Two-stage risk scheduling method and device for active power distribution network, medium and product

Abstract

The invention discloses an active power distribution network two-stage risk scheduling method and device, a medium and a product, and relates to the technical field of power distribution network scheduling, and the method comprises the steps: constructing a source-load typical scene set considering prediction error space-time correlation; constructing an active power distribution network two-stage risk scheduling model based on the source-load typical scene set; the active power distribution network two-stage risk scheduling model comprises a day-ahead scheduling stage model and an intra-day scheduling stage model; and converting the two-stage risk scheduling model of the active power distribution network from a mixed integer nonlinear programming problem to a mixed integer second-order cone programming problem through a second-order cone relaxation method, and solving the two-stage risk scheduling model of the active power distribution network to obtain a scheduling scheme of the active power distribution network. The method can effectively cooperate with the source-grid-load storage to flexibly adjust the resources so as to reduce the power grid dispatching operation risk caused by the source-load uncertainty.

Description

Technical Field

[0001] The present application relates to the technical field of distribution network dispatching, and particularly to a two-stage risk dispatching method, device, medium and product for an active distribution network. Background Technique

[0002] In an active distribution network, the uncertainties of wind power - photovoltaic output and load demand, that is, the errors in day-ahead prediction data, lead to high grid operation risks, such as voltage violation risks and power violation risks. In the current risk dispatching of active distribution networks, a stochastic optimization risk dispatching model is usually constructed based on the scenario method. However, the spatio-temporal correlation of source-load power prediction errors is less considered when generating typical operation scenarios of sources and loads, so the source-load uncertainties cannot be accurately described. At the same time, the existing risk dispatching models rarely consider the collaborative optimization of resources on each side of the source, network, load, and storage in the distribution network. There is little flexibility in adjusting the network topology based on the capabilities of the network-side intelligent soft switch (Softopenpoint, SOP) and network dynamic reconfiguration, the continuous power adjustment of SOP, and the provision of reactive voltage support, etc., making it impossible to reduce the system dispatching operation risks brought by source-load uncertainties. Summary of the Invention

[0003] The purpose of the present application is to provide a two-stage risk dispatching method, device, medium and product for an active distribution network, which can effectively coordinate the flexible adjustment resources of the source, network, load, and storage to reduce the grid dispatching operation risks caused by source-load uncertainties.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] In the first aspect, the present application provides a two-stage risk dispatching method for an active distribution network, including:

[0006] Construct a typical scenario set of sources and loads considering the spatio-temporal correlation of prediction errors;

[0007] Construct a two-stage risk scheduling model for active distribution networks based on a set of typical source-load scenarios; the two-stage risk scheduling model for active distribution networks includes: a day-ahead scheduling stage model and an intra-day pre-scheduling stage model; the day-ahead scheduling stage model includes: a day-ahead scheduling stage objective function and day-ahead scheduling stage constraints. The day-ahead scheduling stage objective function is a function aiming at minimizing the daily comprehensive operation loss value of the active distribution network. The day-ahead scheduling stage constraints include: day-ahead power output constraints, network dynamic reconfiguration constraints, day-ahead SOP constraints, OLTC constraints, CB constraints, day-ahead SVC constraints, day-ahead DR constraints, day-ahead ESS constraints, day-ahead distribution network power flow constraints, and day-ahead system security constraints; the intra-day scheduling stage model includes: an intra-day scheduling stage objective function and intra-day scheduling stage constraints. The intra-day scheduling stage objective function is a function aiming at minimizing the weighted sum of the daily comprehensive operation loss value considering multiple scenarios and the loss value of high-risk scenarios of the active distribution network. The intra-day scheduling stage constraints include: intra-day power output constraints, intra-day SOP constraints, intra-day SVC constraints, intra-day DR constraints, intra-day ESS constraints, intra-day distribution network power flow constraints, and intra-day system security constraints;

[0008] By means of the second-order cone relaxation method, the two-stage risk scheduling model for active distribution networks is transformed from a mixed-integer non-linear programming problem into a mixed-integer second-order cone programming problem, and the two-stage risk scheduling model for active distribution networks is solved to obtain a scheduling plan for the active distribution network.

[0009] In a second aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor executes the computer program to implement the two-stage risk scheduling method for active distribution networks described in any one of the above.

[0010] In a third aspect, the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the two-stage risk scheduling method for active distribution networks described in any one of the above.

[0011] In a fourth aspect, the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the two-stage risk scheduling method for active distribution networks described in any one of the above.

[0012] According to the specific embodiments provided by the present application, the following technical effects are disclosed in the present application:

[0013] The present application discloses a two-stage risk dispatching method, device, medium and product for active distribution networks. First, a typical source-load scenario set considering the spatio-temporal correlation of prediction errors is constructed; then, a two-stage risk dispatching model for active distribution networks is constructed based on the typical source-load scenario set; finally, through the method of second-order cone relaxation, the two-stage risk dispatching model for active distribution networks is transformed from a mixed-integer non-linear programming problem into a mixed-integer second-order cone programming problem, and the two-stage risk dispatching model for active distribution networks is solved to obtain the dispatching scheme for active distribution networks. The present application constructs a two-stage risk dispatching model for active distribution networks based on the typical source-load scenario set, accurately describes the source-load uncertainty caused by prediction errors, and can effectively coordinate the flexible adjustment resources of the source network, load and storage to reduce the grid dispatching operation risk caused by source-load uncertainty. Description of the Drawings

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0015] Figure 1 Schematic diagram of the process of the two-stage risk dispatching method for active distribution networks provided by an embodiment of the present application;

[0016] Figure 2 Schematic diagram of the probability density curve of the node voltage amplitude;

[0017] Figure 3 Schematic diagram of the framework of the two-stage risk dispatching model for active distribution networks;

[0018] Figure 4 Schematic diagram of the structure of a computer device provided by an embodiment of the present application. Detailed Embodiments

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0020] The purpose of the present application is to provide a two-stage risk dispatching method, device, medium and product for active distribution networks, aiming to effectively coordinate the flexible adjustment resources of the source network, load and storage and reduce the grid dispatching operation risk caused by source-load uncertainty.

[0021] To make the above objects, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] In an exemplary embodiment, as Figure 1 shown, a two-stage risk scheduling method for an active distribution network is provided. Before performing the two-stage risk scheduling method for the active distribution network, first analyze using S1.

[0023] S1 includes: The operation risk of the power grid is usually comprehensively measured from two aspects: "the probability of the occurrence of risk events" and "the severity of the harm caused to the power grid after the occurrence of risk events" to quantitatively evaluate the safety level of the power grid operation state. The operation risks considered in the present application are: the high over-limit risk of the day-ahead scheduling plan of the power grid caused by the uncertainty of wind power-photovoltaic output and load demand in the active distribution network (the error of day-ahead prediction data). Among them, the over-limit risk includes voltage over-limit risk and power over-limit risk. Therefore, a risk scheduling model based on stochastic optimization (SO) is adopted for risk defense of the active distribution network to reduce the potential risk of the day-ahead scheduling plan of the power grid caused by source-load uncertainty. That is, first introduce the conditional value-at-risk theory (CVaR) to quantify low-probability, high-risk (voltage and power over-limit risks) operation scenarios. By increasing the weight of high-risk scenarios through CVaR, the tolerance of the risk scheduling plan of the active distribution network to such scenarios can be improved; secondly, construct voltage and power over-limit risk indicators to quantitatively evaluate the voltage and power over-limit risks of the risk scheduling plan of the active distribution network. Specifically include:

[0024] S1.1: Quantification of high-risk scenarios based on CVaR.

[0025] The present application uses scenario-based SO to address the operation risks brought by source-load uncertainty in the active distribution network and performs risk scheduling on two time scales of day-ahead and intra-day. For the problem that individual extreme high-risk scenarios may be ignored due to their low probability, resulting in potential large risks in the power grid scheduling plan, the present application introduces CVaR to quantify low-probability, high-risk operation scenarios. VaR represents the potential maximum loss value under a given confidence level β and is related to the tail loss probability; CVaR represents the expected value of the loss exceeding VaR under a given confidence level β and is related to the expectation of the tail loss.

[0026] Let L(x, y) be the loss function, x be the decision variable, y be the random variable, and assume f(y) is the probability density function of the random variable y. Then the VaR and CVaR of the loss can be expressed as:

[0027] η β[L(x, y)] = min{η ∈ R, ∫ L(x,y)≤η f(y)dy ≥ β}(1)

[0028] C β [L(x, y)] = E[L(x, y)|L(x, y) ≥ η β [L(x, y)]](2)

[0029] where η β [L(x, y)] is the VaR value of the loss function L when the confidence level is set to β; C β [L(x, y)] is the CVaR value of the loss function L when the confidence level is set to β; η is the optimization variable, and its optimized value is the VaR value; E[·] is the mathematical expectation.

[0030] As can be seen from Equation (2), to calculate the CVaR value of the loss function, the VaR value needs to be obtained first. The calculation process is complex, and the VaR value lacks convexity, making it difficult to handle in the optimization process. Therefore, the expected value of the truncated distribution is used here to approximate the CVaR, resulting in:

[0031]

[0032] Further considering the discrete scenarios generated by the random variable, the CVaR expression based on the discrete random variable can be obtained:

[0033]

[0034] where C' β [L(x, y)] represents the approximate value of the CVaR; N * represents the number of scenarios; p s represents the probability of scenario s; y s represents the value of the random variable corresponding to scenario s; η β is the optimization variable.

[0035] CVaR quantifies the low-probability, high-risk scenarios in the optimization scenario. From the perspective of optimization, its essence is to defend against risks by increasing the weights of low-probability, high-risk scenarios. The loss value of the high-risk scenario based on CVaR can be added to the objective function of the scheduling model to reduce risks.

[0036] S1.2: Calculation of the over-limit risk index.

[0037] The over-limit risk index of the active distribution network voltage and power can be calculated by the product of the over-limit scenario probability and its severity. Taking the voltage over-limit as an example below, the calculation process of the over-limit risk index is illustrated.

[0038] 1. Assume that the voltage amplitude of node i at time t is a continuous random variable Vi,t , V i,t The probability density curve of is f(V i,t ), and the schematic diagram of f(V i,t ) is as shown in Figure 2 . Figure 2 Among them: V i,t,av respectively represent the average over-limit voltage and the average under-limit voltage of node i in the t period, which are used to characterize the average degree of voltage over-limit; V i,t,max respectively represent the maximum over-limit voltage and the minimum under-limit voltage of node i in the t period, which are used to characterize the most serious degree of voltage over-limit; V N represents the rated voltage.

[0039] Since the typical operation scenarios of source-load on which risk dispatching is based are all discrete scenarios, when V i,t is a discrete random variable, V i,t,av is calculated as shown in Equation (5), V i,t,max is calculated as shown in Equation (6):

[0040]

[0041]

[0042] Equations (5)-(6) can characterize the average degree and the most serious degree of voltage over-limit.

[0043] 2. Calculate the voltage over-limit risk index by combining the average voltage over-limit risk and the maximum voltage over-limit risk:

[0044]

[0045]

[0046]

[0047] Among them, R i,t,av,VR , R i,t,max,VR respectively represent the average voltage over-limit risk and the maximum voltage over-limit risk of node i in the t period; R i,t,VR represents the voltage over-limit risk index of node i in the t period. It can be seen from Equation (9) that the voltage over-limit risk index is obtained by the product of the over-limit scenario probability and the average and maximum over-limit degrees, and can effectively evaluate the voltage over-limit risk of the power grid.

[0048] 3. Starting from the perspectives of the average value and the maximum value, combine the voltage over-limit risk indexes of all nodes and all periods to obtain the system voltage over-limit risk index:

[0049]

[0050]

[0051] Among them, R t,VR represents the system voltage over-limit risk index at time period t; R VR represents the system voltage over-limit risk index; ω n is the set of distribution network node sets.

[0052] 4. By analogy with 1-3 in S1, the system power over-limit risk index R PR can be calculated. This application believes that power over-limit occurs when the branch power is greater than 0.9 of the power limit. Combining the system voltage over-limit risk index and the system power over-limit risk index, the system over-limit risk index can be obtained:

[0053] R = R VR + R PR (12)

[0054] Among them, R is the system over-limit risk index that comprehensively considers the voltage, the probability of the grid over-limit scenario, and its severity.

[0055] As Figure 1 shown, the two-stage risk dispatching method for the active distribution network in this embodiment includes:

[0056] Step 1: Construct a typical source-load scenario set considering the spatio-temporal correlation of prediction errors.

[0057] As an optional implementation manner, Step 1 includes:

[0058] Step 11: Use the historical output data and prediction data of wind power and photovoltaic power to statistically obtain the change curves of the prediction errors of wind power and photovoltaic power output and the change curves of the standard deviations of the prediction errors of wind power and photovoltaic power output at different time periods, and determine the source-load prediction error distribution considering the time sequence correlation based on the change curves of the prediction errors of wind power and photovoltaic power output and the change curves of the standard deviations of the prediction errors of wind power and photovoltaic power output at each time period.

[0059] Step 12: Based on the source-load prediction error distribution, use the Latin hypercube sampling method to generate multiple source-load operation scenarios for each time period to obtain the wind power and photovoltaic operation scenario set.

[0060] Step 13: Rearrange the wind power and photovoltaic operation scenario set so that the sorted wind power and photovoltaic operation scenario set.

[0061] Step 14: Use the fast forward reduction method based on probability distance to generate a typical source-load scenario set considering the spatio-temporal correlation of prediction errors based on the sorted wind power and photovoltaic operation scenario set.

[0062] Specifically, in step 1, a probability model of the source-load prediction error is first constructed, and time-series segmented fitting of the source-load prediction error and generation of operation scenarios are carried out based on historical data; then, combined with the correlation information of historical data, the wind power-photovoltaic operation scenario set considering the time-series correlation of the prediction error is rearranged by the eigenvalue decomposition sorting method to obtain the wind power-photovoltaic operation scenario set considering the spatio-temporal correlation of the prediction error; finally, all source-load operation scenario sets are reduced to generate typical scenarios to effectively describe the characteristics of source-load uncertainty.

[0063] The operation scenario generation and reduction method fully combines the probability distribution function with historical data, reflecting the process from data to distribution and then from distribution to samples. It can accurately describe the source-load uncertainty caused by prediction errors and transform the uncertainty optimization problem into a deterministic optimization problem under a few typical scenarios. Specifically, it includes:

[0064] S2.1: Time-series correlation of source-load prediction error and operation scenario generation.

[0065] 1. Time-series segmented fitting of source-load prediction error.

[0066] Model the output of wind power and photovoltaic and the load demand, and express the actual value as the sum of the predicted value and the prediction error, where the prediction error is a random variable.

[0067]

[0068] Among them, respectively represent the actual values of photovoltaic output, wind power output and load demand at node i; P i PV , P i W , P i L respectively represent the day-ahead predicted values of photovoltaic output, wind power output and load demand; respectively represent the prediction errors of photovoltaic output, wind power output and load demand.

[0069] This application uses the TSL distribution (t Location-scale distribution) with location-scale parameters to describe the prediction error of wind power-photovoltaic output, and uses the normal distribution with a mean of 0 (Normal distribution) to describe the prediction error of load demand.

[0070] The prediction error of wind power-photovoltaic output The probability density function (Probability density function, PDF) is as follows:

[0071]

[0072] Among them, v, respectively represent the shape parameter, scale parameter, and location parameter of the wind power - photovoltaic TLS distribution; Γ(·) represents the gamma function.

[0073] Load demand prediction error The PDF of can be expressed as:

[0074]

[0075] Among them, respectively represent the scale parameter and location parameter of the normal distribution.

[0076] Based on the above - mentioned model, considering the characteristic that the prediction error increases with the increase of the prediction time - length, a time - series segmented fitting of the source - load prediction error is carried out: Using the historical output data and prediction data of wind power - photovoltaic, the change curves of the wind power - photovoltaic output prediction error and the change curve of the prediction error standard deviation are statistically obtained at different time periods; the two change curves respectively correspond to the values of the location parameter and scale parameter of the source - load prediction error PDF at different moments Substituting them into Equation (13) - Equation (14), the source - load prediction error distribution considering time - series correlation can be obtained.

[0077] 2. Generation of source - load operation scenarios based on Latin hypercube sampling.

[0078] According to the above - mentioned prediction error distribution considering time - series correlation, this application uses the Latin hypercube sampling method (Latinhypercube sampling, LHS) to generate multiple source - load operation scenarios for each time period. Compared with the traditional Monte Carlo sampling method, the LHS method equally divides the interval of the cumulative distribution function (Cumulative distribution function, CDF) and samples within each interval, realizing a more uniform sample distribution with a smaller sampling scale, and the sampling results are more representative. The calculation formula for the k - th sampling value of the LHS method at time period t is as follows:

[0079] Δξ k,t = F -1 [(1 / N)U d +(k - 1) / N](16)

[0080] Among them, Δξ k,t represents the k - th sampling at time period t, k = 1, 2, ……, N; N represents dividing the vertical axis of the random variable CDF into N equally - spaced intervals (that is, using the LHS method to perform N samplings in total). Since the value range of the vertical axis of the CDF is 0 - 1, the width of each interval is (1 / N); Ud represents a random number uniformly distributed within the range of 0 - 1; F -1 (·) represents the inverse function of the CDF.

[0081] Substituting the sampling result of Equation (16) into Equation (13), the source - load operation scenario set S considering the temporal correlation of prediction errors can be generated:

[0082]

[0083] where:

[0084]

[0085] where nw, npv, and nl represent the number of nodes connected to wind power, the number of nodes connected to photovoltaic power plants, and the number of load nodes, respectively.

[0086] S2.2: Spatial correlation modeling of wind power - photovoltaic prediction errors.

[0087] When multiple wind power - photovoltaic power plants are connected to the same regional distribution network, the proximity of their geographical locations results in similar but not identical wind speeds and light intensities they experience, leading to spatial correlation in wind power - photovoltaic output. When generating the source - load operation scenario set, the spatial correlation of prediction errors also needs to be considered to improve the accuracy of the scenario set. The spatial correlation considered in this application is the correlation between wind power plants and between photovoltaic power plants, without considering the correlation between wind power plants and photovoltaic power plants or between different node load demands.

[0088] Calculate the prediction error correlation coefficient matrix of adjacent wind power - photovoltaic power plants based on historical data, and use the permutation method based on eigen - decomposition (PED) to rearrange the wind power - photovoltaic operation scenario set considering the temporal correlation of prediction errors generated in S2.1, so that the sorted scenario set contains both spatio - temporal correlation information. The specific modeling steps are as follows:

[0089] 1. Calculate the Pearson correlation coefficient ρ of the historical data of the output of wind power - photovoltaic power plants at a single time period m,n and the correlation coefficient matrix C 0 , where C 0 is the target correlation matrix for permutation.

[0090]

[0091]

[0092] where, Respectively represent the historical wind power output data matrix in the m-th column and the n-th column, matrix has a dimension of N×nw; σ m and σ n are respectively standard deviations.

[0093] 2. Calculate the Pearson correlation coefficient and the correlation coefficient matrix C W of the wind power operation scenario set S W with a single time period dimension of N×nw, and the calculation method is analogous to equations (19)-(20).

[0094] 3. Perform eigenvalue decomposition on the correlation coefficient matrix C 0 and C W :

[0095]

[0096] where Λ 0 is the diagonal matrix composed of the eigenvalues of the correlation coefficient matrix C 0 ; is the transpose of the orthogonal matrix Q 0 composed of the eigenvectors of the correlation coefficient matrix C 0 ; Λ W is the diagonal matrix composed of the eigenvalues of the correlation coefficient matrix C W ; is the transpose of the orthogonal matrix Q W composed of the eigenvectors of the correlation coefficient matrix C W ;

[0097] 4. Generate a new matrix S1 W through the matrix transformation of equation (22):

[0098]

[0099] Let the matrix C1 W be the correlation coefficient matrix of S1 W , and the matrix S1 W has the same correlation coefficient matrix as the matrix , that is, C1 W = C 0 .

[0100] 5. Sort each column of data in the single-time-period wind power operation scenario set S W according to the matrix S1 W . Let be the element in the i1-th column of the n-th row in S W , be the element in the i1-th column of the n1-th row in S1 W , find Element value Then The position in matrix S W is changed from the n-th row to the n1-th row. Among them, rank(·) represents the position of the element in the matrix.

[0101] 6. Repeating the above steps 1-5 for different time period wind power operation scenario sets can obtain the wind power and photovoltaic operation scenario sets S W* , S PV* . Replace the wind power and photovoltaic operation scenario sets S W* , S PV* in formula (5) with the scenario sets S W , S PV , and obtain the wind power-photovoltaic operation scenario set S * considering the spatio-temporal correlation of prediction errors.

[0102] It should be noted that when the correlation coefficient matrix in 3 of S2.2 is a negative definite matrix, it may cause the matrix S1 W to have complex elements. At this time, since the imaginary part in the complex elements is much smaller than the real part, the column elements can be sorted by the modulus values of these elements. Therefore, applying the PED method to generate the wind power-photovoltaic operation scenario set containing spatio-temporal correlation information can overcome the requirement of the traditional Cholesky decomposition sorting method for the positive definite symmetry of the correlation coefficient matrix, and the method is more universal.

[0103] S2.3: Source-load operation scenario reduction.

[0104] The number of source-load operation scenarios generated by S2.1 and S2.2 is still large, and a large number of operation scenarios will lead to a large amount of calculation for solving the risk scheduling model. Therefore, scenario reduction is required to describe the source-load uncertainty of the active distribution network using a small number of typical scenarios.

[0105] This application generates the source-load typical operation scenarios through the fast forward reduction method based on probability distance. The basic idea of this method is: by calculating the probability distance between different scenarios, the scenarios with smaller probability distance are merged or eliminated, so as to reduce the number of scenarios. The specific steps are as follows:

[0106] 1. For the source-load operation scenario set S * considering the spatio-temporal correlation of prediction errors, initialize the probability (1 / N) of each scenario to generate a probability matrix p, and the dimension of the matrix p is N×1; calculate the geometric distance between each scenario in the scenario set S * respectively, and obtain a geometric distance matrix d, and the dimension of the matrix d is N×N.

[0107] 2. Calculate the scenario set S *The sum of the probability distances between each scenario and the remaining scenarios in [[]] is used to obtain the probability distance matrix g (with the same dimension as matrix d), and the scenario d with the smallest probability distance from the remaining scenarios is selected.

[0108] 3. Let the scenario set be S * Replace scenario d with the scenario r in [[]] that has the smallest geometric distance from scenario d, add the probability of scenario d to the probability of scenario r, eliminate scenario d, and update the operating scenario set S * , the geometric distance matrix d, the probability matrix p, and the probability distance matrix g.

[0109] 4. Repeat the above steps 1 - 3 until the number of reduced scenarios meets the requirements, obtaining N * scenario sets, that is, obtaining the source - load typical scenario set S' considering the spatio - temporal correlation of prediction errors and the probability matrix p' of each scenario.

[0110] Step 2: Construct a two - stage risk dispatch model for active distribution networks based on the source - load typical scenario set.

[0111] According to the main characteristics of active distribution networks, the controllable resources therein can be divided into five categories: distributed generation (DG), network structure, grid - side flexible resources, controllable loads, and energy storage systems (ESS). Through the risk dispatch model for day - ahead and intra - day dispatch of various controllable resources, the comprehensive operating loss value of the distribution network can be effectively reduced, and low - probability and high - risk scenarios caused by source - load uncertainty can be reduced.

[0112] Such as Figure 3As shown in the figure, in the day-ahead scheduling stage of the two-stage risk scheduling model for active distribution networks, discrete decision variables are optimized considering the day-ahead predicted values of sources and loads (initial scenario, without considering the prediction errors of sources and loads), such as variables reflecting the opening and closing states of tie switches, variables reflecting the tap positions of on-line tap changers (OLTCs) and the switching numbers of capacitor banks (CBs), variables reflecting whether controllable loads participate in the scheduling plan (demand response), and variables reflecting the charge and discharge states of energy storage systems (ESSs). Such variables are called Here-and-Now (HN) variables; in the intra-day scheduling stage, continuous decision variables are optimized based on the optimized values of the discrete variables obtained in the day-ahead scheduling and the typical operation scenarios of sources and loads (considering the prediction errors of sources and loads, that is, taking into account the uncertainties of sources and loads), such as the output powers of distributed generations (DGs, including wind power and photovoltaic power) in each operation scenario, the powers and operation losses of soft open points (SOPs), the reactive power compensated by static var compensators (SVCs), the charge and discharge powers of ESSs, and the output powers of generators. Such variables are called Wait-and-See (WS) variables, and the continuous state variables in each operation scenario, such as node voltages, branch currents, and branch active and reactive powers, are obtained.

[0113] The time resolution of the two-stage risk scheduling model for active distribution networks is 1 hour. In the intra-day scheduling stage, the values of the HN variables determined in the day-ahead stage are taken as known for solution, that is, the intra-day scheduling stage is allowed to perform power regulation based on the day-ahead scheduling plan to cope with the uncertainties of sources and loads; at the same time, separately solving the optimization problems in the two stages can reduce the huge computational amount brought by numerous discrete decision variables in multi-scenarios.

[0114] Among them, the two-stage risk scheduling model for active distribution networks includes: a day-ahead scheduling stage model and an intra-day scheduling stage model; the day-ahead scheduling stage model includes: a day-ahead scheduling stage objective function and day-ahead scheduling stage constraints. The day-ahead scheduling stage objective function is a function with the minimum daily comprehensive operation loss value of the active distribution network as the goal. The day-ahead scheduling stage constraints include: day-ahead power output constraints, network dynamic reconfiguration constraints, day-ahead SOP constraints, OLTC constraints, CB constraints, day-ahead SVC constraints, day-ahead DR constraints, day-ahead ESS constraints, day-ahead distribution network power flow constraints, and day-ahead system security constraints; the intra-day scheduling stage model includes: an intra-day scheduling stage objective function and intra-day scheduling stage constraints. The intra-day scheduling stage objective function is a function with the minimum weighted sum of the daily comprehensive operation loss value considering multi-scenarios and the loss value of high-risk scenarios of the active distribution network as the goal. The intra-day scheduling stage constraints include: intra-day power output constraints, intra-day SOP constraints, intra-day SVC constraints, intra-day DR constraints, intra-day ESS constraints, intra-day distribution network power flow constraints, and intra-day system security constraints.

[0115] As an alternative implementation, the objective function in the day-ahead scheduling stage includes:

[0116] minf = ω L ·(f P + f ESS + f SOP + f DR ) + ω R ·f VP (23)

[0117]

[0118]

[0119]

[0120]

[0121]

[0122] Among them, f is the daily comprehensive operation loss value of the active distribution network; ω L is the weight coefficient of the cost item; f P is the main network power purchase cost; f ESS is the ESS operation cost; f SOP is the SOP loss cost; f DR is the DR cost; ω R is the weight coefficient of the power-voltage deviation term; f VP is the voltage-power deviation; T is the total number of time periods; c t is the electricity price at time period t; is the active power value of the power source node at time period t in the initial scenario; Δt is the time interval; ω ess is the set of ESS access nodes; c ess is the ESS cost coefficient; is the charging power of the ESS of node i at time period t in the initial scenario; is the discharging power of the ESS of node i at time period t in the initial scenario; ω sop is the set of nodes on both sides of the SOP access branch; c sop is the SOP cost coefficient at time period t; is the active power of the SOP converter loss of node i at time period t in the initial scenario; is the active power of the SOP converter loss of node j at time period t in the initial scenario; c dr is the DR cost coefficient at time period t; ω dr is the set of DR access nodes; is the interruptible load demand response power of node i at time t in the initial scenario; ω n is the set of distribution network nodes; V i,0,t is the voltage magnitude of node i at time t in the initial scenario; V ref is the node voltage reference value; ω l is the set of branches; I ij,0,t is the current of branch ij at time t in the initial scenario; r ij is the resistance value of branch ij.

[0123] As an alternative implementation, the day-ahead power output constraints of power sources include:

[0124]

[0125]

[0126]

[0127]

[0128] Among them, is the day-ahead power output of the photovoltaic power at node i at time t in the initial scenario; is the day-ahead prediction value of the power output of the photovoltaic power at node i at time t; ω pv is the set of photovoltaic nodes; is the day-ahead power output of the wind power at node i at time t in the initial scenario; is the day-ahead prediction value of the power output of the wind power at node i at time t; ω wind is the set of wind power nodes; is the lower limit of the active power of the power source node; is the upper limit of the active power of the power source node; is the lower limit of the reactive power of the power source node; is the reactive power value of the power source node at time t in the initial scenario; is the upper limit of the reactive power of the power source node.

[0129] Network dynamic reconfiguration refers to reconstructing the network topology at different times by controlling the switch states to achieve the purpose of reasonably regulating the power flow, which can improve the voltage violation problem and power violation problem caused by the source-load uncertainty in the active distribution network. The time resolution of the two-stage risk scheduling model is 1 hour, and the network dynamic reconfiguration proposed in this application is also an hourly adjustment strategy for the distribution network, which is compatible with the risk scheduling. The network dynamic reconfiguration constraints include: radial network topology constraints and switch operation constraints.

[0130] The radial network topology constraint means that the distribution network needs to remain radial during operation and network dynamic reconfiguration, and loops and islands are not allowed. The radial network topology constraint includes:

[0131]

[0132]

[0133]

[0134] Among them, χ ij,0,t and χ ji,0,t are 0-1 variables describing the parent-child relationship between node i and node j on both sides of branch ij in the initial scenario at time t. If node i is the parent node of node j, χ ij,0,t = 1. If node j is the parent node of node i, χ ji,0,t = 1; α ij,0,t is a 0-1 variable describing whether branch ij is open or closed at time t. If branch ij is closed, α ij,0,t = 1; δ(j) represents the set of parent nodes of node j; n 0 is the power supply node.

[0135] The switches in the distribution network can flexibly adjust the topology of the distribution network by opening and closing contacts, but frequent switch operations will seriously reduce the service life of the switches. In addition, the frequent change of the distribution network topology will reduce the operation reliability of the distribution network.

[0136] Therefore, it is necessary to limit the maximum number of operations of a single switch, the minimum operation interval of a single switch, and the total number of operations of all switches per day in the day-ahead optimization.

[0137] The switch operation constraint includes:

[0138]

[0139]

[0140]

[0141] Among them, α ij,0,t+1 is a 0-1 variable describing whether branch ij is open or closed at time t+1; n switch is the maximum number of operations of a single switch per day; m is the shortest number of hours for the change of the open / close state of a single switch; α ij,0,t+a is a 0-1 variable describing whether branch ij is open or closed at time t+a; N switch is the maximum number of operations of all switches per day.

[0142] The SOP is a back-to-back voltage-sourced converter (B2B-VSC). As a fully controlled power electronic device, the SOP has the advantages of continuously regulating power and providing reactive voltage support. During the risk dispatching process of the active distribution network, the PQ-VdcQ control is selected as the SOP control mode. Usually, the day-ahead SOP constraints need to consider the day-ahead SOP active power balance constraint, the day-ahead SOP reactive power output and access capacity constraint. The day-ahead SOP constraints include: the day-ahead SOP active power balance constraint and the day-ahead SOP reactive power output and access capacity constraint.

[0143] The day-ahead SOP active power balance constraint includes:

[0144]

[0145]

[0146] Among them, is the active power injected by the SOP into node i at time t under the initial scenario; is the active power injected by the SOP into node j at time t under the initial scenario; η sop is the active power loss coefficient of the SOP's single-side VSC; is the reactive power injected by the SOP into node i at time t under the initial scenario; is the reactive power injected by the SOP into node j at time t under the initial scenario.

[0147] The day-ahead SOP reactive power output and access capacity constraint includes:

[0148]

[0149]

[0150] Among them, is the lower limit of the reactive power injected by the SOP into the node; is the upper limit of the reactive power injected by the SOP into the node; is the capacity of the VSC on the i side of the node; is the capacity of the VSC on the j side of the node.

[0151] The OLTC constraints include the upper and lower limit constraints of the tap changer position and the tap changer position change times constraint. The OLTC constraints include:

[0152]

[0153]

[0154]

[0155]

[0156] Among them, R ij,t is the voltage regulation rate between node i and node j during period t; V j,0,t is the voltage amplitude of node j during period t under the initial scenario; ω oltc represents the set of nodes on both sides of the branch where the OLTC is connected; R0 ij is the initial regulation rate between node i and node j; O ij,t is the tap position of the OLTC between node i and node j during period t; ΔR is the adjustable change amount between two adjacent tap positions of the OLTC; O is the lower limit value of the tap position of the OLTC; is the upper limit value of the tap position of the OLTC; O ij,t-1 is the tap position of the OLTC between node i and node j during period t - 1; N oltc is the maximum number of adjustments of the OLTC within a day.

[0157] The CB can perform discrete reactive power compensation. The CB constraints include:

[0158]

[0159]

[0160]

[0161] Among them, is the reactive power injected by the CB at node i to the CB during period t under the initial scenario; is the number of CBs put into operation at node i during period t under the initial scenario; is the unit capacity of the CB at node i; ω cb is the set of nodes where the CB is connected; is the number of CBs put into operation at node i during period t - 1 under the initial scenario; N cb is the maximum number of changes of the CB within a day; is the maximum number of CBs that can be put into operation at node i.

[0162] The SVC can perform continuous reactive power compensation. The day-ahead SVC constraints include:

[0163]

[0164] Among them, is the lower limit of the reactive power injected by the SVC into the node; is the reactive power injected by the SVC at node i into the SVC at the initial scenario during period t; is the upper limit of the reactive power injected by the SVC into the node; ω svc is the set of nodes where the SVC is connected.

[0165] Considering the participation of interruptible loads in demand response, its constraint conditions mainly include response power constraint, interruption times constraint, maximum allowable interruption duration constraint, and day-ahead DR constraint, including:

[0166]

[0167]

[0168]

[0169] Among them, is the state variable of the interruptible load at node i during period t in the initial scenario, being equal to 0 indicates no load interruption; is the maximum interruptible amount at node i during period t; is the maximum number of interruptions of the interruptible load at node i in a day; is the maximum interruption duration of the interruptible load.

[0170] Day-ahead ESS constraints, including:

[0171]

[0172]

[0173]

[0174]

[0175]

[0176] Among them, is the minimum charging power of the ESS; is a 0-1 variable describing the charging state of the ESS at node i during period t in the initial scenario; is the maximum charging power of the ESS; is the minimum discharging power of the ESS; is a 0-1 variable describing the discharging state of the ESS at node i during period t in the initial scenario; is the maximum discharging power of the ESS; is the capacity value of the ESS at node i during period t in the initial scenario; is the capacity value of the ESS at node i during period t-1 in the initial scenario; η ess,chThe charging efficiency of the ESS; η ess,dch The discharging efficiency of the ESS; The minimum value of the ESS capacity; The maximum value of the ESS capacity; The capacity value of the ESS at node i in the initial time period under the initial scenario; The capacity value of the ESS at node i in the final time period under the initial scenario.

[0177] The day-ahead distribution network power flow constraints, including:

[0178]

[0179]

[0180]

[0181]

[0182] Among them, P jh,0,t The day-ahead active power of branch jh at time t under the initial scenario; The set of child nodes of node j; P ij,0,t The day-ahead active power of branch ij at time t under the initial scenario; I ij,0,t The current of branch ij at time t under the initial scenario; The injected active power of node j under the initial scenario; Q ij,0,t The day-ahead reactive power of branch ij at time t under the initial scenario; x ij The reactance of branch ij; The injected reactive power of node j under the initial scenario; The day-ahead output of the PV at node j at time t under the initial scenario; The day-ahead output of the wind power at node j at time t under the initial scenario; The charging power of the ESS at node j at time t under the initial scenario; The discharging power of the ESS at node j at time t under the initial scenario; The active load at node j at time t under the initial scenario; The reactive power injected by the CB at node j at time t under the initial scenario; The reactive power injected by the SVC at node j at time t under the initial scenario; The reactive load at node j at time t under the initial scenario.

[0183] The day-ahead system security constraints, including:

[0184]

[0185]

[0186]

[0187]

[0188] Among them, I N is the rated current; α i,0,t is a 0-1 variable characterizing whether the voltage of node i at time t in the initial scenario is lower than the lower limit. When the node voltage is lower than the lower limit, α i,0,t = 1; γ i,0,t is a 0-1 variable characterizing whether the voltage of node i at time t in the initial scenario is higher than the upper limit. When the node voltage is higher than the upper limit, γ i,0,t = 1; V lim is the lower limit of the ultimate tolerance of the node voltage; is the upper limit of the ultimate tolerance of the node voltage; V N is the rated voltage.

[0189] Equation (65) represents the branch power constraint, and equations (66)-(68) represent the node voltage constraint. From equations (66)-(68), it can be seen that when α i,0,t and γ i,0,t take different values, the node voltage is constrained in different intervals. Therefore, there are three cases for the voltage security constraint: (1) The voltage exceeds the upper limit, that is, α i,0,t = 0 and γ i,0,t = 1, and the node voltage constraint is (2) The voltage exceeds the lower limit, that is, α i,0,t = 1 and γ i,0,t = 0, and the node voltage constraint is V lim 2 ,(0.94V N ) 2 ; (3) The voltage does not exceed the limit, that is, α i,0,t = 0 and γ i,0,t = 0, and the node voltage constraint is [(0.94V N ) 2 ,(1.06V N ) 2 .

[0190] As an optional implementation manner, based on the objective function in the day-ahead scheduling stage, by probabilistically weighting the daily comprehensive operation loss values of multiple source-load typical scenarios, the daily comprehensive operation loss value f m considering multiple scenarios is obtained; at the same time, based on the quantification of low-probability and high-risk (voltage and power over-limit risks) operation scenarios by CVaR, the high-risk scenario loss value f r is obtained. The fm With f r weighted as the objective function for the intraday scheduling stage. The objective function for the intraday scheduling stage includes:

[0191] min{ω p ·f m +ω e ·f r}(69)

[0192]

[0193]

[0194]

[0195]

[0196]

[0197]

[0198]

[0199] where ω p is the weight coefficient of f m ; f m is the daily comprehensive operation loss value of the active distribution network considering multiple scenarios; ω e is the weight coefficient of f r , ω p +ω e = 1; f r is the loss value of the high-risk scenario; N * is the total number of scenarios; p s is the probability of scenario s; f P,s is the main grid power purchase cost of scenario s; f SOP,s is the SOP loss cost of scenario s; f DR,s is the DR cost of scenario s; f VP,s is the voltage power deviation of scenario s; C' β [·] is the approximation of CVaR; η β is the VaR value of the loss function L when the confidence level is set to β; β is the confidence level; E[·] is the mathematical expectation; is the active power value of the power source node at time t under scenario s; is the charging power of the ESS at node i at time t under scenario s; is the discharging power of the ESS at node i at time t under scenario s; is the active power of the SOP converter loss at node i at time t under the s-th initial scenario; Active power of the SOP converter loss of node j in time period t under scenario s; Interruptible load demand response power of node i in time period t under scenario s; V i,s,t Voltage amplitude of node i in time period t under scenario s; I ij,s,t Current of branch ij in time period t under scenario s.

[0200] As an optional implementation, the in-day power output constraints of power sources include:

[0201]

[0202]

[0203]

[0204]

[0205] Among them, Day-ahead power output of the photovoltaic at node i in time period t under scenario s; Day-ahead prediction value of the power output of the photovoltaic at node i in time period t under scenario s; Day-ahead power output of the wind power at node i in time period t under scenario s; Day-ahead prediction value of the power output of the wind power at node i in time period t under scenario s; for the s-th; for the s-th; Reactive power value of the power source node in time period t under scenario s.

[0206] In-day SOP constraints include: include:

[0207]

[0208]

[0209] Among them, Reactive power injected by the SOP into node i in time period t under scenario s; Reactive power injected by the SOP into node j in time period t under scenario s; Active power injected by the SOP into node i in time period t under scenario s; Active power injected by the SOP into node j in time period t under scenario s.

[0210] In-day SVC constraints include:

[0211]

[0212] Among them, Reactive power injected by the SVC at node i by the SVC in time period t under scenario s.

[0213] Intraday DR constraints include:

[0214]

[0215] Intraday ESS constraints include:

[0216]

[0217]

[0218]

[0219]

[0220] Among them, is the capacity value of the ESS at node i in scenario s at time t; is the capacity value of the ESS at node i in scenario s at time t - 1; is the capacity value of the ESS at node i in scenario s at the initial time period; is the capacity value of the ESS at node i in scenario s at the final time period.

[0221] Intraday distribution network power flow constraints include:

[0222]

[0223]

[0224]

[0225]

[0226]

[0227]

[0228] Among them, P jh,s,t is the day-ahead active power of branch jh in scenario s at time t; P ij,s,t is the day-ahead active power of branch ij in scenario s at time t; I ij,s,t is the current of branch ij in scenario s at time t; is the injected active power of node j in scenario s; Q jh,s,t is the day-ahead reactive power of branch ij in scenario s at time t; is the injected reactive power of node j in scenario s; V j,s,t is the voltage magnitude of node j in scenario s at time t; is the day-ahead output of the PV at node j in scenario s at time t; The day-ahead output of wind power at node j under scenario s in period t; The charging power of the ESS at node j in period t under scenario s; The discharging power of the ESS at node j in period t under scenario s; The active load at node j in period t under scenario s; The reactive power injected by the CB at node j by the CB in period t under scenario s; The reactive power injected by the SVC at node j by the SVC in period t under scenario s; The reactive load at node j in period t under the s-th scenario.

[0229] Intra-day system security constraints, including:

[0230]

[0231]

[0232]

[0233]

[0234] Among them, α i,s,t is a 0-1 variable representing whether the voltage at node i in period t under scenario s is lower than the lower limit; γ i,s,t is a 0-1 variable representing whether the voltage at node i in period t under scenario s is higher than the upper limit.

[0235] Step 3: By means of second-order cone relaxation, transform the two-stage risk dispatch model of the active distribution network from a mixed-integer non-linear programming problem into a mixed-integer second-order cone programming problem, and solve the two-stage risk dispatch model of the active distribution network to obtain the dispatch plan of the active distribution network.

[0236] Specifically, from the non-linear terms in the above SOP constraints and distribution network power flow constraints, it can be seen that the established two-stage risk dispatch model is essentially a complex mixed-integer non-linear programming problem, which is difficult to optimize and solve. Therefore, in this application, the mixed-integer non-linear programming problem is transformed into a mixed-integer second-order cone programming problem (Mixed-integer second-order cone programming, MISOCP) by means of second-order cone relaxation. Second-order cone programming has wide-range convergence and is easy to obtain the global optimal solution. Taking the day-ahead SOP constraints and distribution network power flow constraints in the day-ahead dispatch stage as an example, the transformation process of MISOCP is illustrated, and the same applies to the intra-day stage.

[0237] 1. Perform second-order cone relaxation on equation (40) in the day-ahead SOP constraints.

[0238]

[0239] 2. Perform second-order cone relaxation on equations (59) - (62) in the current distribution network power flow constraints. Define v i,0,t and l ij,0,t :

[0240]

[0241] Substituting them into equations (59) - (62) gives:

[0242]

[0243]

[0244]

[0245]

[0246] Further perform second-order cone relaxation on equation (105) to obtain the following second-order cone constraints:

[0247]

[0248] For the MISOCP problem, a commercial solver GUROBI can be used for solution.

[0249] Finally, the risk reduction effect of the dispatching scheme can also be evaluated through the over-limit risk index constructed in S1.2.

[0250] Technical effects of this application:

[0251] (1) A source-load operation scenario generation method considering the spatio-temporal correlation of prediction errors is proposed. When modeling the temporal and spatial correlations of prediction errors, this method fully combines the probability distribution function with historical data, more accurately describing the source-load uncertainty caused by prediction errors; in addition, when modeling the spatial correlation of wind and light prediction errors, the used eigen-decomposition sorting method overcomes the requirement of the mainstream Cholesky decomposition sorting method for the positive definite symmetry of the correlation coefficient matrix, and is more universal.

[0252] (2) When constructing a two-stage risk dispatching model for an active distribution network, the risk of the power grid dispatching scheme is reduced through the joint optimization of network-side intelligent soft switches and network dynamic reconfiguration, fully utilizing the advantages of the continuous power regulation of intelligent soft switches, providing reactive power and voltage support, and the flexible adjustment of the network topology by network dynamic reconfiguration. The joint optimization of the two can effectively reduce the high operation loss value of the distribution network caused by source-load uncertainty and the risk of voltage and power over-limit.

[0253] In an exemplary embodiment, a computer device is provided, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to implement a two-stage risk scheduling method for an active distribution network.

[0254] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, it implements a two-stage risk scheduling method for an active distribution network.

[0255] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, it implements a two-stage risk scheduling method for an active distribution network.

[0256] In an exemplary embodiment, a computer device is provided. The computer device can be a server or a terminal, and its internal structural diagram can be as Figure 4 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a two-stage risk scheduling method for an active distribution network.

[0257] Those skilled in the art can understand that Figure 4 the structure shown in

[0258] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0259] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0260] The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0261] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0262] In the present application, specific examples are used to elaborate on the principles and implementation manners of the present application. The descriptions of the above embodiments are only used to help understand the methods and core ideas of the present application; at the same time, for those of ordinary skill in the art, according to the ideas of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. A two-stage risk dispatching method for active distribution network, characterized in that: The two-stage risk dispatching method for active distribution network includes: Construct a set of typical source-load scenarios that consider the spatiotemporal correlation of prediction errors; Based on the source-load typical scenario set, a two-stage risk dispatch model of an active distribution network is constructed; the two-stage risk dispatch model of an active distribution network includes: a day-ahead dispatch stage model and a day-ahead dispatch stage model; the day-ahead dispatch stage model includes: a day-ahead dispatch stage objective function and a day-ahead dispatch stage constraint condition, the day-ahead dispatch stage objective function is a function with the minimum daily comprehensive operation loss value of the active distribution network as the target, the day-ahead dispatch stage constraint condition includes: day-ahead power output constraint, network dynamic reconstruction constraint, day-ahead SOP constraint, OLTC constraint, CB constraint, day-ahead SVC constraint The intraday dispatching stage model includes: an intraday dispatching stage objective function and intraday dispatching stage constraint conditions, the intraday dispatching stage objective function is a function that takes the weighted sum of the daily comprehensive operation loss value of the active distribution network taking into account multiple scenarios and the high-risk scenario loss value as the minimum, and the intraday dispatching stage constraint conditions include: intraday power output constraint, intraday SOP constraint, intraday SVC constraint, intraday DR constraint, intraday ESS constraint, intraday distribution network flow constraint and intraday system safety constraint; Through the second-order cone relaxation method, the two-stage risk scheduling model of the active distribution network is transformed from a mixed integer nonlinear programming problem to a mixed integer second-order cone programming problem, and the two-stage risk scheduling model of the active distribution network is solved to obtain a scheduling plan for the active distribution network.

2. The two-stage risk dispatching method for active distribution network according to claim 1 is characterized in that: Construct a typical source-load scenario set that takes into account the spatiotemporal correlation of prediction errors, including: By using the historical output data and forecast data of wind power and photovoltaic power, the wind power and photovoltaic power output forecast error change curve and the wind power and photovoltaic power output forecast error standard deviation change curve in different time periods are statistically obtained, and based on the wind power and photovoltaic power output forecast error change curve and the wind power and photovoltaic power output forecast error standard deviation change curve in each time period, the source-load forecast error distribution considering the time series correlation is determined; Based on the source-load prediction error distribution, a Latin hypercube sampling method is used to generate multiple source-load operation scenarios in each time period to obtain a wind power-photovoltaic operation scenario set; Rearranging the wind power-photovoltaic operation scenario set to make a sorted wind power-photovoltaic operation scenario set; The fast forward reduction method based on probability distance is used to generate the source-load typical scenario set considering the spatiotemporal correlation of prediction errors based on the sorted wind power-photovoltaic operation scenario set.

3. The two-stage risk dispatching method for active distribution network according to claim 1 is characterized in that: The objective function of the day-ahead scheduling phase includes: minf=ω L ·(f P +f ESS +f SOP +f DR )+ω R ·f VP ; Where, f is the daily comprehensive operation loss value of the active distribution network; ω L is the weight coefficient of the cost item; f P The cost of purchasing electricity for the main grid; f ESS is the ESS operating cost; f SOP is the SOP loss cost; f DR is the DR cost; R is the power voltage deviation term weight coefficient; f VP is the voltage power deviation; T is the total number of time periods; c t is the electricity price in period t; is the active power value of the power node in period t under the initial scenario; Δt is the time interval; ω ess is the set of ESS access nodes; c ess is the ESS cost coefficient; is the charging power of the ESS of node i in period t under the initial scenario; is the discharge power of the ESS at node i in period t under the initial scenario; ω sop is the node set on both sides of the SOP access branch; c sop is the SOP cost coefficient for period t; is the active power lost by the SOP converter of node i in period t under the initial scenario; is the active power loss of the SOP converter of node j in period t under the initial scenario; c dr is the DR cost coefficient in period t; ω dr It is a set of DR access nodes; is the interruptible load demand response power of node i in period t under the initial scenario; ω n is the set of distribution network nodes; V i,0,t is the voltage amplitude of node i in period t under the initial scenario; V ref is the node voltage reference value; ω l I is the branch set; ij,0,t is the current of branch ij in period t under the initial scenario; r ij is the resistance value of branch ij.

4. The two-stage risk dispatching method for active distribution network according to claim 3 is characterized in that: The day-ahead power output constraints include: in, is the day-ahead output of the PV at node i in period t under the initial scenario; is the day-ahead forecast value of the photovoltaic power output at node i in period t; ω pv is a collection of photovoltaic nodes; is the day-ahead output of wind power at node i in period t under the initial scenario; is the day-ahead forecast value of wind power output at node i in period t; ω wind is a set of wind power nodes; is the lower limit of the active power of the power node; The upper limit of the active power of the power node; is the lower limit of reactive power of power node; is the reactive power value of the power node in period t under the initial scenario; is the upper limit of reactive power of power node; The network dynamic reconfiguration constraints include: radial network topology constraints and switch operation constraints; The radial network topology constraints include: Among them, χ ij,0,t and χ ji,0,t is a 0-1 variable describing the parent-child relationship between nodes i and j on both sides of branch ij in the initial scenario in time period t. If node i is the parent node of node j, χ ij,0,t =1, if node j is the parent node of node i, ji,0,t =1;α ij,0,t is a 0-1 variable describing whether branch ij is open or closed in period t in the initial scenario. If branch ij is closed, α ij,0,t =1; δ(j) represents the parent node set of node j; n0 is the power node; The switch operation constraints include: Among them, α ij,0,t+1 is a 0-1 variable describing whether the branch ij is open or closed at the time period t+1 in the initial scenario; n switch is the maximum number of operations per day for a single switch; m is the shortest number of hours for a single switch to change its on / off state; α ij,0,t+a N is a 0-1 variable describing whether the branch ij is open or closed in the t+a period under the initial scenario; switch The maximum number of operations per day for all switches; The day-ahead SOP constraint includes: the day-ahead SOP active power balance constraint and the day-ahead SOP reactive power output and access capacity constraint; The active power balance constraints of the day-ahead SOP include: in, is the active power injected into node i by SOP in period t in the initial scenario; is the active power injected by SOP into node j in period t in the initial scenario; η sop is the active power loss coefficient of the VSC on one side of SOP; is the reactive power injected by SOP to node i in period t under the initial scenario; is the reactive power injected by SOP to node j in period t under the initial scenario; The day-ahead SOP reactive power output and access capacity constraints include: in, The lower limit of reactive power injected into the node by SOP; The upper limit of reactive power injected into the node by SOP; is the capacity of the VSC on the node i side; is the capacity of the VSC at the node j side; The OLTC constraints include: Among them, R ij,t is the voltage regulation rate between node i and node j during period t; V j,0,t is the voltage amplitude of node j in period t under the initial scenario; ω oltc Indicates the node set on both sides of the OLTC access branch; R0 ij is the initial adjustment rate between node i and node j; ij,t is the gear position of the OLTC tap changer between node i and node j in time period t; ΔR is the adjustable change between two adjacent gear positions of the OLTC tap changer; O is the lower limit value of the gear position of the OLTC tap changer; is the upper limit of the OLTC tap changer position; ij,t-1 N is the position of the OLTC tap changer between node i and node j in time period t-1; oltc The maximum number of adjustments of the OLTC tap changer in one day. The CB constraints include: in, is the reactive power injected by the CB to the CB at node i in period t in the initial scenario; is the amount of CB investment of node i in period t under the initial scenario; is the unit capacity of the CB at node i; ω cb is the set of CB access nodes; N is the amount of CB invested by node i in period t-1 under the initial scenario; cb is the maximum number of changes in CB in one day; is the maximum number of CBs that node i is allowed to invest; The day-ahead SVC constraints include: in, The lower limit of reactive power injected by SVC into the node; is the reactive power injected by the SVC to the SVC at node i in period t in the initial scenario; The upper limit of reactive power injected by SVC into the node; ω svc is a set of SVC access nodes; The day-ahead DR constraints include: in, is the state variable of the interruptible load of node i in period t under the initial scenario, Equal to 0 means no interruption of load; is the maximum interruption amount of node i in period t; is the maximum number of interruptions that can interrupt the load of node i in one day; is the maximum interruption duration of the interruptible load; The day-ahead ESS constraints include: in, is the minimum charging power of ESS; is a 0-1 variable describing the ESS charging state of node i in period t under the initial scenario; is the maximum charging power of ESS; is the minimum discharge power of ESS; is a 0-1 variable describing the ESS discharge state of node i in period t under the initial scenario; is the maximum discharge power of ESS; is the capacity value of ESS at node i in period t under the initial scenario; is the capacity value of the ESS at node i in the t-1 period under the initial scenario; η ess,ch is the charging efficiency of ESS; η ess,dch is the discharge efficiency of ESS; is the minimum value of ESS capacity; is the maximum value of ESS capacity; is the capacity value of ESS at node i in the initial period under the initial scenario; is the capacity value of ESS at node i in the final period under the initial scenario; The day-ahead distribution network flow constraints include: Among them, P jh,0,t is the day-ahead active power of branch jh in period t under the initial scenario; is the set of child nodes of node j; P ij,0,t is the day-ahead active power of branch ij in period t under the initial scenario; I ij,0,t is the current of branch ij in period t under the initial scenario; is the injected active power of node j in the initial scenario; Q ij,0,t is the reactive power of branch ij in period t in the initial scenario; x ij is the reactance of branch ij; is the injected reactive power of node j in the initial scenario; is the day-ahead output of the PV at node j in period t under the initial scenario; is the day-ahead output of wind power at node j in period t under the initial scenario; is the charging power of the ESS of node j in period t under the initial scenario; is the discharge power of the ESS at node j in period t under the initial scenario; is the active load of node j in period t under the initial scenario; is the reactive power injected by the CB to the CB at node j in period t in the initial scenario; is the reactive power injected by the SVC to the SVC at node j in period t in the initial scenario; is the reactive load of node j in period t under the initial scenario; The day-ahead system security constraints include: Among them, I N is the rated current; α i,0,t is a 0-1 variable that represents whether the voltage of node i in period t exceeds the lower limit in the initial scenario. When the node voltage exceeds the lower limit, α i,0,t =1;γ i,0,t is a 0-1 variable that represents whether the voltage of node i exceeds the upper limit in period t in the initial scenario. When the node voltage exceeds the upper limit, γ i,0,t =1; V lim is the lower limit of the node voltage tolerance; V is the upper limit of the node voltage tolerance. N is the rated voltage.

5. The two-stage risk dispatching method for active distribution network according to claim 4 is characterized in that: The objective function of the intraday scheduling stage includes: min{ω p ·f m +oh e ·f r }; Among them, ω p f m The weight coefficient of m The daily comprehensive operating loss value of the active distribution network taking into account multiple scenarios; ω e f r The weight coefficient, ω p +ω e =1; f r N is the loss value of high-risk scenario; * is the total number of scenes; p s is the probability of scene s; f P,s is the main grid electricity purchase cost for scenario s; f SOP,s is the SOP loss cost of scene s; f DR,s is the DR cost of scene s; f VP,s is the voltage power deviation of scenario s; C' β [·] is the approximate value of CVaR; η β is the VaR value of the loss function L when the confidence level is set to β; β is the confidence level; E[·] is the mathematical expectation; is the active power value of the power node in period t under scenario s; is the charging power of the ESS at node i in period t under scenario s; is the discharge power of the ESS at node i in period t under scenario s; is the active power lost by the SOP converter of node i in period t under the s-th starting scenario; is the active power lost by the SOP converter of node j in period t under scenario s; is the interruptible load demand response power of node i in period t under scenario s; V i,s,t is the voltage amplitude of node i in time period t under scenario s; I ij,s,t is the current of branch ij in period t under scenario s.

6. The two-stage risk dispatching method for active distribution network according to claim 5 is characterized in that: The intraday power output constraints include: in, is the day-ahead output of the PV at node i in period t under scenario s; is the day-ahead forecast value of the PV output at node i in period t under scenario s; is the day-ahead output of wind power at node i in period t under scenario s; is the day-ahead forecast value of wind power output at node i in period t under scenario s; is the sth; is the sth; is the reactive power value of the power node in period t under scenario s; The intraday SOP constraints include: including: in, is the reactive power injected by SOP to node i in time period t under scenario s; is the reactive power injected by SOP to node j in time period t under scenario s; is the active power injected into node i by SOP in period t under scenario s; is the active power injected into node j by SOP in period t under scenario s; The intraday SVC constraints include: in, is the reactive power injected by the SVC to the SVC at node i in time period t under scenario s; The intraday DR constraints include: The intraday ESS constraints include: in, is the capacity value of ESS at node i in time period t under scenario s; is the capacity value of ESS at node i in time period t-1 under scenario s; is the capacity value of ESS at node i in the initial period under scenario s; is the capacity value of ESS at node i in the last period under scenario s; The intraday distribution network flow constraints include: Among them, P jh,s,t is the day-ahead active power of branch jh in period t under scenario s; P ij,s,t is the day-ahead active power of branch ij in period t under scenario s; ij,s,t is the current of branch ij in period t under scenario s; is the injected active power of node j in scenario s; Q jh,s,t is the reactive power of branch ij in period t in scenario s; is the injected reactive power of node j in scenario s; V j,s,t is the voltage amplitude of node j in time period t under scenario s; is the day-ahead output of the PV at node j in period t under scenario s; is the day-ahead output of wind power at node j in scenario s during period t; is the charging power of the ESS at node j in period t under scenario s; is the discharge power of the ESS at node j in period t under scenario s; is the active load of node j in period t under scenario s; is the reactive power injected by the CB to the CB at node j in period t under scenario s; is the reactive power injected by the SVC to the SVC at node j in time period t under scenario s; is the reactive load of node j in the sth scenario during period t; The intraday system safety constraints include: Among them, α i,s,t is a 0-1 variable that represents whether the voltage of node i in time period t exceeds the lower limit in scenario s; γ i,s,t It is a 0-1 variable that represents whether the voltage of node i in time period t exceeds the upper limit in scenario s.

7. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the two-stage risk scheduling method for an active distribution network as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the two-stage risk scheduling method for an active distribution network described in any one of claims 1 to 6 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the two-stage risk scheduling method for an active distribution network described in any one of claims 1 to 6 is implemented.

Images