Method and dapparatus for evaluating available transfer capability of power system with multiple regions

Abstract

An available transfer capability evaluation method and apparatus for a multi-region power system, belonging to the technical field of new energy grid connection. The method comprises the steps: in view of multi-dimensional uncertainty of new energy output and a load demand, on the basis of a conditional generative adversarial network method, determining a typical daily source-load scenario set; constructing an initial operation point set on the basis of the typical daily source-load scenario set, and determining a limit operation point of a multi-region power system; on the basis of the initial operation point set and the limit operation point, constructing an ATC evaluation model on the basis of safety indexes of multi-region power grid operation; and, on the basis of the ATC evaluation model and the typical daily source-load scenario set, determining available transfer capability probability distribution of the multi-region power system.

Description

[0001] This application claims priority to Chinese Patent Application No. 202410281521.2 filed with the China National Intellectual Property Administration (CNIPA) on Mar. 13, 2024, the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to a method and an apparatus for evaluating an available transfer capability of a power system with multiple regions, belonging to the technical field of renewable energy grid connection. BACKGROUND

[0003] In the power market environment, how to achieve economically optimal operation while ensuring system safety constraints has become an urgent problem to be solved by power grid managers and power market participants. Available transfer capability (ATC) represents the remaining power transfer capacity on the premise of ensuring the safe and stable operation of the power system. It approximately measures the safety and stability margin of the current operating point of the power grid. Therefore, ATC is not only an important basis for all parties involved in the power market to conduct transfer rights transactions, but also a boundary condition for power system expansion planning. In this context, proposing an ATC evaluation method that coordinates safety and economy is of great significance for meeting the reasonable operation of the power market and improving the efficiency of power grid utilization.

[0004] In early studies, domestic and foreign scholars proposed a variety of calculation methods such as linear distribution factor method, continuous power flow method, repeated power flow method and optimal power flow (OPF) for the ATC evaluation problem with deterministic source and load parameters. Among those methods, the OPF method models the ATC calculation as a mathematical optimization problem with maximizing the transfer power of inter-regional channels as the objective function and with grid power flow balance and safety and stability criteria as constraints. By solving the OPF optimization problem, the ATC under a certain operation state of the system can be obtained. The new power system with renewable energy as the main body has the characteristics of high renewable energy penetration rate and complex load structure. Therefore, the uncertainty of the output of large-scale wind farms and photovoltaic power stations, as well as the uncertainty of load, makes the transfer power between different regions uncertain, increasing the difficulty of ATC evaluation. In response to the uncertainty of renewable energy and load, scholars at home and abroad have proposed methods based on robust optimization, interval optimization, and optimization with chance constraints and the like.

[0005] Affected by the timing uncertainty of renewable energy and load, the ATC of the new power system actually has the characteristics of timing probability distribution. Therefore, the key difficulty of ATC evaluation lies in the generation of source-load random scenarios and the construction of ATC optimization models. Traditional methods based on robust optimization or interval optimization have less computational complexity, but only provide the distribution range of random variables and fail to fully characterize the probability distribution of ATC. Moreover, the uncertainty of renewable energy and load in large-scale power systems has high-dimensional uncertainty characteristics. Although the method based on the traditional Monte Carlo simulation can generate a large number of random scenarios, it has a large amount of calculations and can hardly effectively classify and screen the scenarios, which brings certain difficulties to the probability distribution evaluation of ATC. Therefore, how to comprehensively characterize the probability distribution of ATC is a technical problem that needs to be solved urgently by the person skilled in the art. SUMMARY

[0006] It is an object of embodiments of the present application to provide a thermoregulation device which is capable of rapidly adjusting the temperature of a to-be-thermoregulated component, with a high heat exchange efficiency.

[0007] In order to solve the above problems, a method and apparatus for evaluating an available transfer capability of a power system with multiple regions are provided according to the present application, which can accurately and efficiently calculate the ATC of the renewable energy transfer section online, and facilitate improvement of renewable energy consumption capability.

[0008] The technical solution adopted by the present application to solve the technical problem is as follow.

[0009] In a first aspect, a method for evaluating an available transfer capability of a power system with multiple regions is provided according to embodiments of the present application, which includes the following steps:

[0010] in consideration with multi-dimensional uncertainty of a renewable energy output and load demand, determining a typical daily source-load scenario set based on the conditional generative adversarial network method;

[0011] according to the typical daily source-load scenario set, constructing an initial operating point set, and determining a limit operating point of the power system with multiple regions;

[0012] according to the initial operating point set and the limit operating point, constructing an available transfer capability (ATC) evaluation model based on an operation safety index of the power system with multiple regions; and

[0013] according to the ATC evaluation model and the typical daily source-load scenario set, determining a probability distribution of the available transfer capability of the power system with multiple regions.

[0014] As a possible implementation of this embodiment, the determining a typical daily source-load scenario set based on the conditional generative adversarial network method includes as follows.

[0015] Based on the conditional generative adversarial network method, deep neural network models are used to characterize a nonlinear relationship generator and a classification signal discriminator, and conditional information is transmitted as an input layer to the classification signal discriminator and the nonlinear relationship generator. The conditional information includes: historical meteorological data with time attributes, spatial characteristics of the power system, and wind farm output characteristics and load demand characteristics.

[0016] Based on historical data of a renewable energy output and load demand, a typical daily source-load scenario set S0 is constructed: = {    ,    }, = (pW, pW, ..., pW], SD ={P1, P2, . , P2] ,

[0017] in the formulas:   1 represents an output of a wind farm connected to a node i, represents a set of 1 in the power system, represents a load of a wind farm connected to the node i,     represents a set of in the power system, W represents a set of wind farm nodes in the power system, and D represents a set of load nodes in the power system.

[0018] As a possible implementation of this embodiment, the according to the typical daily source-load scenario set, constructing an initial operating point set, and determining a limit operating point of the power system with multiple regions includes:

[0019] according to the initial operating point set and based on the operation safety index of the power system with multiple regions, continuously increasing exchange power between the multiple regions of the power system by adjusting power generation and load power, so as to obtain a limit operating point of the power system with multiple regions under limit operating conditions.

[0020] As a possible implementation of this embodiment, the according to the initial operating point set and the limit operating point, constructing an ATC evaluation model based on an operation safety index of the power system includes as follows.

[0021] An objective function is established based on objectives of minimizing overall generation cost and maximizing ATC between different regions of the power system: aCiPiG - ^Z(PiGm - PiG) i ^G Gm         G

[0022] in the formula, Pi,t and Pi,t are an active output of a thermal generator at a node i in a limit operation state and an active output of the thermal generator at the node i at an initial operating point respectively; G is a set of power generation nodes to be adjusted at a transfer end in the power system; Ci is unit power generation cost of unit i, a and 0 are weight coefficients of multi-objective optimization.

[0023] A power system benchmark (or reference) operation state model is constructed, and the power system benchmark operation state model includes a grid structure, a startup mode, load power, operating parameters of power growth mode, and safety constraint conditions.

[0024] By adjusting power generation and load power, exchange power between the multiple regions is continuously increased until any safety constraint condition is exceeded, and thus, the limit operating point of the power system is obtained, and the ATC evaluation model is constructed.

[0025] As a possible implementation of this embodiment, the safety constraint conditions include as follow.

[0026] (1) An active power balance constraint and a reactive power balance constraint of the node in the benchmark operation state of the power system are respectively: GWD Pi,t = Pi,t + Pi,t    Pi,t Qi,t=QiG,t+0-QiD,t

[0027] in the formulas, Pi,t and Qi,t are an active input power and a reactive input power of PG      QG a node i at time t respectively, i,t and i,t are an active output and a reactive output of a PW thermal power generator unit at the node i at the time t respectively, i,t is an active output of a PD     QD wind farm at the node i at the time t; i,t and i,t are an active load and a reactive load of the node i at the time t respectively, and Xi is a ratio of the reactive load to the active load.

[0028] A linearized power flow constraint in the benchmark operation state of the power system is: Pi,t Qi,t NB =Z[ Bj(s.t. -s,,.)+B (V,.. -j=1 j * i NB =E[-B^i., - ^j,.)+B2 ( v ,. j=1 j * i Vj..)] - Vj..)]

[0029] The amplitude and phase angle of the voltage of a node can both be solved through the linear equation, and an equivalent admittance 1 and 2 of the power flow equation is shown as follows: B1 = lJ    r 2 , v2 + B 2 = -^ J    r2j+xij

[0030] A capacity constraint of the line in the benchmark operation state of the power system is: max      2 -Sj  < Bj(s<4 -0.5Simj ax -Bi1j -s,..)+Bj( v. . - Vj.. )■ Sima (^. - s„)+Bj( Vi. - V,.. )< 0.5 st '

[0031] A node voltage constraint in the benchmark operation state of the power system is: ymin ^ Vi,t < ymax

[0032] An output constraint of a generator unit in the benchmark operation state of the power system is: Gmin P 1 om" Gmax

[0033] The above formula indicates that at time t, the active output of an nth generator unit must be between its upper limit and lower limit.

[0034] The following formulas represent increase or decrease in output of a generator unit per unit time respectively: G i,t,s G       Gmax Pi, t+1 — rP i rP Gmax G i,t G i,t +1

[0035] in the formulas, , and , are the phase angle and the amplitude of a voltage at the node i at the time t respectively; and are a resistance and reactance of a line (i, j) respectively;         is an upper capacity limit of the line (i, j);         and are an upper limit and a lower limit of a voltage at a PQ node respectively;          and         represent an upper limit and a lower limit of power generation of each generator unit respectively, and r is an upper limit of a ramp rate.

[0036] (2) An active power balance constraint and a reactive power balance constraint of the node in a limit operation state of the power system are respectively: iGtm + iiWt -iiDtm Gm        Dm i,t +0 -Qi,t

[0037] A linearized power flow constraint in the limit operation state of the power system is: NB p m = sr B2 Bm - bt)+Bj (Vi m - vm)] j=1 j * i NB Qmt =S[-B, (Bmm, — Bmt) + B2( Vi m, - Vm )] j=1

[0038] A capacity constraint of a line in the limit operation state of the power system is: max -Sij < 5.2(5- - 5;t)+B ( v, - - Vt) <S-ax -0.5S-x < -B' (5- - 5- ) + B2 ( V- -i                   ij \ i, t J, t / ij \ i, t vj-,t) < 0.5Si-j ax

[0039] A node voltage constraint in the limit operation state of the power system is: -in <vi,-t <v -ax G-in Pi QG-in G- G-      G-ax i,t,s -Pi,t+' <rPi G-ax ri G- G-< Pi,t  -Pi,t+'

[0040] An output constraint of a generator unit in the limit operation state of the power system is: G G-Pit <Pit , G G-Pit = Pit , i g SR i g SK ' p, DD = p, DD-, i g sr i [ P, D < P, D-, i g SK

[0041] P-     Q in the formulas, ,,t and ,,t are an active input power and a reactive input power of PG-      QG- the node i at the time t respectively, ,,t and ,,t are an active output and a reactive output of PW a thermal power generator unit at the node i at the time t respectively,  ,,t is an active output of PD-      Q Da wind farm at the node i at the time t; ,,t   and ,,t are an active load and a reactive load of 5-      v - the node i at the time t respectively, ,,t and ,,t are a phase angle and an amplitude of a voltage at the node i at time t respectively; and are a resistance and a reactance of a line (i, j) respectively;         is an upper capacity limit of the line (i, j);         and are an upper limit and a lower limit of a voltage at a PQ node respectively;          and         represent an upper limit and a lower limit of power generation of each generator unit respectively, and r is an upper limit of a ramp rate; SR represents a transfer end region in the power system, and SK represents a receiving end region in the power system.

[0042] As a possible implementation of this embodiment, the according to the ATC evaluation model and the typical daily source-load scenario set, determining a probability distribution of the available transfer capability of the power system with multiple regions includes as follows.

[0043] The nonlinear relationship generator receives random noises and conditional values of the power system with multiple regions as inputs, the classification signal discriminator receives wind power output curves or load curves and the conditional values as inputs, and decomposes wind power output data or load data of each wind farm into a matrix form.

[0044] The random noise, conditional values and real data of the power system with multiple regions are input into the nonlinear relationship generator and the classification signal discriminator, and fake sample data generated by the nonlinear relationship generator is input together with the real data into the classification signal discriminator for discrimination. The classification signal discriminator outputs a Wasserstein distance as a discrimination result to determine the probability distribution of the available transfer capability of the power system with multiple regions.

[0045] As a possible implementation of this embodiment, the mean value of the probability distribution of the available transfer capability of the power system with multiple regions is: ATI = 1^=1^ Q,s.

[0046] in the formula. ATCs is an available transfer capability within a day of the power system in a typical day scenario s. t is a daily time period. T is a set of daily time periods. and ATCt,s is an available transfer capability of the power system in the typical day scenario s at the time t.

[0047] In a second aspect. an apparatus for evaluating an available transfer capability of a power system with multiple regions according to embodiments of the present application includes: a source-load scenario set determination module, a limit operating point determination module, an evaluation model constructing module and an available transfer capability evaluation module.

[0048] The source-load scenario set determination module is configured to, in consideration with multi-dimensional uncertainty of a renewable energy output and load demand, determine a typical daily source-load scenario set based on the conditional generative adversarial network method.

[0049] The limit operating point determination module is configured to, according to the typical daily source-load scenario set, construct an initial operating point set, and determine a limit operating point of the power system with multiple regions.

[0050] The evaluation model constructing module is configured to, according to the initial operating point set and the limit operating point, construct an ATC evaluation model based on an operation safety index of the power system with multiple regions.

[0051] The available transfer capability evaluation module is configured to, according to the ATC evaluation model and the typical daily source-load scenario set, determine a probability distribution of the available transfer capability of the power system with multiple regions.

[0052] In a third aspect, an electronic device is provided according to embodiments of the present application, which includes a processor, a memory and a program stored in the memory and executable on the processor. When the electronic device is running, the processor, when executing the program, implements any of the steps of the method for evaluating an available transfer capability of a power system with multiple regions as described above.

[0053] In a fourth aspect, it is further provided according to embodiments of the present application a storage medium on which a program is stored. The program, when being executed by a processor, implements the steps of the method for evaluating an available transfer capability of a power system with multiple regions as described above. BRIEF DESCRIPTION OF DRAWINGS

[0054] FIG. 1 is a flow chart of a method for evaluating an available transfer capability of a power system with multiple regions shown according to an exemplary embodiment;

[0055] FIG. 2 is a block diagram of an apparatus for evaluating an available transfer capability of a power system with multiple regions shown according to an exemplary embodiment; and

[0056] FIG. 3 is a schematic diagram of the structure of a conditional generative adversarial network shown according to an exemplary embodiment. DETAILED DESCRIPTION

[0057] The present application is further described hereinafter in conjunction with the drawings and embodiments:

[0058] In order to clearly illustrate the technical features of the present solution, the present application is described in detail hereinafter through embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the present application. In order to simplify the disclosure of the present application, components and arrangements of specific examples are described below. Furthermore, the present application may repeat reference numerals and / or letters in different examples. This repetition is for the purpose of simplicity and clarity and does not in itself dictate relationships between the various embodiments and / or settings discussed. It should be noted that the components illustrated in the drawings are not necessarily drawn to scale. The present application omits descriptions of well-known components and processing techniques and processes to avoid unnecessarily limiting the present application.

[0059] As shown in FIG. 1, a method for evaluating an available transfer capability of a power system with multiple regions is provided according to embodiments of the present application, which includes the following steps:

[0060] in consideration with multi-dimensional uncertainty of a renewable energy output and load demand, determining a typical daily source-load scenario set based on the conditional generative adversarial network method;

[0061] according to the typical daily source-load scenario set, constructing an initial operating point set, and determining a limit operating point of the power system with multiple regions;

[0062] according to the initial operating point set and the limit operating point, constructing an available transfer capability (ATC) evaluation model based on an operation safety index of power system with multiple regions; and

[0063] according to the ATC evaluation model and the typical daily source-load scenario set, determining a probability distribution of the available transfer capability of the power system with multiple regions.

[0064] As a possible implementation of this embodiment, the determining a typical daily source-load scenario set based on the conditional generative adversarial network method includes as follows.

[0065] Based on the conditional generative adversarial network method, deep neural network models are used to characterize a nonlinear relationship generator and a classification signal discriminator, and conditional information is transmitted as an input layer to the classification signal discriminator and the nonlinear relationship generator. The conditional information includes: historical meteorological data with time attributes, spatial characteristics of the power system, and wind farm output characteristics and load demand characteristics.

[0066] Based on historical data of a renewable energy output and load demand, a typical daily source-load scenario set S0 is constructed: 0 = {    ,    }, V =     , P™, ..., pW], SD ={P^, P2, . , P°] ,

[0067] in the formulas:   1 represents an output of a wind farm connected to a node i, SW D represents a set of 1 in the power system, Pi  represents a load of a wind farm connected to D the node i, SD represents a set of Pi  in the power system, W represents a set of wind farm nodes in the power system, and D represents a set of load nodes in the power system.

[0068] As a possible implementation of this embodiment, the according to the typical daily source-load scenario set, constructing an initial operating point set, and determining a limit operating point of the power system with multiple regions includes:

[0069] according to the initial operating point set and based on the operation safety index of the power system with multiple regions, continuously increasing exchange power between the multiple regions of the power system by adjusting power generation and load power, so as to obtain a limit operating point of the power system with multiple regions under limit operating conditions.

[0070] As a possible implementation of this embodiment, the according to the initial operating point set and the limit operating point, constructing an ATC evaluation model based on the operation safety index of the power system includes as follows.

[0071] An objective function is established based on objectives of minimizing overall generation cost and maximizing ATC between different regions of the power system: minY acPf - ^Y (pfm - pf) i i,t                             i,t               i,t tgT _                 igG

[0072] pGm in the formula,   i,t pG and i,t are an active output of a thermal generator at a node i in a limit operation state and an active output of the thermal generator at the node i at an initial operating point respectively; G is a set of power generation nodes to be adjusted at a transfer end in the power system; ci is unit power generation cost of unit i, a and P are weight coefficients of multi-objective optimization.

[0073] A power system benchmark operation state model is constructed, and the power system benchmark operation state model includes a grid structure, a startup mode, load power, operating parameters of a power growth mode, and safety constraint conditions.

[0074] By adjusting power generation and load power, exchange power between the multiple regions is continuously increased until any safety constraint condition is exceeded, and thus, the limit operating point of the power system is obtained, and an ATC evaluation model is constructed.

[0075] As a possible implementation of this embodiment, the safety constraint conditions include as follow.

[0076] (1) An active power balance constraint and a reactive power balance constraint of the node in the benchmark operation state of the power system are respectively: GWD Pi,t = Pi,t + Pi,t - Pi,t Qi,t=QiG,t+0-QiD,t

[0077] in the formulas, Q and i,t are an active input power and a reactive input power PG QG of a node i at time t respectively,   i,t and   i,t are an active output and a reactive output of a PW thermal power generator unit at the node i at the time t respectively,   i,t is an active output of PD     QD a wind farm at the node i at the time t; i,t and i,t are an active load and a reactive load of the node i at the time t respectively, and Ai is a ratio of the reactive load to the active load.

[0078] A linearized power flow constraint in the benchmark operation state of the power system is: Pi,t Qi,t NB =Z[ Bj (s, t - S, t)+B1 (Vi, t -j=1 j * i NB =E[-$& - S,.)+B2 (Vi, . j=1 j * i V,.)] - Vj,.)]

[0079] The amplitude and phase angle of the voltage of a node can both be solved through the 12 linear equation, and an equivalent admittance 1 and 2 of the power flow equation is shown as follows: BI = nj ll „2 , v2 + D 2 _ xu r 2-+x2-. +

[0080] A capacity constraint of the line in the benchmark operation state of the power system is: _'-j < Bj (^, - Sj,. ) + B1 (Vi,. - Vh. ) < Sma -0.5S-"- < -B' (S, - S,) + B2 (V, - V,,) < 0.5sma . ij                     ij i,.           j,.              ij       i,.          j,.                       ij

[0081] A node voltage constraint in the benchmark operation state of the power system is: ymin < y. < ymax

[0082] An output constraint of a generator unit in the benchmark operation state of the power system is: Gmin G Gmax Pi    < Pi,. <Pi QGmin <QG < QGmax

[0083] The above formula indicates that at the time t, the active output of an nth generator unit must be between its upper limit and lower limit.

[0084] The following formulas represent increase or decrease in output of a generator unit per unit time respectively: G      G       Gmax i, t, s     Pi, t+1 — rPi rPi Gmax GG i,t           i,t +1

[0085] in the formulas, , and , are the phase angle and the amplitude of a voltage at the node i at the time t respectively;      and are a resistance and reactance of a line (i, j) respectively;         is an upper capacity limit of the line (i, j);         and are an upper limit and a lower limit of a voltage at a PQ node respectively;          and         represent an upper limit and a lower limit of power generation of each generator unit respectively, and r is an upper limit of a ramp rate.

[0086] (2) An active power balance constraint and a reactive power balance constraint of the node in a limit operation state of the power system are respectively: Gm W Dm Pi,t = Pi,t  + Pi,t    Pi,t Qi,t=QiG,tm+0-QiD,tm

[0087] A linearized power flow constraint in the limit operation state of the power system is: NB Pm = V B 2 (3m - 3m ) + B1 (Vm - Vm) i, t                   j \ i, t j,t J i y I, t j, t J j=1 J * i NB Qm =z - Bl (3^: - 3m)+Bj( Vi m, - vm) j=1 j*i

[0088] A capacity constraint of a line in the limit operation state of the power system is: max 2m m       1m m      max -Sij   — Bij (3i,t -3j,t) + Bij (Vi,t -Vj,t)—Sij -0.5Simjax—-Bi1j(3im,t-3jm,t)+Bi2j(Vi,mt-Vjm,t)—0.5Simjax

[0089] A node voltage constraint in the limit operation state of the power system is: min — Vi mm — V max Gmin Pi Gm Gmax < Pi,t <Pi QGmin QGmax Gm Gm      Gmax Pi, t, ^ Pi, t+1 — rPi rPGmax Gm Gm < Pi,t  - Pi,t+1

[0090] An output constraint of a generator unit in the limit operation state of the power system is: ' pG i,t PG i, t <PGm Dm =P i g SR i g SK Dm i g SR i gSK mm

[0091] in the formulas, Pi,t and Qi,t are an active input power and a reactive input power of PGm      QGm the node i at the time t respectively, i,t and i,t are an active output and a reactive output of PW a thermal power generator unit at the node i at the time t respectively,  i,t is an active output of PDm      Q Dm a wind farm at the node i at the time t; i,t   and i,t are an active load and a reactive load of ........    ...  3m the node i at the time t respectively,   i,t Vm and i,t are a phase angle and an amplitude of a voltage at the node i at time t respectively; and are a resistance and a reactance of a line (i, j) respectively;         is an upper capacity limit of the line (i, j);         and are an upper limit and a lower limit of a voltage at a PQ node respectively;          and         represent an upper limit and a lower limit of power generation of each generator unit respectively, and r is an upper limit of a ramp rate; SR represents a transfer end region in the power system, and SK represents a receiving end region in the power system.

[0092] As a possible implementation of this embodiment, the according to the ATC evaluation model and the typical daily source-load scenario set, determining a probability distribution of the available transfer capability of the power system with multiple regions includes as follows.

[0093] The nonlinear relationship generator receives random noise and conditional values of the power system with multiple regions as inputs, the classification signal discriminator receives wind power output curves or load curves and conditional values as inputs, and wind power output data or load data of each wind farm are decomposed into a matrix form.

[0094] The random noise, conditional values and real data of the power system with multiple regions are input into the nonlinear relationship generator and the classification signal discriminator respectively, and fake sample data generated by the nonlinear relationship generator is input together with the real data into the classification signal discriminator for discrimination. The classification signal discriminator outputs a Wasserstein distance as a discrimination result to determine the probability distribution of the available transfer capability of the power system with multiple regions.

[0095] As a possible implementation of this embodiment, the mean value of the probability distribution of the available transfer capability of the power system with multiple regions is: ATCS = 18=1^  ,

[0096] in the formula, ATCs is an available transfer capability within a day of the power system in a typical day scenario s, t is a daily time period, T is a set of daily time periods, and ATCt,s is an available transfer capability of the power system in the typical day scenario s at the time t.

[0097] As shown in FIG. 2, an apparatus for evaluating an available transfer capability of a power system with multiple regions according to embodiments of the present application includes: a source-load scenario set determination module, a limit operating point determination module, an evaluation model constructing module and an available transfer capability evaluation module.

[0098] The source-load scenario set determination module is configured to, in consideration with multi-dimensional uncertainty of a renewable energy output and load demand, determine a typical daily source-load scenario set based on the conditional generative adversarial network method.

[0099] The limit operating point determination module is configured to, according to the typical daily source-load scenario set, construct an initial operating point set, and determine a limit operating point of the power system with multiple regions.

[0100] The evaluation model constructing module is configured to, according to the initial operating point set and the limit operating point, construct an ATC evaluation model based on an operation safety index of the power system with multiple regions.

[0101] The available transfer capability evaluation module is configured to, according to the ATC evaluation model and the typical daily source-load scenario set, determine a probability distribution of the available transfer capability of the power system with multiple regions.

[0102] The specific process of evaluating the available transfer capability of a power system with multiple regions by using the present application is as follows.

[0103] In step S1, according to the multi-dimensional uncertainty of a renewable energy output and load demand, a typical daily source-load scenario set is determined by the conditional generative adversarial network method; the typical daily source-load scenario set is formed by using a conditional generative adversarial network model to simulate multiple factors of the renewable energy output and load demand and collecting scenarios of load and power source output that are highly representative in historical data.

[0104] Based on the conditional generative adversarial network method, deep neural network models are used to characterize a nonlinear relationship generator which can determine complex nonlinear relationships and a classification signal discriminator which can classify complex signals, and the conditional information is transmitted to the discriminator and the generator as the input layer. The generator and the discriminator in the conditional generative adversarial network are both deep neural network models, random noise vectors and conditional information are input into the generator, and then a series of nonlinear transformations and mappings are performed to generate output samples that meet the conditions. The goal of the generator is to generate realistic data samples as much as possible to deceive the discriminator. Historical real data samples and conditional information are input into the discriminator, and through a series of nonlinear transformations and classification operations, the discriminator determines whether the input data samples are real samples or fake samples generated by the generator. The goal of the discriminator is to correctly distinguish between real samples and generated samples. Conditional information mainly includes three categories: 1) historical meteorological data with time attributes; 2) spatial characteristics such as site location, topography and landform; 3) wind farm output and load demand characteristics. The conditional information is fed as the input layer into the discriminator and the generator.

[0105] The power system structure (including element parameters, on / off status), initial startup method, and load power of the power system constitute a balanced power flow solution point, which is called the initial operating point. The initial operating point of the ATC model is a set of typical scenarios with a range of parameters. In order to conduct a comprehensive evaluation on ATC, it is necessary to construct a set of typical scenarios for both the renewable energy output and load demand based on massive historical data of the renewable energy output and load.

[0106] Based on the historical data of the renewable energy output and load demand, a typical daily source-load scenario set is constructed as follows: = {    ,    }, SW ={ PW, P2W, L , PW}, i e W, S D ={ P1 D, PD, L , pD}, i e D

[0107] In the formulas:   1 represents an output of a wind farm connected to a node i, SW D represents a set of 1 in the system, Pi represents a load of a wind farm connected to the D node i, SD represents a set of Pi in the power system, W represents a set of wind farm nodes in the power system, and D represents a set of load nodes in the power system.

[0108] In step S2: according to the typical daily source-load scenario set, an initial operating point set is constructed, and a limit operating point of the power system is determined.

[0109] The difference of the exchange powers between regions in the power system at the initial operating point and the limit operating point is just ATC. According to the initial operating point set and based on the operation safety index of the power system with multiple regions, exchange power between multiple regions of the power system is continuously increased by adjusting power generation and load power, so as to obtain a limit operating point of the power system under limit operating conditions.

[0110] In an embodiment of the present application, the power system meets node balance constraints, a power flow constraint, a node voltage constraint and an output constraint of a thermal power generator unit in the benchmark operation state.

[0111] Optionally, an active power balance constraint and a reactive power balance constraint of the node in the benchmark operation state of the power system are: GWD Pi,t = Pi,t + Pi,t    Pi,t Qi,t=QiG,t+0-QiD,t

[0112] PQ in the formulas,   i,t and i are an active input power and a reactive input power PG    QG of a node i at time t respectively,   i,t and    i,t are an active output and a reactive output of PW a thermal power generator set at the node i at the time t respectively,   i,t is an active output of PD   QD a wind farm at the node i at the time t;   i,t and    i,t are an active load and a reactive load of the node i at the time t respectively, and X. is a ratio of the reactive load to the active load.

[0113] A linearized power flow constraint in the benchmark operation state of the power system is: NB P,t = Z BX^i-t - Sj.) + Bj (V - Vj.t) j=1 j * . NB Qu = Z — Bj (6., — 3j,,) + B.2 (V>, — Vj, t) j=1 j *.

[0114] The amplitude and phase angle of the voltage of a node can both be solved through the linear equation, and an equivalent admittance 1 and 2 of the power flow equation is shown as follows: B1 = nj ll r2 , „2 + B 2 = -^ lJ    r2j+xij

[0115] A capacity constraint of the line in the benchmark operation state of the power system is: -j < Bj (S,.. — 6,, ) + B ( V.. — Vj.. )< Sma -0.5Smax <-B1 (S - 6,) + B2 (V. - V ,)< 0.5Smax ij                     ij       i .t           j .t              ij       i.t           j .t                      ij

[0116] A node voltage constraint in the benchmark operation state of the power system is: ymin ^ ymax

[0117] An output constraint of a generator unit in the benchmark operation state of the power system is: Gmin G Gmax Pi    — Pi, t — Pi QGmin — qG — QGmax

[0118] where, this formula indicates that at the time t, the active output of an nth generator unit must be between its upper limit and lower limit. The following formulas represent increase or decrease in output of a generator unit per unit time respectively: PG — Gmax Pi,t+1 —rPi rPGmax GG — Pi,t    Pi,t+1

[0119] in the formulas, , and , are the phase angle and the amplitude of a voltage at the node i at the time t respectively; and are a resistance and reactance of a line (i, j) respectively;         is an upper capacity limit of the line (i, j);         and are an upper limit and a lower limit of a voltage at a PQ node respectively;          and         represent an upper limit and a lower limit of power generation of each generator set respectively, and r is an upper limit of a ramp rate.

[0120] The power system meets node balance constraints, a power flow constraint, a node voltage constraint and an output constraint of a thermal power generator unit even in the limit operation state.

[0121] An active power balance constraint and a reactive power balance constraint of the node in the limit operation state of the power system are respectively: Gm W Dm Pi,t = Pi,t  + Pi,t    Pi,t Qi,t=QiG,tm+0-QiD,tm

[0122] A linearized power flow constraint in the limit operation state of the power system is: NB Pi m=z b2^ - et)+bj (V m - Vm) j=1 j * i NB Qm =Y -Bj (¢: -Sm.t) + B^V,mm - Vjmt) j=1 j*i

[0123]

[0124] A capacity constraint of a line in the limit operation state of the power system is: f Smax — B2 / em - ¢: \ + B1 / Vm - Vm A — Smax ij                   ij i,t j,t ij i,t j,t ij s -0.5S“ < -B1 (Am - Am | + B2 (Vm - Vm) < 0.5Smax • ij                   ij y i, t J, t / j y i, t          J, t )                 ‘J A node voltage constraint in the limit operation state of the power system is: Vmin max Gmin Gm Gmax Pi    < Pi,t <Pi QGmin <QGm < QGmax Gm Gm      Gmax i,t,s - Pi,t+1 <rPi rPGmax Gm Gm ,t - Pi,t+1

[0125] is: An output constraint of a generator unit in the limit operation state of the power system <P Gm = PGm i e S^ i e SK / , D = P Dm D Dm t Pi, t — Pi, t i e SR i eSK

[0126] in the formulas, Pm Qm and i,t are an active input power and a reactive input power PGm     QGm of the node i at the time t respectively, i,t and i,t are an active output and a reactive output PW of a thermal power generator set at the node i at the time t respectively,   i,t is an active output PDm     QDm of a wind farm at the node i at the time t;   i,t   and   i,t are an active load and a reactive load of the node i at the time t respectively, Xi is the ratio of reactive load to active load, Am     vm i,t and i,t are a phase angle and an amplitude of a voltage at the node i at time t respectively; and are a resistance and a reactance of a line (i, j) respectively;         is an upper capacity limit of the line (i, j);         and are an upper limit and a lower limit of a voltage at a PQ node respectively;           and          represent an upper limit and a lower limit of power generation of each generator set respectively, and r is an upper limit of a ramp rate; SR represents a transfer end region in the power system, and SK represents a receiving end region in the power system. The output constraint formula of the generator unit in the limit operation state of the power system represents the changing relationships of the generator unit and the load between in the benchmark state and in the limit state. Since ATC is generally used to evaluate the increasing potential of interregional transfer capability, it is assumed that the increase in power generation output between the limit operating point and the basic operating point is completely provided by the thermal power generator units at the transfer end, and the output of the thermal power generator units at the receiving end does not change. In another aspect, the increase in load between the limit operating point and the basic operating point is completely caused by the load at the receiving end, while the load at each node at the transfer end remains unchanged.

[0127] In step S3: according to the initial operating point set and the limit operating point, an ATC evaluation model is constructed based on an operation safety index of power system with multiple regions.

[0128] In step S31, an objective function is established based on objectives of minimizing overall generation cost and maximizing ATC between different regions of the power system.

[0129] The objective function under the limit operating condition is: min X acP, - ^(PGm — PiG ) t^T _                i^G                  _ , Gm     PG

[0130] in the formula, i,t and i,t are an active output of a thermal generator at the node i in a limit operation state and an active output of the thermal generator at the node i at an initial operating point respectively; G is a set of power generation nodes to be adjusted at a transfer end; ci is unit power generation cost of unit i, a and P are weight coefficients of multi-objective optimization respectively.

[0131] In step S32, a power system benchmark operation state model is constructed, which includes a grid structure, startup mode, load power, operating parameters of power growth mode, and safety constraint conditions.

[0132] In step S33, by adjusting power generation and load power, exchange power between regions of the power system is continuously increased until any safety constraint condition is exceeded, and thus, the limit operating point of the power system is obtained, and an ATC evaluation model is constructed.

[0133] In step S4, according to the ATC evaluation model and the typical daily source-load scenario set, a probability distribution of the available transfer capability of the power system with multiple regions is determined.

[0134] According to the typical daily source-load scenario set, a calculation formula of mean value of the probability distribution of the available transfer capability of the power system with multiple regions is: atcs = Ig=v4r  ,

[0135] in the formula, ATCs is an available transfer capability within a day of the power system in a typical day scenario s, t is a daily time period, T is a set of daily time periods, and ATCt,s is an available transfer capability of the system in the typical day scenario s at the time t.

[0136] The generator receives random noises and conditional values of the power system with multiple regions as inputs, while the discriminator receives wind power output curves or load curves and conditional values as inputs, and the wind power output data or load data of each wind farm are decomposed into a matrix of size NXT.

[0137] One-hot encoding is used to convert discrete data in the power system with multiple regions into binary vectors, and conditional information is introduced into the training process of the generator and discriminator to ensure that both the generator and the discriminator can obtain conditional information during training to generate and discriminate sample data.

[0138] The random noise, conditional values and real data of the power system with multiple regions are input into the generator and the discriminator respectively, and the fake sample data generated by the generator is input together with the real data into the discriminator for discrimination. The discriminator outputs a Wasserstein distance whose size is used as a discrimination result to determine the probability distribution of the available transfer capability of the power system with multiple regions.

[0139] As shown in FIG. 3, the present application provides a data-driven controllable generation method for scenarios based on the conditional generative adversarial network (CGAN). It aims to set up a two-person zero-sum game based on the minimax theorem between the generator neural network and the discriminator neural network. In the training process of each time, the generator continuously updates its weights to generate "fake" samples in an attempt to "deceive" the discriminator network, while the discriminator tries to distinguish between real historical samples and generated samples. This training process will continue until the discriminator can no longer determine whether the output result of the generator is real. CGAN is to, based on the generative adversarial network, transmit additional conditional information to the discriminative model and the generative model as part of the input layer.

[0140] It is assumed that observation values xj (j = 1, ..., N) of renewable energy at time t^ T are used for each power farm. The real historical data distribution is represented by , is unknown and it is difficult to model and solve . It is assumed that a group of noise vector input Z that obeys a known distribution can be obtained (which can be obtained by the joint Gaussian method), and denoted as ~  . Converting the sample Z extracted from to make it obey the (historical data) distribution requires training simultaneous / both, that is, training the generator network and the discriminator network at the same time / both. Let G denote the generating function parameterized by 0 ^^, denoted as G (Z; 0 ^^ ) ; let D denote the generating function parameterized by 0 l) , it is written as D(x',0 l) ) . Here, 0 G and 0 ^^ are weights of the two neural networks respectively.

[0141] Generator: the generator is continuously trained to obtain a batch of random variable Z as input and then output the real scenario through a series of up-sampling operations. It is assumed that Z is a random variable that obeys the Pz distribution, then G(Z;0 G ) is a new random variable, and its distribution is expressed as .

[0142] Discriminator: the discriminator acquires samples from real historical data and, by using another deep neural network to perform a series of down-sampling operations, the discriminator outputs a continuous value for measuring how similar the input sample is to ^. The discriminator can be expressed as: Preal = D(x;0 (D) ),

[0143] in the formula: D(x; 0 ^^ ) represents the generating function parameterized by 0 (D) ; X is taken from the historical data Pt. The discriminator is continuously trained to discriminate and distinguish between and , and maximize the difference between [ ( )] (real data) and E[ ( ( ))] (generated data).

[0144] After defining the discriminator and generator, it is required to formulate loss functions (denoted by and respectively) for the generator and discriminator. The smaller the value of , the more realistic the samples generated by the generator are from the perspective of the discriminator, and the smaller the value of , the stronger the discriminator's capability to distinguish between generated scenarios and historical scenarios is. and are: = -EZ[ ( ( ))],     = -EX[ ( )] +    [ ( ( ))],

[0145] in the formula: E represents the mathematical expectation.

[0146] In order to set up the game between the generator and the discriminator, the minimum and maximum objectives of the game are described by using the Wasserstein distance. In CGAN, two random variables ( ( )) and ( ( ( ))) are acquired and are made close to each other. The Wasserstein distance is defined as: ( ) ( ) =EX[ ( | )] -EZ[ ( ( | ))],

[0147] in the formula: y is conditions of different classes, EX[ ( | )] is the expected value of the random variable PX ( ( )) , E [ ( | )] is the expected value of the random variable PZ ( ( ( ))) , and the class labels are assigned according to the user-defined classification indicators. The class labels are just representations of the sample event reflected by the distribution of power generation data. CGAN should be able to learn the conditional distribution and measure generated samples based on any given meaningful conditions.

[0148] The specific steps for conditional scenario generation are as follows.

[0149] The generator accepts random noises and conditional values as inputs, while the discriminator accepts wind power output curve or load curve and the conditional values as inputs. The wind power output data or load data of each wind farm is decomposed into a matrix of size N*T. For wind power and load, five numbers and three numbers are selected as classification labels respectively, and one-hot encoding is used to convert these discrete classification labels into binary vectors. This method can effectively introduce conditional information into the training process of the generator and discriminator, thus ensuring that both the generator and the discriminator can obtain conditional information during training, so as to better generate and discriminate sample data.

[0150] Subsequently, these label values are horizontally concatenated into the historical data matrix to input the real data and conditional values into the discriminator as a training set for training. Similarly, Gaussian noises randomly sampled from the normal distribution are set to have the same dimension, i.e., N*T, and the Gaussian noises are horizontally concatenated with the one-hot encoded label values, and the concatenated result is fed as input into the generator for training. In each training, an appropriate number of batches containing noises, conditional values, and real data and conditional values are selected and fed into the generator and discriminator respectively. The fake sample data generated by the generator is input together with the real data into the discriminator for discrimination. The discriminator outputs the size of the Wasserstein distance as its discrimination result. In order to improve the accuracy of the discriminator network and reduce the number of updates of network parameters, to make the training process more stable, it chooses to train the generator once after every four times of training the discriminator in the process of training. This alternating updating method can balance the training of the generator and the discriminator and ensure that they influence each other to achieve better training results. Through continuous iterative training, the Wasserstein distance will gradually approach 0, and at this point, the generator can more accurately generate wind power output or load scenarios under different conditions.

[0151] Aiming at the multi-dimensional uncertainty of the renewable energy output and load demand, in the present application, the conditional generative adversarial network method is used to generate and screen typical daily scenario sets, and then the initial operating point set of the power system is constructed; a system limit operating point calculation method considering the operation safety index of power system is proposed, and an ATC evaluation model is constructed; and according to the generated scenario set, the probability distribution of the available transfer capability is obtained, which provides an important basis for the expansion planning of new power systems and the power market trading mechanism.

[0152] In the present application, a method for evaluating an available transfer capability of a power system with multiple regions is proposed based on multi-dimensional uncertainty of renewable energy output and load demand caused by large-scale renewable energy grid connection; the typical daily source-load scenario set established by the conditional generative adversarial network can cover various possible operation scenarios, and thus the operation of the power system can be truly reflected, which improves the accuracy of evaluation of the available transfer capability of the power system with multiple regions. Moreover, the ATC of the renewable energy transfer section can be accurately and efficiently calculated online, which improves the renewable energy consumption capacity on the premise of maintaining safe and stable operation of the system; the ATC evaluation model takes minimizing the overall generation cost of thermal power plants and maximizing ATC between connection lines as a multi-objective function, and is carried out under a group of scenarios of random renewable energy output and random load demand. Under the condition of operation uncertainty, the transfer margins of the connecting lines can be fully evaluated, which overcomes the problem of being too conservative in the traditional available transfer capability evaluation method.

[0153] An electronic device is provided according to embodiments of the present application, which includes a processor, a memory and a program stored in the memory and executable on the processor. When the electronic device is running, the processor, when executing the program, implements any of the steps of the method for evaluating an available transfer capability of a power system with multiple regions as described above.

[0154] Optionally, the above-mentioned memory and processor can be general-purpose memory and processor, which are not specifically limited here. The processor, when running the computer program stored in the memory, can implement the method for evaluating an available transfer capability of a power system with multiple regions described above.

[0155] Corresponding to the method for starting the above-mentioned application program, it is further provided according to embodiments of the present application a storage medium on which a program is stored. The program, when being executed by a processor, implements the steps of the method for evaluating an available transfer capability of a power system with multiple regions as described above.

[0156] The application startup apparatus provided in the embodiment of the present application can be specific hardware on the device or software or firmware installed on the device. The implementation principle and technical effects of the apparatus provided in this embodiment of the present application are the same as those of the method embodiments described above. For the sake of brief description, for matters not mentioned in the apparatus embodiment, reference may be made to the method embodiments described above for corresponding contents. The person skilled in the art can clearly understand that, for the convenience and brevity of description, reference can be made to the corresponding processes in the above method embodiments for the specific working processes of all the systems, apparatuses and units described above, which will not be repeated here.

[0157] The embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0158] The above embodiments are only used to illustrate the technical solutions of the present application rather than limiting it. Although the present application has been described in detail with reference to the above embodiments, the person of ordinary skills in the art should understand that the embodiments of the present application can still be modified or equivalently replaced. Any modification or equivalent replacement that does not depart from the spirit and scope of the present application should be covered within the scope of protection of the claims of the present application.

Claims

1. A method for evaluating an available transfer capability of a power system with multiple regions, comprising:in consideration with multi-dimensional uncertainty of a renewable energy output and load demand, determining a typical daily source-load scenario set based on a conditional generative adversarial network method;according to the typical daily source-load scenario set, constructing an initial operating point set, and determining a limit operating point of the power system with multiple regions;according to the initial operating point set and the limit operating point, constructing an available transfer capability (ATC) evaluation model based on an operation safety index of the power system with multiple regions; andaccording to the ATC evaluation model and the typical daily source-load scenario set, determining a probability distribution of the available transfer capability of the power system with multiple regions, wherein the probability distribution of the available transfer capability represents a remaining power transfer capacity on the premise of ensuring a safe and stable operation of the power system, and the probability distribution of the available transfer capability is used for the power system to measure a safety and stability margin of a current operating point of the power system;wherein the determining a typical daily source-load scenario set based on a conditional generative adversarial network method comprises:based on the conditional generative adversarial network method, using deep neural network models to characterize a nonlinear relationship generator and a classification signal discriminator, and transmitting conditional information as an input layer to the classification signal discriminator and the nonlinear relationship generator;wherein the conditional information comprises: historical meteorological data with time attributes, spatial characteristics of the power system, and wind farm output characteristics and load demand2024205215   22 Jul 2026characteristics;based on historical data of a renewable energy output and load demand, constructing a typicaldaily source-load scenario set S0:= {sW,sD},= {^ z^'. -, pW},$D = {P?, P?, -, P?},wherein: P^represents an output of a wind farm connected to a node i, SW represents a set ofP™ in the power system, PD represents a load of a wind farm connected to the node i, D° represents a set of P® in the power system, W represents a set of wind farm nodes in the power system, and D represents a set of load nodes in the power system;wherein the according to the initial operating point set and the limit operating point, constructing an ATC evaluation model based on an operation safety index of the power system with multiple regions comprises:establishing an objective function based on objectives of minimizing overall generation cost and maximizing ATC between different regions of the power system:min ^ acrf - f£(P' - PG) teT _                i^Gwherein,GmPi,tPG and i,tare an active output of a thermal generator at a node i in a limitoperation state and an active output of the thermal generator at the node i at an initial operating point respectively; G is a set of power generation nodes to be adjusted at a transfer end in the power system; ci is unit power generation cost of unit i, a and P are weight coefficients of multiobjective optimization;constructing a power system benchmark operation state model, wherein the power system2024205215   22 Jul 2026benchmark operation state model comprises a grid structure, a startup mode, load power, operating parameters of a power growth mode, and safety constraint conditions; andby adjusting power generation and load power, continuously increasing exchange power between the multiple regions until any safety constraint condition of the safety constraint conditions is exceeded, to obtain the limit operating point of the power system and construct the ATC evaluation model; andwherein the according to the ATC evaluation model and the typical daily source-load scenario set, determining a probability distribution of the available transfer capability of the power system with multiple regions comprises:receiving, by the nonlinear relationship generator, random noise and conditional values of the power system with multiple regions as inputs, receiving, by the classification signal discriminator, wind power output curve or load curve and the conditional values as inputs, and decomposing wind power output data or load data of each wind farm into a matrix form; andinputting the random noise, the conditional values and real data of the power system with multiple regions into the nonlinear relationship generator and the classification signal discriminator, and inputting fake sample data generated by the nonlinear relationship generator together with the real data into the classification signal discriminator for discrimination, and outputting, by the classification signal discriminator, a Wasserstein distance as a discrimination result to determine the probability distribution of the available transfer capability of the power system with multiple regions.

2. The method for evaluating an available transfer capability of a power system with multiple regions according to claim 1, wherein the according to the typical daily source-load scenario set, constructing an initial operating point set, and determining a limit operating point of the power system with multiple regions comprises:according to the initial operating point set and based on the operation safety index of the power system with multiple regions, continuously increasing exchange power between the multiple2024205215   22 Jul 2026regions of the power system by adjusting power generation and load power, to obtain a limit operating point of the power system with multiple regions under limit operating conditions.

3. The method for evaluating an available transfer capability of a power system with multiple regions according to claim 1, wherein the safety constraint conditions comprise:(1) an active power balance constraint and a reactive power balance constraint of a node in a benchmark operation state of the power system respectively:'p = PG+pW - PiD ' Q., = QG + 0 - QDDqD = a. pd, i,, i i,, ,wherein,pi,,and Qi,,are an active input power and a reactive input power of a node i at timePG t respectively,   i,,QG and    i,,are an active output and a reactive output of a thermal powerPWgenerator unit at the node i at the time t respectively, i,, is an active output of a wind farm atPD    QDthe node i at the time t;   i,, and i,, are an active load and a reactive load of the node i at thetime t respectively, and A / is a ratio of the reactive load to the active load;a linearized power flow constraint in the benchmark operation state of the power system:NBPu =Z[ jv s„)+BM, - j)]<NBQ,=E[-BM,-p„)+b;(v, - pt)] j=1 j *‘wherein an amplitude and a phase angle of a voltage of a node are both solved through a linear.-          i                i . i                n 1       < n2                        riequation, and an equivalent admittance B-j and Bof the power flow equation is:2024205215   22 Jul 2026B * = rij u ij „2 , v2 ' ij + xijR 2 = -^2-ij    r-+x-,'ij^xija capacity constraint of the line in the benchmark operation state of the power system is:[-Smax < B2 S, - S. t) + B1 (V t - V. t) < Smaxj              j \ i ,t j ,t / ij y i ,t j ,t / ij1-0.5Smax <-B1 S -S +B2 V -V   <0.5Smax ,ij                    j y 1 ,t          j ,t /          ij y i ,t          j ,t /                  ija node voltage constraint in the benchmark operation state of the power system is:Vmin < y.,t < Vmax,n output constraint ofa generator unit in the benchmark operation state of the power system is:GminPiGi,t iGmaxQGmin< QG < QGmaxwherein theabove formula indicates thatat time t, theactive output ofan nth generator unit must be between its upper limitand lower limit,the following formulas represent increase or decrease in output ofa generator unit per unit timerespectively:G i,t,sPG i,t +1< rPiGmax-rPGmax ri< PG i,tPG i,t +1wherein 8[ t and Vi t are a phase angle and an amplitude of a voltage at the node i at the time t respectively; T^j and Xjj are a resistance and a reactance of a line (i, j) respectively; Sm1^ is an upper capacity limit of the line (i, j); ymax and ymin are an upper limit and a lower limit of a voltage at a PQ node respectively; Pmmxx and PiGmin represent an upper limit and a lower limit of power generation of each generator unit respectively, and r is an upper limit of a ramp rate; and2024205215   22 Jul 2026(2) an active power balance constraint and a reactive power balance constraint of the node in a limit operation state of the power system respectively are:f Pi, t Gm = Pi , t + PiW,t Dm i,t 1 L Qi, t Gm = Qi,t +0- Dm Qi,ta linearized power flow constraint in the limit operation state of the power system is:mPi,tNBZ[ BJ(smt- sz)+Bj (vm - j.)] j=1J * iQim,tNBS[-Bj(s,mt- sjm.)+B^,': - Vj’)]J=jJ*,a capacity constraint of a line in the limit operation state of the power system is:S max< B 2 S' - S ” )+ B1 (V ” - V” ) < S'ax,J ,,. J,. ,J ,,. J,. ,J-0.5Smax <-Bj sm -sm  +B2 Vm -Vm <0.5Smax ,,J                       ,J        ,,.            J ,. ,J        ,,.             J,.                         ,Ja node voltage constraint in the limit operation state of the power system is:m,n< V,,m. < VmaxGm,nPiQ^minGm Gmax < P,,. < P,< QGm < QGmaxPGm ,,.,sGmP,,.+j< rP, Gmax-rPGmax< P,G.mGmP,,.+jan output constraint of a generator unit in the limit operation state of the power system is:PG,,.< P Gm ,,.Gm= P,,.i G SRi g SK2024205215   22 Jul 2026wherein,mPi,tandQim,tD DmPi,t = Pi,tD Dm^ Pit — Pit ,i e SRi e SKare an active input power and a reactive input power of the node i atPGm     QGmthe time t respectively, i,t and   i,t are an active output and a reactive output of a thermalPWpower generator unit at the node i at the time t respectively,   i,t  is an active output of a windfarm at the node i at the time t;PDm     QDmi,t and     i,tare an active load and a reactive load of thenode i at the time t respectively,m^i,tVm and i,tare a phase angle and an amplitude of a voltageat the node i at time t respectively; r^y and X / y are a resistance and a reactance of a line (i, j)respectively; mJXxx is an upper capacity limit of the line (i, j); mmax and mmin are an upper limit and a lower limit of a voltage at a PQ node respectively; p?™-11-*- and P / Gm / n represent an upper limit and a lower limit of power generation of each generator unit respectively, and r is an upper limit of a ramp rate; SR represents a transfer end region in the power system, and SK represents a receiving end region in the power system.

4. The method for evaluating an available transfer capability of a power system with multiple regions according to any one of claims 1 to 3, wherein a mean value of the probability distribution of the available transfer capability of the power system with multiple regions is:1ATCs =-^1=1 ATCt,s,wherein, ATCs is an available transfer capability within a day of the power system in a typical day scenario s, t is a daily time period, T is a set of daily time periods, and ATCt,s is an available transfer capability of the power system in the typical day scenario s at the time t.

5. An apparatus for evaluating an available transfer capability of a power system with multiple regions, comprising:2024205215   22 Jul 2026a source-load scenario set determination module configured to, in consideration with multidimensional uncertainty of a renewable energy output and load demand, determine a typical daily source-load scenario set based on a conditional generative adversarial network method;a limit operating point determination module configured to, according to the typical daily sourceload scenario set, construct an initial operating point set, and determine a limit operating point of the power system with multiple regions;an evaluation model constructing module configured to, according to the initial operating point set and the limit operating point, construct an available transfer capability (ATC) evaluation model based on an operation safety index of the power system with multiple regions; andan available transfer capability evaluation module configured to, according to the ATC evaluation model and the typical daily source-load scenario set, determine a probability distribution of the available transfer capability of the power system with multiple regions, wherein the probability distribution of the available transfer capability represents a remaining power transfer capacity on the premise of ensuring a safe and stable operation of the power system, and the probability distribution of the available transfer capability is used for the power system to measure a safety and stability margin of a current operating point of the power system;wherein the source-load scenario set determination module is further configured to:based on the conditional generative adversarial network method, use deep neural network models to characterize a nonlinear relationship generator and a classification signal discriminator, and transmit conditional information as an input layer to the classification signal discriminator and the nonlinear relationship generator;wherein the conditional information comprises: historical meteorological data with time attributes, spatial characteristics of the power system, and wind farm output characteristics and load demand characteristics;based on historical data of a renewable energy output and load demand, construct a typical dailysource-load scenario set 5°:2024205215   22 Jul 2026$W = {p^, P2W, -, / f}sD = {p?, P2^f -, p?},wherein: P^represents an output of a wind farm connected to a node i, SW represents a set of P™ in the power system, PD represents a load of a wind farm connected to the node i, D° represents a set of PD) in the power system, W represents a set of wind farm nodes in the power system, and D represents a set of load nodes in the power system;wherein the evaluation model constructing module is further configured to:establish an objective function based on objectives of minimizing overall generation cost and maximize ATC between different regions of the power system:min£ acP -^(PT — PG) teT L                i^Gwherein,GmPi,tPGand i,tare an active output of a thermal generator at a node i in a limitoperation state and an active output of the thermal generator at the node i at an initial operating point respectively; G is a set of power generation nodes to be adjusted at a transfer end in the power system; Ci is unit power generation cost of unit i, a and 0 are weight coefficients of multiobjective optimization;construct a power system benchmark operation state model, wherein the power system benchmark operation state model comprises a grid structure, a startup mode, load power, operating parameters of a power growth mode, and safety constraint conditions; andby adjusting power generation and load power, continuously increase exchange power between the multiple regions until any safety constraint condition of the safety constraint conditions is exceeded, to obtain the limit operating point of the power system and construct the ATC evaluation model; andwherein the available transfer capability evaluation module is further configured to:2024205215   22 Jul 2026receive, by the nonlinear relationship generator, random noise and conditional values of the power system with multiple regions as inputs, receive, by the classification signal discriminator, wind power output curve or load curve and the conditional values as inputs, and decompose wind power output data or load data of each wind farm into a matrix form; andinput the random noise, the conditional values and real data of the power system with multiple regions into the nonlinear relationship generator and the classification signal discriminator, and input fake sample data generated by the nonlinear relationship generator together with the real data into the classification signal discriminator for discrimination, and output by the classification signal discriminator, a Wasserstein distance as a discrimination result to determine the probability distribution of the available transfer capability of the power system with multiple regions.

6. An electronic device, comprising a processor, a memory and a program stored in the memory and executable on the processor, wherein when the electronic device is running, the processor, when executing the program, implements the method for evaluating an available transfer capability of a power system with multiple regions according to any one of claims 1 to 4.

7. A storage medium on which a program is stored, wherein the program, when being executed by a processor, implements the method for evaluating an available transfer capability of a power system with multiple regions according to any one of claims 1 to 4.

Images