coefficient and constant coefficient of the virtual power plant operation cost respectively; They are respectively the quadratic coefficient, linear coefficient and constant coefficient of the operating cost of conventional units in the electric-hydrogen integrated energy system.

[0104] The cost of a hydrogen production plant is generally divided into fixed costs and variable costs. Fixed costs include equipment investment costs C in and operation and maintenance costs C om Variable costs include electricity consumption costs during hydrogen production C power The unit cost calculation formula for hydrogen production by water electrolysis is as follows:

[0105] C P2H =C in +C om +C power (34);

[0106]

[0107] Where: represents the total rated power of the hydrogen production plant; p in is the unit price of equipment investment; p omis the average annual operation and maintenance cost per unit capacity of the electrolyzer; p price is the unit electricity cost of hydrogen production, is the total unit power consumption of the hydrogen plant at time t.

[0108] The main transportation cost is labor cost. Assuming that labor cost is proportional to the total travel time and the energy consumption during transportation is very low and can be ignored, the hydrogen transportation cost C RT It can be expressed as:

[0109]

[0110] Among them, c RT represents the transportation labor cost per unit time. In the present invention, a labor cost of 20 yuan per hour is used; T represents the total scheduling period; n represents the hydrogen production node identifier, and m represents the hydrogen demand node identifier; Ω hs represents the set of hydrogen production nodes, Ω hd represents the set of hydrogen demand nodes; γ n,m,t is a 0-1 variable indicating the trip, which is 1 when the hydrogen vehicle is on the way from the hydrogen production node n to the hydrogen demand node m at time t. Formula (36) mainly calculates the hydrogen transportation cost by traversing whether there is a hydrogen vehicle going from the hydrogen production node n to the hydrogen demand node m at time t.

[0111] The constraints include grid constraints, hydrogen grid constraints, and the electric-hydrogen dispatch boundary constraints of the electric-to-hydrogen virtual power plant converted by opportunity constraints. The grid constraints are mainly composed of branch flow constraints, system balance constraints, and flow upper and lower limit constraints. The hydrogen grid constraints are mainly composed of hydrogen production node constraints, hydrogen refueling station constraints, hydrogen vehicle storage constraints, hydrogen vehicle status constraints, and hydrogen vehicle driving time constraints. The process of converting the cumulative distribution function of the external characteristics of electric hydrogen into dispatchable domain constraints using opportunity constraints is as follows:

[0112] After obtaining the probability distribution of the dispatch boundary, in order to effectively deal with the uncertainty of the virtual power plant, the opportunity constraint can be used to model the virtual power plant dispatch boundary constraint, as shown in formula (37):

[0113]

[0114] Where: represents the power output of the i-th electricity-to-hydrogen virtual power plant to the grid at time t; It indicates that the i-th electricity-to-hydrogen virtual power plant outputs hydrogen to the hydrogen grid at time t; and They respectively represent the minimum and maximum values ​​of the dispatching boundary of the i-th virtual power plant at time t obtained by the probability distribution characterization method.

[0115] α max and α minare the confidence levels for satisfying the upper and lower boundaries of virtual power plant scheduling respectively; ζ represents the hydrogen-electricity conversion efficiency at different output powers; and The probability distribution function of is a nonlinear function, and constraint (37) is a nonlinear constraint. To ensure the efficiency of the coordinated optimization scheduling solution of the multi-virtual power plant-electric hydrogen energy system, the present invention converts the opportunity constraint into a linear constraint (38) based on the quantile conversion technology:

[0116]

[0117] Where: and They represent the minimum and maximum values ​​of the virtual power plant dispatch boundary obtained by quantile transformation; and The cumulative distribution function and the given confidence level α can be obtained min and α max To obtain.

[0118] In step 5, the multi-virtual power plant-electric-hydrogen integrated energy system scheduling model in step 4 is built based on the MATLAB platform and the YALMIP toolbox, and the day-ahead scheduling data and system equipment parameters are substituted into it. After determining the minimum cost obtained by the operator as the objective function, the gurobi solver is used to solve it.

[0119] The present invention provides a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, and the technical effects are as follows:

[0120] 1) The method of the present invention can obtain a piecewise analytical mapping between the hydrogen production rate and external output power of the electricity-to-hydrogen virtual power plant under different new energy outputs, and then calculate the cumulative distribution function of the electricity-hydrogen external characteristics based on the full probability principle, which can accurately reflect the probabilistic characteristics of the scheduling boundary fluctuations of the electricity-to-hydrogen virtual power plant.

[0121] 2) The system operation optimization proposed in the present invention taking into account the dynamic efficiency characteristics of the hydrogen production plant can not only enable the hydrogen production boundary to better follow the fluctuations of new energy, but also give full play to the complementary advantages of electricity and hydrogen energy. The hydrogen network operation strategy that takes into account the dynamic hydrogen production efficiency of the electrolyzer is more conducive to practical engineering guidance.

[0122] 3) The present invention applies the probability distribution of the external characteristics of electricity and hydrogen in a power-to-hydrogen virtual power plant to the coordinated optimization scheduling of multiple power-to-hydrogen virtual power plants and an electricity-to-hydrogen integrated energy system, which not only significantly improves the optimization scheduling calculation rate, but also avoids the uncertainty risks of distributed new energy and ensures the safe operation of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0123] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0124] Figure 1 Schematic diagram of the scheduling method for a virtual power plant involving electricity-to-hydrogen conversion.

[0125] Figure 2 Schematic diagram of linear fitting of hydrogen production rate.

[0126] Figure 3 Schematic diagram of the electric-hydrogen external characteristics of the virtual power plant under different new energy output conditions.

[0127] Figure 4 Schematic diagram of the coordinated scheduling method of the multi-electricity-to-hydrogen virtual power plant-electricity-hydrogen energy system.

[0128] Figure 5 Schematic diagram of hydrogen demand at a typical daily hydrogen load node.

[0129] Figure 6 This is the overall system structure diagram.

[0130] Figure 7 Schematic diagram of the probability information of the electricity-hydrogen external characteristics of the 3-node virtual power plant in period 1.

[0131] Figure 8 This is a typical daily dispatchable domain of a 3-node virtual power plant. DETAILED DESCRIPTION

[0132] An efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, comprising the following steps: first, a virtual power plant resource aggregation optimization model and a hydrogen production model of a gateway hydrogen production plant considering dynamic hydrogen production efficiency are established; second, based on a multi-parameter planning method, the external output power variable of the virtual power plant is used as a planning parameter to analyze the segmented analytical mapping of the external characteristics of electricity-hydrogen in the virtual power plant containing electricity-to-hydrogen, and the cumulative distribution function of the external characteristic curve of the virtual power plant containing electricity-to-hydrogen is calculated based on a full probability formula; finally, the cumulative distribution function of the external characteristic curve of the virtual power plant containing electricity-to-hydrogen is converted into a dispatchable domain constraint condition by using a chance constraint, and is applied to the coordinated optimization scheduling of the virtual power plant containing electricity-to-hydrogen and the electric-hydrogen energy system. Compared with the integrated coordinated scheduling, the method proposed in the present invention has a solution mean error of less than 0.01%, and the solution speed is accelerated by at least one order of magnitude, which is suitable for online scheduling of the system.

[0133] Example:

[0134] The 3-node electricity-to-hydrogen virtual power plant consists of a 3-node distribution network and a gateway hydrogen production plant with 2 electrolyzers. Node 1 is a gateway node, node 2 has loads and distributed renewable energy, and node 3 has loads and adjustable units. Assume that the power factor of the load and renewable energy are both 0.9. Assume that the active power of distributed renewable energy all obeys a normal distribution with the predicted value being the expected value and the standard deviation being 10% of the expected value. The maximum active power of distributed renewable energy at node 2 is Active power demand of node 2 load P D2 ~N(2,0.2 2 ), the active load demand of node 3 is P D3 ~N(1.5,0.15 2 ), node 3 adjustable unit P Gmax =4MW, P Gmin =1MW, Q Gmax =-Q Gmin =2.5MVar; electrolytic cell rated power P rate =2MW.

[0135] Using the above parameters, the analytical mapping calculation method of the dynamic hydrogen production rate and external output power of the gateway hydrogen production plant based on the multi-parameter planning theory is used to solve the segmented analytical mapping between the external output electricity-hydrogen characteristics of the three-node power-to-hydrogen virtual power plant. The external electricity-hydrogen characteristics of the three-node virtual power plant under different new energy output conditions can be obtained. Figure 3 shown.

[0136] In order to verify the overall accuracy of the probability distribution of the scheduling boundary of the 3-node test system, the present invention compares the calculation results with the Monte Carlo method calculation results, with a sample number of 10,000 times. The root mean square error is used to measure the error, that is, the following formula:

[0137]

[0138] Where: N f is the value of the curve's horizontal coordinate; f PM and f MC The corresponding points of the curve obtained by the proposed method and the curve obtained by Monte Carlo simulation are taken as values ​​respectively. The root mean square error values ​​of the cumulative distribution function curves obtained by the method proposed in the present invention are all less than 0.01%, and the results are very close to the results of the Monte Carlo simulation method. It can be seen that the method proposed in the present invention can effectively characterize the probability distribution of the virtual power plant scheduling boundary. At the same time, the calculation time of the virtual power plant scheduling boundary probability distribution characterization method taking into account the uncertainty of distributed new energy is about 20s, while the calculation time of the Monte Carlo simulation method is about 3000s. Therefore, the proposed method is significantly better than the Monte Carlo method in terms of calculation time. In short, the scheduling boundary probability distribution characterization method proposed in the present invention is superior in both calculation time and calculation accuracy.

[0139] The establishment of a multi-virtual power plant-electricity-hydrogen integrated energy system scheduling model Figure 4As shown, the power grid in the electric hydrogen energy system is an improved IEEE-30 node transmission network, which is connected to 3, 39, and 69 node virtual power plants at 5, 17, and 24. The hydrogen network is composed of 3 hydrogen production nodes and 15 hydrogen load nodes. The hydrogen production node is the gateway hydrogen production plant part of the electric-to-hydrogen virtual power plant. The time period is set to 24 time periods. In the embodiment, the main network IEEE-30 node transmission network system is not equipped with a new energy unit, the 3-node virtual power plant is the virtual power plant mentioned above, the IEEE-39 node virtual power plant is equipped with distributed new energy units at 4 nodes of 30, 33, 36, and 39, and the IEEE-69 node virtual power plant is equipped with distributed new energy units at 8 nodes of 7, 19, 25, 29, 33, 45, 51, and 58. And the rated power of all distributed new energy units in the 39-node and 69-node virtual power plants is shown in Table 1. In the same virtual power plant, the new energy fluctuation characteristics are the same. The typical daily new energy fluctuation characteristics of each virtual power plant are shown in Table 2; the hydrogen demand of the hydrogen load node is as shown in Table 2. Figure 5 As shown; the overall system structure diagram is as follows Figure 6 shown.

[0140] Table 1 Rated power of distributed renewable energy units in virtual power plants

[0141]

[0142] Table 2 Typical daily renewable energy fluctuation characteristics of each virtual power plant

[0143]

[0144] Based on the above data, after the cumulative distribution function curve is obtained through analytical calculations in steps 1 to 3, the opportunity constraint is used to model the uncertainty of the virtual power plant dispatch boundary, and the opportunity constraint is converted into an output boundary constraint based on the quantile conversion technology, as shown in the following figure: Figure 7 The proposed method is used to analyze the 24-period dispatchable domain of virtual power plant 1 in the electricity-hydrogen space. The results are shown in Figure 8 shown.

[0145] Furthermore, the applicability of the mentioned multi-electricity-to-hydrogen virtual power plant-electricity-hydrogen energy system collaborative scheduling method based on the external characteristics of electricity-hydrogen was analyzed, and the following three methods were used for comparative verification.

[0146] M1: Multi-electricity-to-hydrogen virtual power plant-integrated coordinated scheduling method of electric-hydrogen energy system.

[0147] M2: A coordinated scheduling method for multi-electric-to-hydrogen-electric-hydrogen energy systems based on the external characteristics of electricity-hydrogen, which does not consider the uncertainty of distributed new energy within the virtual power plant.

[0148] M3: A coordinated scheduling method for multi-electric-to-hydrogen-electricity-hydrogen energy systems based on the external characteristics of electricity-hydrogen.

[0149] Table 3 gives the coordinated dispatch optimization dispatch results of the three methods. As can be seen from the table, the boundary exchange dispatch result of M1 is slightly larger than that of M3. This is because the M1 method uses the interval constraint method to deal with the uncertainty of distributed renewable energy within the virtual power plant. The output of renewable energy varies between the upper and lower bounds of the prediction. In the coordinated dispatch calculation, the output of renewable energy tends to the upper bound of the prediction. Therefore, the boundary nodes of the virtual power plant can output more power. In the M3 method, the output of distributed renewable energy satisfies the normal distribution with its predicted value as the expectation. In the coordinated dispatch calculation, the output of renewable energy tends to vary near the predicted value. Therefore, compared with the M1 method, the boundary nodes of the virtual power plant output less power. Therefore, due to the reduction in the exchange power capacity of the boundary nodes of the virtual power plant, the M3 method will dispatch more conventional units of the main grid, and the M3 method will cost more operating costs. Although the operating cost of the M3 method increases, the increased operating cost is 0.677% of the operating cost of the M3 method, and the increase in operating cost is acceptable. In terms of computing time, the M3 method treats the virtual power plant as a whole and transmits some of its important information to the main network, reducing the number of variables and constraints in the collaborative scheduling model. The computing time is only 15.9% of that of the M1 method, so the M3 method can guarantee a high computing efficiency. In addition, compared with the M1 method, the M3 method avoids the calculation and storage of massive virtual power plant internal resource data by the electric hydrogen dispatch center, easing the computing and communication pressure of the unified dispatch center. At the same time, the virtual power plant can also effectively protect the data privacy of internal resources.

[0150] Table 3 Collaborative optimization scheduling results for time period 1 under different methods

[0151]

[0152] Furthermore, for methods M2 and M3, random sampling is performed on the scheduling boundary of the 69-node test system, and the probability of not satisfying constraint formula (38) is statistically analyzed. The probability of constraint exceeding the limit is lower, and the system security is higher. As can be seen from Table 4, the scheduling boundary of the 69-node test system of M3 can satisfy the opportunity constraint formula (38). For the scheduling boundary of M2, the probability of the 69-node test system scheduling boundary of M2 not satisfying constraint formula (38) is about 4.27%, and the probability of constraint exceeding the limit is high, and the system security is difficult to guarantee. Therefore, the multi-electric-to-hydrogen virtual power plant-electric-hydrogen energy system collaborative optimization scheduling method based on the probability distribution of the virtual power plant scheduling boundary proposed in the present invention can avoid the uncertainty risk of new energy and ensure the safe operation of the system.

[0153] Table 4 Scheduling boundary security verification

[0154]

Claims

1. An efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a power-to-hydrogen virtual power plant, characterized in that The following steps are involved: Step 1: Establish a hydrogen production plant model and a virtual power plant resource aggregation optimization model taking into account dynamic electrolysis efficiency, and determine the limit exchange power of the virtual power plant gateway node; Step 2: Analytical expression of the external characteristics of electricity and hydrogen of the power-to-hydrogen virtual power plant based on the multi-parameter planning method; Step 3: Solve the cumulative distribution function of the electricity-hydrogen external characteristics of the power-to-hydrogen virtual power plant; Step 4: Collect day-ahead dispatch data and system equipment parameters, and use opportunity constraints to convert the cumulative distribution function of the electricity-hydrogen external characteristics of the power-to-hydrogen virtual power plant obtained in step 3 into dispatchable domain constraints, and establish a multi-virtual power plant-electricity-hydrogen integrated energy system dispatch model with the operator obtaining the minimum cost as the objective function; Step 5: Based on the multi-virtual power plant-electric-hydrogen integrated energy system scheduling model in step 4, call the solver for solution.

2. According to claim 1, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: In step 1, a hydrogen production plant model taking into account the dynamic electrolysis efficiency with hydrogen production rate as the target is established as follows: 1). Electrolyzer operating conditions constraints: The operating state of the electrolytic cell and the logical constraints of its operating state changes are expressed as follows: O i,t +R i,t +F i,t +D i,t =1 (1); Among them, O i,t , R i,t 、F i,t , D i,t are all 0-1 variables, representing the overload state, full load state, fluctuation state and shutdown state of the i-th electrolytic cell at time t respectively; Formula (1) indicates that the electrolytic cell can only be in one operating state; in order to extend the service life of the electrolytic cell as much as possible, the electrolytic cell should be operated in the rated state as much as possible; Formula (2) indicates that the number of electrolytic cells in the fluctuation or overload state is at most one; 2). Electrolyzer power constraints: The power of the electrolyzer under fluctuating conditions is defined as fluctuating power, and the power under overload conditions is defined as overload power, satisfying the following constraints: 0.2F i,t P rate ≤P f,i,t <F i,t P rate (3); THE i,t P rate <P o,i,t ≤1.2O i,t P rate (4); Where P rate Indicates the rated power of the electrolytic cell, P f,i,t represents the power consumption of the ith electrolytic cell under the fluctuating state at time t, P o,i,t Indicates the power consumption of the electrolytic cell under overload condition at time t; The input power P of the hydrogen production plant at time t t elen It is expressed as: Where N h Indicates the number of electrolyzers. In addition, the sum of the input power of the hydrogen plant and the external output power of the virtual power plant should be within the upper and lower limits of the exchange power of the virtual power plant gateway node, as shown below: Where P out,t It represents the external output power of the virtual power plant at time t; and They represent the upper and lower limits of the gateway node exchange power, respectively, and are obtained by solving the gateway node limit exchange power; 3). Dynamic hydrogen production efficiency: Power consumption of proton exchange membrane electrolyzer P t e and hydrogen production rate As shown below: P t e =N E U cell,t I cell,t (7); Where P t e I is the power consumption of the proton exchange membrane electrolyzer at time t; cell,t is the operating current of the proton exchange membrane electrolyzer at time t, U cell,t is the voltage of the proton exchange membrane electrolyzer at time t, N E is the number of proton exchange membrane electrolyzers connected in series; is the hydrogen production rate of the proton exchange membrane electrolyzer at time t; k1 represents the conversion coefficient, specifically the molar mass of hydrogen; η f,t is the Faraday efficiency at time t; F represents the Faraday constant; Further, the curve fitting is performed based on the relevant variables of equations (7) and (8). The relationship between the hydrogen production rate of the electrolyzer in a fluctuating state and the input power is nonlinear. The input power is divided into N k Segments, each of which is linearized and superimposed, as shown in the following formula: In the formula, The minimum input power of the ith proton exchange membrane electrolyzer is the power of the kth segment of the ith electrolyzer at time t; s i,k represents the slope of the i-th electrolytic cell in the k-th segment interval, represents the maximum hydrogen production rate of the i-th electrolyzer in the k-th segment interval, represents the minimum hydrogen production rate of the i-th electrolyzer in the k-th segment interval, △P i k represents the segmented power of the i-th electrolytic cell; represents the minimum hydrogen production rate of the ith electrolyzer, H f,i,t represents the hydrogen production rate of the ith electrolyzer under the fluctuating state at time t; N k Indicates the number of segments; The hydrogen production rate is considered to be the sum of the hydrogen production rates of all electrolyzers under overload, rated and fluctuating conditions. The total hydrogen production rate of the electrolyzer is shown as follows: In the formula, H total，t Indicates the total hydrogen production rate of the hydrogen production plant at time t; indicates the hydrogen production rate of the electrolyzer under rated conditions; n h Indicates the hydrogen production rate of the electrolyzer under overload conditions.

3. According to claim 2, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: In step 1, the virtual power plant resource aggregation optimization model specifically includes the following contents: In the virtual power plant resource aggregation optimization model, the gateway node limit interactive power is used as the objective function as shown in formula (13): Where P in,t is the interactive power of the virtual power plant at time t.

4. According to claim 3, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: The optimization model includes: branch power flow constraints, node power balance constraints, and upper and lower limit constraints of some power flow variables, as shown below; a). Branch power flow constraint: Line active power P L,t It should vary within the allowable transmission upper and lower limits, namely: P Lmin ≤P L,t ≤P Lmax (15); Where: P L,t and Q L,t are respectively the active and reactive power vectors of the internal branches of the virtual power plant at time t; U t and θ t are the node voltage amplitude and phase angle at time t respectively; G V , G θ , B V and B θ is the linearized power flow equation coefficient matrix; P Lmax and P Lmin They are the upper and lower limit vectors of branch active power transmission respectively; b) Node power balance constraint: The sum of the inflow power of each node is equal to the sum of the outflow power, that is: Where: P N,t and Q N,t are the active and reactive power vectors injected into the node at time t; P G,t and Q G,t are the active and reactive power injected by the adjustable unit at time t; P R,t and Q R,t are the active and reactive power vectors of distributed renewable energy at time t; P D,t and Q D,t are the active and reactive power demand vectors of the load at time t; P in,t is the interactive power of the virtual power plant at time t; e G 、e R 、e D and e in are the association matrices of adjustable units, new energy units, loads and virtual power plant boundaries and nodes respectively; T L is the node-branch association matrix; G0 and B0 are the earth conductance matrix and earth susceptance matrix respectively; c) Upper and lower limit constraints of the flow variables: variable U t ,θ t , P G,t and Q G,t The change of should not exceed its upper and lower limits, namely: Where: U max and U min are the upper and lower limit vectors of node voltage amplitude respectively; θ max and θ min are the upper and lower limit vectors of node voltage phase angle respectively; P Gmax and P Gmin are the upper and lower limits of the maximum active power adjustment range of the adjustable unit; Q Gmax and Q Gmin They are respectively the upper and lower limits of the maximum reactive power adjustment range of the adjustable units.

5. According to claim 1, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: In step 2, firstly, taking the maximum scheduling boundary of a scheduling period as an example, the scheduling boundary optimization problem of a period is expressed in a general form: max H total (20); In the formula, H total represents the total hydrogen production rate of the gateway hydrogen plant during the characterization period; the constraint is expressed by the following formula (21): M(H total ,x,y)=Ax+By+CH total +DP out +Ew n -I≤0 (21); In the formula, M(H total ,x,y) is the general form of the system constraints of the optimization model of the gateway hydrogen plant, where x = [O i R i F i D i ] are binary optimization variables, which respectively represent the overload state, full load state, fluctuation state and shutdown state of the i-th electrolytic cell in the specified characterization period; y=[P f,i P o,i P elen H f,i ] and H total is a continuous optimization variable; P f,i represents the fluctuating power of the i-th electrolytic cell within the specified characterization period; P o,i represents the overload power of the i-th electrolytic cell in the specified characterization period; P elen Represents the total input power of the hydrogen production plant during the specified characterization period; H f,i represents the total hydrogen production rate of the hydrogen production plant during the specified characterization period; A, B, C, D, E and I represent constant coefficient matrices; Assume that the optimal solution of the optimization model (20) and (21) is x * ,y * and H total * , then at the optimal solution, the system constraint (21) is divided into the effective constraint G a (H total ,x,y)=0 and the constraint G is invalid b (H total ,x,y)<0, and then the KKT optimality condition of the original problem is expressed as follows: Where: A a , B a , C a , D a 、E a and I a A is the constant coefficient matrix representing the constraints in effect; b , B b , C b , D b 、E b and I b The constant coefficient matrix representing the ineffective constraints; λ a represents the dual variable vector of the constraints in equation (21); b represents the dual variable vector of the ineffective constraints in equation (21); * represents the value of variable x when seeking the optimal solution to the optimization problem; y * Indicates the value of variable y when seeking the optimal solution to the optimization problem; The variable H represents the optimal solution of the optimization problem. total The value of P out represents the external output power of the virtual power plant; w n represents the random variable of renewable energy output; Obtaining the optimal solution through Gaussian elimination and external output power P out and new energy power w n The linear analytical relationship is: In the formula, F w,i , L w,i and G w,i is the corresponding coefficient matrix. Substituting the above formula into the inequality constraint of formula (22), we can obtain the random variable [w n ,P out ] A partition of the variation space is called the critical region CR i : In the formula, It means that the left side of the symbol is defined as the right side of the symbol, that is, formula (24) defines the critical region on the left side of the symbol as the inequality form on the right side of the symbol; At the same time, the critical region CR of the random variable is obtained, and the corresponding objective function virtual power plant dispatch boundary and external output power P out and new energy power w n The linear analytical relationship is: Based on the multi-parameter programming method, by enumerating all the constraints that work and do not work, we can obtain The hydrogen production rate boundary z * The piecewise linear analytical function of the random variable is denoted as: Where: CR i is the i-th critical region, CR j is the jth critical region; and the intersection of two critical regions is an empty set n CR Indicates that the number of critical regions is nth; represents the empty set; Respectively represent the hydrogen production rates of the gateway nodes in the 1st, 2nd, and nth critical regions; F1, L1, G1 represent the relationship matrix between the hydrogen production rate at the gateway node and the random variable obtained according to the linear analytical relationship in the first critical domain; F2, L2, G2 represent the relationship matrix between the hydrogen production rate at the gateway node and the random variable obtained according to the linear analytical relationship in the second critical domain; represents the relationship matrix between the hydrogen production rate at the gateway node and the random variable obtained according to the linear analytical relationship in the nth critical region; (w n ,P out )∈CR1,(w n ,P out )∈CR2, They represent random variables (w n ,P out ) belongs to the 1st, 2nd, and nth critical regions.

6. According to claim 5, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: In step 3, the process of solving the cumulative distribution function of the electricity-hydrogen external characteristics of the power-to-hydrogen virtual power plant based on the full probability formula is as follows: The distribution function of the electricity-to-hydrogen virtual power plant’s external output electricity-hydrogen boundary is the dispatch boundary z * The probability that the value Z is less than or equal to a given value Z can be obtained from formula (26). n When z is a fixed value, * and P out The electricity-hydrogen dispatch curve is formed, so only z * About w n The probability distribution of , that is, the probability characteristics of the external characteristics of electricity and hydrogen in the power-to-hydrogen virtual power plant are obtained; To simplify the probability derivation process, set P out =0, so the total probability formula gives z as shown in formula (27): * About w n The probability distribution of Where: Pr{z * ≤Z|w n ∈CR i } is to obtain the random variable w n Falling in the critical region CR i On, satisfy z * The conditional probability of being less than or equal to a given value Z; Pr(w n ∈CR i ) indicates that the random variable falls in the critical region CR i probability.

7. According to claim 6, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: From formula (27), we can know that the cumulative distribution function Pr(z * (w n )≤Z), we need to obtain: ① The new energy power w derived from the internal resources of the virtual power plant n The probability distribution of ②The conditional probability Pr{z * ≤Z|w n ∈CR i }; ③Probability Pr(w n ∈CR i ); (1) Assuming that the probability information of renewable energy output in the virtual power plant distribution network follows a normal distribution, the renewable energy power w aggregated by the internal resources of the virtual power plant n The probability distribution of can be derived by the following process; the active power exchanged by the virtual power plant gateway node is affected by the sum of the active powers of the connected branches, and the branch active power P L The distribution factor matrix F is transferred through the internal lines of the virtual power plant and is affected by the node injection power, that is: Where: F P is the submatrix related to branch active power in the virtual power plant internal line transfer distribution factor matrix F; e in is the branch-node association matrix connected to the gateway node, and It is the node-branch correlation matrix after removing the active injection power related sub-matrix of the virtual power plant gateway node; Furthermore, according to formula (28), the distribution factor corresponding to the random variable w is extracted, and the distributed new energy random variable w is constructed. d On the transfer distribution factor vector F of active power exchange at the virtual power plant gateway P-w : Virtual power plant gateway exchange active power w n The branch active power e of the branch connected to it in P L Influence, branch active power e in P L Affected by the distributed new energy random variable w d Therefore, the gateway node limit exchange power P is derived through formula (29) and the probability information of random variables in the virtual power plant. in The probability information is: Where P re and P sv are the expectation vector and variance vector composed of the expectation and variance of the random variables corresponding to the transfer distribution factor vector of the gateway exchange active power; F p-w Represents the new energy random variable w d Regarding the transfer distribution factor vector of active power exchanged at the virtual power plant gateway; Indicates P re The transpose of Indicates P sv The transpose of (2) Conditional probability Pr{z * ≤Z|w n ∈CR i } is the probability To solve the probability Need to obtain The cumulative distribution function curve that satisfies; w n It obeys normal distribution; by the critical region CR i Linear function on Know It also obeys the normal distribution, and its expectation and variance can be obtained from formula (30); After the expectation and variance of the normal distribution, the probability can be obtained from the cumulative distribution function of the normal distribution. Then we get Pr{z * ≤Z|(w n )∈CR i }; Where: N w is the number of random variables in the random variable vector w; μ j and are the expectation and variance of the jth random variable respectively; L j is the row vector L i The element corresponding to the jth column in represents the linear function The coefficient of the j-th random variable in ; (3) Based on sampling calculation Pr(w n ∈CR i ): Sampling the known probability distribution, the statistics satisfy the subdomain CR i The number of sample points to calculate Pr(w n )∈CR i ); The computational burden of this method is to determine whether the sample falls into the subdomain CR i .

8. According to claim 7, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: In step 4, a multi-virtual power plant-electric-hydrogen integrated energy system scheduling model is established with the operator obtaining the minimum cost as the objective function, and the cumulative distribution function of the electric-hydrogen external characteristics obtained in step 3 is converted into a dispatchable domain constraint condition method using opportunity constraints, as shown below: The multi-virtual power plant-electricity-hydrogen integrated energy system scheduling model takes the minimum total cost as the optimization goal, which includes the power generation costs of conventional units and virtual power plants in the electric-hydrogen integrated energy system, the hydrogen production cost of the gateway hydrogen production plant, and the hydrogen transportation cost, which can be expressed as follows: Where: represents the external output power of the i-th virtual power plant at time t; It represents the external output power of the i-th conventional unit at time t; represents the output power of the i-th virtual power plant The cost function of represents the output power of the i-th conventional unit The cost function of P2H represents the hydrogen production cost of the Guankou hydrogen production plant; C RT represents the hydrogen transportation cost; T represents the total scheduling period; Among them, the power generation cost of conventional units and virtual power plants is a quadratic function of the power generated by the units, which can be expressed as follows: In the formula, They are the quadratic coefficient, linear coefficient and constant coefficient of the virtual power plant operation cost respectively; They are the quadratic coefficient, linear coefficient and constant coefficient of the operating cost of conventional units in the electric-hydrogen integrated energy system; The cost of the hydrogen production plant at the gateway is divided into fixed cost and variable cost. The fixed cost includes the equipment investment cost C in and operation and maintenance costs C om ; Variable costs include the electricity consumption cost Cp in the hydrogen production process ower ; The unit cost calculation formula for hydrogen production by water electrolysis is as follows: C P2H =C in +C om +C power (34); Where: Indicates the total rated power of the hydrogen production plant; p in is the unit price of equipment investment; p om is the average annual operation and maintenance cost per unit capacity of the electrolyzer; p price is the unit electricity cost of hydrogen production, is the total unit power consumption of the hydrogen plant at time t; Hydrogen transportation cost C RT It can be expressed as: Among them, c RT represents the transportation labor cost per unit time, T represents the total scheduling period; n represents the hydrogen production node identifier, m represents the hydrogen demand node identifier; Ω hs represents the set of hydrogen production nodes, Ω hd represents the set of hydrogen demand nodes; γ n,m,t is a 0-1 variable indicating the trip, which is 1 when the hydrogen vehicle is on the way from the hydrogen production node n to the hydrogen demand node m at time t. Formula (36) calculates the hydrogen transportation cost by traversing whether there is a hydrogen vehicle going from the hydrogen production node n to the hydrogen demand node m at time t.

9. According to claim 8, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: The constraints of the multi-virtual power plant-electricity-hydrogen integrated energy system scheduling model include: grid constraints, hydrogen grid constraints, and electricity-hydrogen scheduling boundary constraints of the electricity-to-hydrogen virtual power plant converted from opportunity constraints; Among them, the power grid constraints are composed of branch flow constraints, system balance constraints, and flow upper and lower limit constraints; the hydrogen network constraints are composed of hydrogen production node constraints, hydrogen refueling station constraints, hydrogen vehicle storage constraints, hydrogen vehicle state constraints, and hydrogen vehicle driving time constraints; the process of converting the cumulative distribution function of the external characteristics of electric hydrogen into the dispatchable domain constraint conditions using opportunity constraints is as follows: After obtaining the probability distribution of the dispatch boundary, in order to effectively deal with the uncertainty of the virtual power plant, the opportunity constraint can be used to model the virtual power plant dispatch boundary constraint, as shown in formula (37): Where: represents the power output of the i-th electricity-to-hydrogen virtual power plant to the grid at time t; It indicates that the i-th electricity-to-hydrogen virtual power plant outputs hydrogen to the hydrogen grid at time t; and They represent the minimum and maximum values ​​of the dispatch boundary of the ith virtual power plant at time t obtained by the probability distribution characterization method respectively; α max and α min are the confidence levels for satisfying the upper and lower boundaries of virtual power plant scheduling respectively; ζ represents the hydrogen-electricity conversion efficiency at different output powers; and The probability distribution function of is a nonlinear function, and constraint (37) is a nonlinear constraint. In order to ensure the efficiency of the coordinated optimization scheduling solution of the multi-virtual power plant-electric hydrogen energy system, the opportunity constraint is transformed into a linear constraint (38) based on the quantile conversion technology: Where: and They represent the minimum and maximum values ​​of the virtual power plant dispatch boundary obtained by quantile transformation; and The cumulative distribution function obtained and the given confidence level α min and α max To obtain.

10. According to claim 1, a highly efficient probabilistic characterization method for the external characteristics of electricity-hydrogen in a virtual power plant containing electricity-to-hydrogen, characterized in that: In step 5, the multi-virtual power plant-electric-hydrogen integrated energy system scheduling model in step 4 is built based on the MATLAB platform and the YALMIP toolbox, and the day-ahead scheduling data and system equipment parameters are substituted into it. After determining the minimum cost obtained by the operator as the objective function, the gurobi solver is used to solve it.

Images