A method and system for generating an intraday scheduling scheme considering source-load cross correlation

Abstract

The application belongs to the technical field of power system operation and dispatching, and provides an intra-day dispatching scheme generation method and system considering source-load cross correlation, and the technical scheme is as follows: based on historical wind speed data and load data, considering source-load cross correlation, random wind speed sequence samples are obtained, the random wind speed sequence samples are converted into wind energy sequences in combination with the autocorrelation of the wind speed time sequence, and a wind energy output scene set is obtained; based on the wind energy output scene set, a rolling optimization framework is established, based on the wind power, load typical scene of the current period and the conventional unit start-stop scheme specified in the day-ahead, the dispatching plan of the future set time is obtained in combination with the intra-day dispatching optimization model optimization, then the prediction interval is rolled to the next time period, the intra-day dispatching optimization model optimization is performed again until the optimal dispatching scheme of each period of the total time interval is generated; under the condition of meeting the system reliability requirement, the method is helpful for the dispatching personnel to make more optimal unit dispatching decision, reduces the system wind curtailment rate, and ensures the supply-demand balance of the high penetration rate wind power system.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of power system operation scheduling, and particularly relates to an intra-day scheduling scheme generation method and system considering source-load cross correlation. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] Currently, the unit scheduling scheme from the time scale includes day-ahead scheduling optimization, intra-day scheduling optimization and real-time coordination;

[0004] The 24-hour plan of the day-ahead scheduling optimization model is phased, and the time scale is relatively large, so the optimization result of a shorter time scale is needed. For the real-time coordination model, the constraint conditions for establishing such a model are more, and the complexity is improved, so it is necessary to promote the research on the intra-day rolling optimization model in order to enable decision makers to make reasonable scheduling plans. The concept of rolling optimization refers to taking the scheduling plan of the day-ahead scheduling optimization model as a reference, taking the optimal value of the current execution period as the optimization target, and solving the next execution period in the next stage rolling time window to obtain the optimal value of the next execution period, and then rolling and solving. The intra-day rolling optimization model is a further refinement of the day-ahead plan, and the scheduling plan becomes a plan with finer granularity.

[0005] The current intra-day rolling optimization scheduling has not yet solved multiple key core problems, including the following: first, the wind power output has significant uncertainty and randomness, and the decision-making of the intra-day operation of the power system needs to consider "reliability" and "economy", which leads to great difficulties in analysis and modeling; second, the existing scheduling scheme does not consider the source-load cross correlation, and the prediction lead time has a greater impact on the prediction error, which will inevitably affect the reliability and intra-day rolling scheduling decision of the system to some extent. How to reveal the relationship between wind and light prediction deviation and system reliability, and realize the intra-day scheduling optimization of the system under the reliability constraint, etc. still needs further exploration; finally, the time scale of the intra-day rolling optimization model is 15 minutes, and the model needs to be solved within 15 minutes. In order to make the data results obtained by the model have timeliness, it is necessary to find a high-efficiency solving method.

[0006] The inventor found that the application No. CN117559523A, the invention name is a kind of intra-day rolling scheduling method and system considering reliability and uncertainty, although it considers reliability and uncertainty, but there is limitation in the scene selection of research object, and the model solving speed is slow, cannot provide scheduling reference for dispatchers in time, therefore does not meet the needs of power dispatching, there are obvious disadvantages. SUMMARY

[0007] In order to solve at least one technical problem in the above background art, the present application provides a method and system for generating an intra-day scheduling scheme considering the cross correlation of source and load, which proposes a wind power output scene reconstruction method considering the autocorrelation of wind speed and the cross correlation of source and load, considers typical scenes of wind power and load, establishes an intra-day rolling scheduling model containing reliability constraints, and finally solves the ultra-short-term prediction scheduling problem containing economy and reliability based on the intra-day rolling scheduling model of Benders decomposition, so as to better guarantee the reliability of the power system with high proportion of wind power access.

[0008] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0009] The first aspect of the present application provides a method for generating an intra-day scheduling scheme considering the cross correlation of source and load, comprising the following steps:

[0010] Based on historical wind speed data and load data, considering the cross correlation of source and load, random wind speed sequence samples are obtained, and the random wind speed sequence samples are converted into wind energy sequences in combination with the autocorrelation of wind speed time series, so as to obtain a wind energy output scene set;

[0011] Based on the wind energy output scene set, a rolling optimization framework is established, based on the wind power and load typical scenes of the current period and the specified conventional unit start-stop scheme of the day ahead, the scheduling plan of the future set time is obtained by combining the intra-day scheduling optimization model, then the prediction interval is rolled to the next time period, and the intra-day scheduling optimization model is optimized again, until the optimal scheduling scheme of each period of the total time interval is generated; wherein the construction process of the intra-day scheduling optimization model comprises:

[0012] An intra-day rolling scheduling model containing reliability constraints and multiple comprehensive uncertain scenes is established;

[0013] According to the Benders decomposition method, the intra-day rolling scheduling model is decomposed into a main problem and multiple comprehensive typical scene sub-problems, wherein the main problem solves the start-stop plan of the fast-starting unit, and the sub-problem solves the output plan of the conventional unit and the output plan of the fast-starting unit;

[0014] The outage quantity index of the comprehensive typical scene is limited, the start-stop plan of the fast-starting unit obtained by solving the main problem is substituted into the sub-problem, the Benders cut is generated, the lower bound obtained by solving the main problem and the upper bound obtained by solving the sub-problem are set as the convergence criterion, the cycle degree is limited, and it is judged whether the multiple sub-problems reach the convergence stage, if not, the Benders cut is added as a constraint condition to the main problem, and the main problem and the sub-problem are iteratively solved, until the multiple sub-problems all satisfy the convergence, and the iteration is ended, and the optimization value obtained at this time is taken as the optimal intra-day scheduling scheme of the current period.

[0015] Further, the random wind speed sequence sample is obtained based on historical wind speed data and load data, considering source load cross correlation, comprising:

[0016] Based on historical wind speed data, a wind speed cumulative distribution function is established in combination with statistical Weibull probability distribution parameters.

[0017] Based on historical wind speed data and load data, a rank correlation coefficient of the source load sequence is calculated, and a random edge distribution value of the wind speed sequence with the rank correlation coefficient of the source load sequence is constructed; in combination with an inverse function of the wind speed cumulative distribution function, the edge distribution value is converted into a random wind speed sequence sample.

[0018] Further, the source load cross correlation matrix is:

[0019]

[0020] Wherein, ρ xy is a product moment correlation coefficient of K pairs of samples (x i , y i ) of random variables,

[0021] Or

[0022] The random wind speed sequence sample is converted into a wind energy sequence, comprising: in the case of known wind speed scenario sequence, the wind speed scenario is projected onto the wind energy sequence scenario, so as to indirectly obtain a wind energy sequence scenario set, and the specific projection relationship is:

[0023]

[0024] Wherein, P W,t is the wind power output, v t is the wind speed at time t, v in is the wind turbine cut-in wind speed, v rate is the rated wind speed, v out is the cut-out wind speed, is the rated wind power output.

[0025] Or

[0026] The autocorrelation of the wind speed time sequence is:

[0027]

[0028] Wherein, ζ i,i+k is the sequence correlation coefficient between any two time values x i and x i+k , μ is the mean of the time sequence, and N is the length of the wind speed time sequence.

[0029] Further, the intra-day rolling scheduling model comprises an objective function and constraint conditions:

[0030] The objective function is:

[0031]

[0032] The constraint conditions are:

[0033] The power flow constraint in the prediction scenario is:

[0034]

[0035] The power flow constraint in the typical scenario is:

[0036]

[0037] The power balance constraint in the prediction scenario is:

[0038]

[0039] The power balance constraint in the multiple typical scenarios is:

[0040]

[0041] The output limit constraint of the fast-start unit and the conventional unit in the prediction scenario is:

[0042]

[0043] The output limit constraint of the fast-start unit and the conventional unit in the typical scenario is:

[0044]

[0045] The reliability constraint is:

[0046]

[0047] Wherein, t0 is the current time period at the time of executing the optimization process, h roll represents the total number of windows, ΔH is the time of a unit window, N G is the total number of conventional units in the system, N S is the total number of generated comprehensive typical scenarios, N GF is the total number of fast-start units in the system, π S is the probability of occurrence in the current scenario, are the conventional unit m and the fast-start unit m F at t1 in the comprehensive typical scenario, respectively represent the start-stop process state of the m F fast-start unit No. C operation1 C operation2 C up C down C l max Pmax,l b N bl ψ Pw,t1 Dw,t1 Pw,t1 Pw,t1 Pmin Pmax S P EENS t1 EENS max EENS

[0048] Further, in the main problem solving process, the original objective function of the intra-day rolling scheduling model is changed to only contain integer problem solving, the rest is represented by a parameter α, and the constraint condition contains the 0-1 constraint of the fast-starting unit; in the sub-problem solving process, the objective function only contains four variables of the unit output of the conventional unit and the fast-starting unit under the comprehensive typical scene and the prediction scene.

[0049] Further, the start-stop plan of the fast-starting unit obtained by solving the main problem is substituted into the sub-problem to generate a Benders cut, which comprises:

[0050] The optimal fast-starting unit start state obtained by the main problem is transmitted to the constraint condition of the sub-problem, the optimal value of the sub-problem is solved, the state of the intra-day rolling scheduling model is judged, if the model has no solution at this time, the polar line value of the model is solved, and the Benders feasible cut is created; if the model is the optimal solution at this time, the pole point value of the model is solved, and the Benders optimization cut is created.

[0051] The second aspect of the present application provides an intra-day scheduling scheme generation device considering the cross correlation of source and load, which comprises:

[0052] a wind energy scene generation module, configured to obtain random wind speed sequence samples based on historical wind speed data and load data, taking into account source load cross correlation, convert the random wind speed sequence samples into wind energy sequences in combination with autocorrelation of the wind speed time sequence, and obtain a wind energy output scene set;

[0053] an intra-day dispatching scheme generation module, configured to establish a rolling optimization framework based on the wind energy output scene set, obtain a dispatching plan for a future set time based on wind power, load typical scenes of a current time period and a conventional unit start-stop scheme designated in day-ahead, and then optimize the dispatching plan by means of an intra-day dispatching optimization model, and then roll the prediction interval to a next time period, and again optimize the dispatching plan by means of the intra-day dispatching optimization model until optimal dispatching schemes for each time period of a total time interval are generated; wherein the intra-day dispatching optimization model comprises:

[0054] establishing an intra-day rolling dispatching model containing reliability constraints and multiple comprehensive uncertain scenes;

[0055] decomposing the intra-day rolling dispatching model into a master problem and multiple comprehensive typical scene sub-problems according to a Benders decomposition method, wherein the master problem is used to solve start-stop plans of fast-starting units, and the sub-problems are used to solve output plans of conventional units and fast-starting units;

[0056] adopting a loss-of-load index of the comprehensive typical scene to limit the start-stop plans of the fast-starting units obtained by solving the master problem, and then generating a Benders cut by substituting the start-stop plans of the fast-starting units into the sub-problems, setting a convergence criterion for a lower bound obtained by the master problem and an upper bound obtained by the sub-problems, limiting a degree of circulation, judging whether the multiple sub-problems reach a convergence stage, adding the Benders cut as a constraint condition to the master problem if the multiple sub-problems do not reach the convergence stage, and iteratively solving the master problem and the sub-problems until the multiple sub-problems all satisfy the convergence, and then ending the iteration and taking an optimized value obtained at this time as an optimal intra-day dispatching scheme of a current time period.

[0057] A third aspect of the present application provides a computer readable storage medium.

[0058] A computer readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the intra-day dispatching scheme generation method considering source load cross correlation.

[0059] A fourth aspect of the present application provides a computer device.

[0060] A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the intra-day dispatching scheme generation method considering source load cross correlation when executing the program.

[0061] A fifth aspect of the present application provides a program product.

[0062] A program product, which is a computer program product, comprises a computer program, characterized in that the computer program is executed by a processor to realize the steps of the method for generating an intraday scheduling scheme considering source-load cross-correlation.

[0063] Compared with the prior art, the present application has the beneficial effects that:

[0064] 1、The present application considers the influence of wind speed autocorrelation and source-load cross-correlation on intraday scheduling in the intraday scheduling process, introduces correlation on the basis of a traditional probability model, constructs a wind power output scene with correlation and probability statistical characteristics, models the correlation of time series based on a correlation coefficient matrix, Cholesky decomposition and Copula theory, obtains a random wind speed sequence considering correlation, then converts the wind speed sequence into a wind power output scene sequence based on a wind speed-wind energy conversion model, then constructs a scheduling model based on the scene, further optimizes, and the obtained intraday scheduling scheme is more conducive to scheduling personnel to make more optimal decisions.

[0065] 2、The present application proposes a rolling optimization framework that can realize dynamic updating of predicted input data, and creates multiple typical scenes, considers typical scenes of wind power and load, establishes an intraday rolling scheduling model containing reliability constraints, and simultaneously considers the economy and reliability of a power system; through dynamic updating of predicted input data, the latest wind power load prediction data is used, and a Benders decomposition algorithm is used to solve the scheduling plan of each time window until the unit scheduling scheme of the total time interval (24 hours) is obtained.

[0066] 3、The present application proposes using a Benders decomposition algorithm to solve the intraday rolling scheduling model, dividing the mixed integer programming problem in the objective function into a main problem and a sub-problem under multiple scenes, and can obtain a unit scheduling scheme considering scene uncertainty and N-1 criteria in the future 4 hours under the current time window, wherein the unit scheduling scheme contains the output plan of a conventional unit and the start-stop and output plan of a fast-starting unit; finally, the scheduling unit plan of the first window in each unit scheduling scheme is executed as the optimal scheduling plan of the next period; after the window rolls to the next time period, a new round of optimization process is performed and the optimal scheduling plan thereafter is executed.

[0067] 4. The model is applied to the RBTS system, and the rationality and effectiveness of the model are verified, the cost-optimal scheduling plan in each execution cycle range of the system can be efficiently solved within a specified time under the condition of meeting the system reliability requirements, the method is more beneficial to the scheduling personnel to make more optimal decisions, improves the consumption rate of the system wind power, and guarantees the reliability of the high proportion of wind power system connected to the power grid.

[0068] Advantages of the additional aspects of the application will be in part apparent from the descriptions below, will in part be apparent from the descriptions below, or will be learned by practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0069] The drawings constituting a part of the specification of the application are used to provide further understanding of the application, the illustrative embodiments of the application and the description thereof are used to explain the application, and do not constitute improper limitation on the application.

[0070] Figure 1 is a kind of day scheduling scheme generation method flow chart considering source load cross correlation provided by the embodiment of the application;

[0071] Figure 2 is the 24-hour wind power output and load demand data example curve provided by the embodiment of the application;

[0072] Figure 3 is the example diagram of the rolling optimization framework provided by the embodiment of the application;

[0073] Figure 4 is the day-ahead start-stop plan diagram of conventional unit provided by the embodiment of the application;

[0074] Figure 5 is the scheduling plan diagram before rolling optimization provided by the embodiment of the application;

[0075] Figure 6 is the scheduling plan diagram after rolling optimization provided by the embodiment of the application. DETAILED DESCRIPTION

[0076] The application will be further described below in combination with the drawings and embodiments.

[0077] It should be pointed out that the following detailed description is all exemplary, and is intended to provide further description of the application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as generally understood by those skilled in the art to which the application belongs.

[0078] It is to be understood that the terms used herein are for the purpose of describing specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, devices, components, and / or combinations thereof, but do not preclude the presence or addition of one or more other features, steps, operations, devices, components, and / or combinations thereof.

[0079] Terminology

[0080] Day-ahead scheduling optimization model: the input values are short-term predictions of new energy power generation and short-term predictions of user load power consumption, the time scale is 1 h, the execution cycle is 24 h, and the day-ahead scheduling takes the total economy or environmental protection of the system in the execution cycle as the optimization target under the premise of meeting the safety constraints. The execution cycle of the day-ahead scheduling is relatively large, which leads to a certain deviation between the predicted wind power and load values and the actual values.

[0081] Intra-day rolling scheduling: refers to inputting the prediction data of the next 4 hours, solving the operation plan of all units in the next 4 hours within 15 minutes, and solving the output plan of the units in the next 4 hours in the next 15-minute stage, and rolling scheduling according to the time sequence logic. This model aims to improve the consumption capacity of new energy output, reduce the deviation of the whole time scheduling data, and reduce the operation pressure of the standby units in the system.

[0082] Real-time coordination model: the real-time coordination control execution cycle is 5 minutes. This model takes the intra-day rolling curve as a reference to real-time coordinate control operation strategy to correct the engineering conditions in the actual problem, which is used to reduce the deviation, which is a further correction and verification of the intra-day rolling optimization scheduling. The disadvantage of this model is the complexity of task scheduling, environmental uncertainty, and the need to establish multiple constraint conditions, which increases the complexity of the model and the difficulty of operation.

[0083] Embodiment one

[0084] As shown in Figure 1 , the embodiment provides a day-ahead scheduling scheme generation method considering source-load cross correlation, including the following steps:

[0085] Step 1: obtaining the start-stop plan of the conventional unit, the basic parameters in the system, the predicted value of the new energy output, and the predicted value of the load demand;

[0086] Step 2: based on the historical wind speed data and load data, considering the source-load cross correlation, obtaining a random wind speed sequence sample, combining the autocorrelation of the wind speed time sequence, converting the random wind speed sequence sample into a wind energy sequence, and obtaining a wind energy output scene set, specifically including:

[0087] Step 201: According to the historical data of typical wind speed in the region, the Weibull probability distribution parameters are calculated, and the wind speed cumulative distribution function model is established based on parameter estimation:

[0088] The probability distribution of wind speed obeys the classical two-parameter Weibull distribution, and its probability density function f W(s,k) (v) and cumulative distribution function F W(s,k) (v) are as follows:

[0089]

[0090] Where v is the wind speed, s is the scale parameter of Weibull distribution, and k is the shape parameter of Weibull distribution.

[0091]

[0092] According to formulas (3) and (4), the average wind speed and the wind speed standard deviation σ can be calculated, where is the gamma function.

[0093] The standard unit value of wind speed standard deviation determines the shape parameter k of Weibull distribution, and the scale parameter of Weibull distribution is proportional to the average wind speed. Therefore, when the historical wind speed data of a certain region is known, the Weibull distribution parameters of wind speed can be approximately estimated by calculating the average wind speed and standard deviation. First, the shape parameter k can be approximately estimated by formula (4), and then the scale parameter s can be estimated by the average wind speed and formula (3).

[0094] Step 202: Based on the historical wind speed data and load data, the rank correlation coefficient ρ xy of the source load sequence is calculated, and the Gaussian Copula function is used to construct the random edge distribution value of the wind speed sequence with the rank correlation coefficient ρ xy ; Based on the inverse function of Weibull cumulative distribution function, the edge distribution value is converted into wind speed sequence sample, and a plurality of random wind speed sequences F considering the cross correlation of source load are obtained.

[0095] Time series cross correlation describes the degree of correlation between different time series, which arranges random variables from low to high, and then calculates their respective rank Pearson correlation coefficients, i.e. product moment correlation coefficients, to obtain the correlation of random variables. If K pairs of samples (x i ,y i ) of random variables (X,Y) are given, the product moment correlation coefficient ρ xy of the samples can be expressed as:

[0096]

[0097] wherein, K is the sample size of the pair (x i ,y i );

[0098] Therefore, the time series cross-correlation coefficient matrix P can be expressed as:

[0099]

[0100] The Copula function is a function that connects the joint distribution of multiple random variables with their respective marginal distributions, and the membership function is as follows

[0101] F(x1,x2,,x n )=C(F(x1),F(x2),…,F(x n )) (7),

[0102] The Gaussian Copula function is used to generate a random number vector with rank correlation coefficient ρ xy , and the edge distribution value of the wind speed sequence is obtained, and then the inverse function of the cumulative distribution function is used to convert the edge distribution value into a random variable distribution sample.

[0103] H=F -1 (U) (8),

[0104] Wherein, H is the random variable distribution sample, F -1 (·) is the inverse function of the cumulative distribution function, and U is the edge distribution value of the wind speed sequence.

[0105] Step 203: Calculate the time series autocorrelation coefficient matrix S of the historical wind speed sequence, perform Cholesky decomposition, and apply the decomposed correlation matrix to the wind speed sequence F.

[0106] The time series autocorrelation coefficient matrix describes the correlation between different time values in the same time series. For a time series {X t} 1×N ={x1,x2,…,x N}, the sequence correlation coefficient ζ i between any two time values x i+k and x i,i+k can be expressed as:

[0107]

[0108] The time series autocorrelation coefficient matrix S g can be expressed as:

[0109]

[0110] By performing Cholesky decomposition on the correlation coefficient matrix S g , a lower triangular matrix D is obtained, and then according to formula (12), a random variable sequence G with the correlation coefficient matrix S g can be obtained.

[0111] S g = DD T (11),

[0112] G = DF (12),

[0113] Step 204: Based on the wind speed-wind energy conversion model, considering the autocorrelation of the wind speed time series and the cross correlation of the source load, the wind speed sequence is converted into the wind energy sequence, and the wind energy output scene set is reconstructed.

[0114] In the case of known wind speed scene sequence, the wind speed scene is projected onto the wind power scene, thereby indirectly obtaining the wind power scene set. The relationship curve between wind turbine output power and wind speed is called the standard power characteristic curve of the wind turbine, and its expression is shown in formula (13). The wind power output P W,t at time t is:

[0115]

[0116] Where v t is the wind speed at time t, v in is the cut-in wind speed of the wind turbine, v rate is the rated wind speed, and v out is the cut-out wind speed.

[0117] Step 3: After generating the typical scene, a rolling optimization framework is established. Since the N-1 safety standard and the uncertainty of multiple integrated scenes of wind power and load scenarios need to be considered in the model, only relying on conventional units cannot guarantee that the system will not lose load. Therefore, the patent increases the fast-starting unit in the model. The main input parameters each time include the latest wind power, load, typical scene and day-ahead specified conventional unit start-stop scheme, as shown in formula (14). At time t0, according to the input wind power output prediction value and load demand prediction value of the [t0, t0+16] time period, the scheduling plan for the next 4 hours (16 time periods) is obtained, but the output value only executes the result of the control window [t0, t0+1]. Then, the prediction interval is rolled to the next time period [t0+1, t0+17], and the scheduling plan for the next 4 hours is obtained by re-optimization, and only the optimization result of 1 window is output. Repeat until the total time interval (24 hours) scheduling plan is generated. Figure 3

[0118] Step 4: Establish an intra-day rolling scheduling model containing reliability constraints and multiple uncertain integrated scenarios. The specific intra-day rolling scheduling model is as follows:​

[0119] The day-ahead rolling scheduling model is to determine the optimal cost of unit output and unit start-stop scheduling plan under the premise of meeting the economic and safe operation of the power system, and improve the accommodation capacity of the system wind power in the process. The application proposes an optimization model with the minimum total cost as the target, and the objective function includes the operation cost of conventional units and the operation cost and start-stop cost of fast-starting units, as follows:

[0120]

[0121] In the formula, t0 is the current time period when the optimization process is executed, h roll represents the total number of windows (the value is 16 in this patent), ΔH is the time of a unit window, N G is the total number of conventional units in the system, N S is the total number of generated comprehensive typical scenarios, N GF is the total number of fast-starting units in the system, π S is the probability of occurrence in the current scenario, respectively, m is the conventional unit and m F is the output of the fast-starting unit under the comprehensive typical scenario at t1, respectively, m F is the start-stop process state of the fast-starting unit No. is 1, is 0; when the fast-starting unit changes from the start state to the stop state, the value is opposite). is the start-stop state of the conventional unit, which is a 0-1 variable determined by the day-ahead scheduling plan input data, C operation1 , C operation2 are the costs of generating unit electric energy of the conventional unit and the fast-starting unit, respectively, C up , C down are the start and stop costs of the fast-starting unit.

[0122] The constraint conditions in the model include system power balance constraint, unit output limit constraint, system network security constraint and reliability constraint, and the specific model constraint conditions are as follows:

[0123] The system network security constraint considers the DC power flow constraint in this embodiment, and uses the power transfer distribution factor PTDF, without considering the intermediate variable node voltage phase angle, reducing the number of variables in the scheduling optimization, and the power flow constraint expression form under the prediction scenario and the typical scenario is as follows:

[0124] s.t.

[0125]

[0126] F = Fmax - Fmin l max Fmax represents the maximum flow of line l, N b N represents the total number of nodes in the system, ψ bl ψ represents the power transfer distribution factor (PTDF) in the system, respectively represent the predicted scenario output value and the comprehensive typical scenario output value of the wth unit at t1 in the dispatching plan, respectively represent the predicted scenario demand value and the comprehensive typical scenario demand value of the bth node at t1 in the system;

[0127] The system power balance constraint includes the power balance constraint in the predicted scenario and the power balance constraint in the multiple typical scenarios.

[0128] s.t.

[0129]

[0130] In the above formula, and respectively represent the unit output of the conventional unit and the fast-start unit in the predicted scenario.

[0131] The unit output limit constraint includes the output limit constraint of the fast-start unit and the conventional unit in the predicted scenario and the typical scenario, and the unit output in the typical scenario is constrained by the unit output in the predicted scenario.

[0132] s.t.

[0133]

[0134] In the formula, Fmin represents the minimum output of the conventional unit, Fmax represents the maximum output of the conventional unit, Fmin represents the minimum output of the fast-start unit, Fmax represents the maximum output of the fast-start unit,

[0135] The reliability constraint is limited by the outage index of the comprehensive typical scenario, and the constraint condition is as follows:

[0136]

[0137] In the formula, P S represents the probability of the occurrence of the typical scenario; represents the load shedding amount of the bth node at t1 in the typical scenario, represents the expected power supply shortage amount EENS in the window period in which t1 is located, max represents the maximum outage index in each window period.

[0138] Among them, the objective function (11) is the economic optimization (minimum cost) of the intraday rolling dispatch model, which consists of the total power generation cost of conventional units, the total power generation cost of fast-start units, and the start-up and shutdown costs in all comprehensive typical scenarios; Equations (12) and (13) are the network security constraints of the system, including the DC power flow constraints under the prediction scenario and the comprehensive typical scenario S; Equations (14) and (15) are the system power balance constraints of the system under the prediction scenario and the comprehensive typical scenario S; Equations (16) and (17) are the upper and lower limit constraints of the output of conventional units and fast-start units under the prediction scenario; Equations (18) and (19) constrain the output of conventional units and fast-start units under different comprehensive typical scenarios S to be within the range of the intraday dispatch plan of the units under the prediction scenario; Equation (20) restricts the power loss index of the dispatch unit output plan obtained by optimization under different comprehensive typical scenarios to be within the reliable range.

[0139] Step 5: Decompose the problem to be solved in the model into the main problem of unit start-up and shutdown and N S This involves multiple sub-problems within a single scenario. By iteratively solving the main problem and its sub-problems over the time interval [t0, t0+16], the solution speed can be improved. This process mainly includes four steps:

[0140] like Figure 5 As shown, step 501: Establish the main problem MP

[0141] The objective function of the model proposed in step 4 is divided into integer and continuous problems. The solution to the problem is divided into two parts: the integer problem is the start-up and shutdown process states of the fast-start units, and the continuous problem is the unit output of the conventional and fast-start units. The main problem is the start-up and shutdown process states of each unit during fast startup. In solving this problem, the original objective function becomes a solution that only includes the integer problem, with the remaining parts represented by α. The constraints include 0-1 constraints for the fast-start units. The specific main problem is as follows:

[0142]

[0143] st

[0144]

[0145] Step 502: Establish subproblem SP

[0146] The subproblem is the solution to the optimal output plan for each unit in the initial objective function, which includes... With 4 variables, write all the constraints in step 4 as follows: The objective function and constraints for the specific subproblems are as follows:

[0147]

[0148] s.t.

[0149]

[0150]

[0151] Step 503: generating Benders cut

[0152] The optimal quick-start unit start state z obtained by the main problem is transmitted to the constraint condition of the sub-problem, and the optimal value of the sub-problem is solved, the model state is judged, if the model has no solution at this time, the polar ray value of the model is solved, and the Benders feasibility cut (Benders feasibility cut) is created; if the model is an optimal solution at this time, the polar point value of the model is solved, and the Benders optimality cut (Benders optimality cut) is created. The specific form of Benders cut is as follows:

[0153] Benders feasibility cut:

[0154]

[0155] In the above formula, R represents the set of polar rays of the dual problem of the sub-problem in the iteration process, represents a polar ray of the dual problem, b represents a constant term in the constraint condition of the sub-problem, and B represents the coefficient value of z in the constraint condition containing z in the sub-problem. Benders optimality cut:

[0156]

[0157] In the above formula, P represents the set of polar points of the dual problem of the sub-problem in the iteration process, represents a polar point of the dual problem, b represents a constant term in the constraint condition of the sub-problem, and B represents the coefficient value of z in the constraint condition containing z in the sub-problem.

[0158] Step 504: setting iteration convergence criterion

[0159] The objective function value obtained by the main problem is taken as the lower bound LB, and the objective function value obtained by the sub-problem is added to the integer programming objective value in the main problem to obtain the upper bound UB. It is judged whether the difference rate of the upper bound and the lower bound satisfies the error value, if not, the Benders cut is added to the constraint condition of the main problem, the main problem is solved again, and the z value obtained by solving is transmitted to the sub-problem to repeat steps 503 and 504; if it is satisfied, the cycle is stopped and the optimal value at this time is output as the optimal scheme of the period [t0, t0+16].

[0160]

[0161] Step 6: The prediction window is moved back to the period [t0+1, t0+17], the updated prediction data is read in, step 5 is repeated, the optimal scheduling scheme of this period is solved iteratively, and so on until the execution plan of all time intervals is generated.

[0162] Next, the superiority of the model established by the present application is proved based on the improved 10-100 machine system, and the specific execution process is as follows:

[0163] First, some basic data of the improved 10-100 machine system is obtained. The improved 10 machine system includes a wind farm composed of 10 identical 12MW wind turbines, 7 conventional generators, and 3 fast-starting units. The total installed capacity of the conventional units is 1497MW, and the total installed capacity of the fast-starting units is 165MW. The peak load is 1500MW. For the system of 20 units, the same units as the 10 machine system are taken, and the system load of each period is doubled. The data of other systems are also determined in the same way. The day-ahead conventional unit start-stop plan is solved based on the day-ahead prediction data and the day-ahead scheduling model considering N-1, and the result is shown in Table 1. Figure 5

[0164] Second, based on the historical wind speed data of the wind farm, a wind power output scenario reconstruction method considering wind speed autocorrelation and source-load cross-correlation is proposed. The correlation is introduced based on the traditional probability model to construct a wind power output scenario with correlation and probability statistical characteristics. The correlation of time series is modeled based on the correlation coefficient matrix, Cholesky decomposition and Copula theory to obtain the random wind speed sequence considering correlation. Then, the wind speed sequence is converted into a wind power output scenario sequence based on the wind speed-wind energy conversion model.

[0165] Third, an intra-day rolling scheduling model containing reliability constraints and multiple comprehensive uncertainty scenarios is established. The optimization model with the minimum total cost as the objective function includes the operation cost of conventional units and the operation cost and start-stop cost of fast-starting units. The constraint conditions in the model include system power balance constraints, unit output limit constraints, system network security constraints and reliability constraints.

[0166] Fourth, a day-ahead scheduling scenario generation technique considering source-load cross-correlation and its fast solving method are applied. The algorithm aims to improve the economy and timeliness of the day-ahead scheduling plan result. The problem to be solved in the model is decomposed into a unit start-stop main problem and an N S ​multiple-scenario sub-problems. Through the interactive iteration of the main problem and the sub-problems in the period [t0, t0+16], the solving rate can be improved, different schemes and corresponding total costs, including power generation cost, start-stop machine cost, fast-start machine cost and reliability cost, and a 24-hour daily rolling scheduling scheme are obtained, and a 10-machine scheduling plan is as shown in FIG. 8, wherein G1-G7 respectively represent seven different conventional units, GF1-GF3 represent three different fast-start units, and WT represents the sum of wind turbine generators, and the vertical coordinate represents the output power of the unit. Figure 6

[0167] Step 5, in order to analyze the performance of the daily rolling scheduling model, the model is set to run under the following three examples. Figure 6 It is shown that due to the elimination of the penalty cost caused by load loss, the daily scheduling is more cost-effective. Through the proposed stochastic optimization method, the coordination of non-fast-start and fast-start units compensates for the supply-demand imbalance in the random scene in the daily scheduling process. By analyzing the above results, the model proposed in the application takes into account the economy and reliability.

[0168] Scenario 1: without considering uncertainty.

[0169] Scenario 2: only considering the uncertainty of single unit failure.

[0170] Scenario 3: considering the uncertain scene, including wind power, load demand and N-1 power interruption emergency.

[0171] The specific results of the three scenarios are shown in Table 1.

[0172] Table 1 Comparison of simulation results of 10-machine example under different scenarios

[0173]

[0174] Step 6, in order to analyze the algorithm performance of the method proposed in the application, Table 2 provides a comparison of the calculation time of the 10-100 machine examples using the Benders decomposition algorithm and other algorithms of the application.

[0175] Table 2 Comparison of calculation time of different algorithms (s)

[0176]

[0177]

[0178] ​Through the above data analysis, the rationality and effectiveness of the method of the application can be verified; and comparative analysis of different algorithms is carried out, and it is found that the algorithm of the application has high computational efficiency. In addition, through the analysis of the calculation results, the corresponding results can be obtained. Under the condition of meeting the system reliability requirement, the method helps the dispatching personnel to make more optimal unit scheduling decisions, reduces the system wind curtailment rate, and ensures the supply-demand balance of the high-penetration wind power system.

[0179] Embodiment two

[0180] The embodiment provides a day-ahead scheduling scheme generation device considering source-load cross correlation, which comprises:

[0181] A wind energy scene generation module is configured to obtain random wind speed sequence samples by considering source-load cross correlation based on historical wind speed data and load data, and convert the random wind speed sequence samples into wind energy sequences by combining the autocorrelation of the wind speed time sequence to obtain a wind energy output scene set;

[0182] A day-ahead scheduling scheme generation module is configured to establish a rolling optimization framework based on the wind energy output scene set, obtain a scheduling plan for a future set time by combining a day-ahead specified conventional unit start-stop scheme and a day-ahead scheduling optimization model based on a current period wind power and load typical scene, and then roll the prediction interval to a next time period, and optimize again through the day-ahead scheduling optimization model until an optimal scheduling scheme for each period of a total time interval is generated; wherein the construction process of the day-ahead scheduling optimization model comprises:

[0183] A day-ahead rolling scheduling model containing reliability constraints and multiple uncertainty comprehensive scenes is established;

[0184] According to the Benders decomposition method, the day-ahead rolling scheduling model is decomposed into a main problem and multiple comprehensive typical scene sub-problems, wherein the main problem solves the start-stop plan of the fast-starting unit, and the sub-problem solves the output plan of the conventional unit and the fast-starting unit;

[0185] The outage quantity index of the comprehensive typical scene is used for limitation, the start-stop plan of the fast-starting unit obtained by solving the main problem is substituted into the sub-problem, a Benders cut is generated, the lower bound obtained by the main problem and the upper bound obtained by the sub-problem are set as a convergence criterion, the circulation degree is limited, whether the multiple sub-problems reach the convergence stage is judged, if not, the Benders cut is added to the main problem as a constraint condition, the main problem and the sub-problem are iteratively solved, until the multiple sub-problems all satisfy the convergence, the iteration is ended, and the optimization value obtained at this time is taken as the optimal day-ahead scheduling scheme of the current period.

[0186] Embodiment three

[0187] The embodiment provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement steps of the method for generating an intraday scheduling scheme considering source-load cross correlation.

[0188] Embodiment four

[0189] The embodiment provides a computer device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements steps of the method for generating an intraday scheduling scheme considering source-load cross correlation when executing the program.

[0190] Embodiment five

[0191] The embodiment provides a program product, which is a computer program product, comprising a computer program, and the computer program is executed by a processor to implement steps of the method for generating an intraday scheduling scheme considering source-load cross correlation.

[0192] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a form of a computer program product implemented on one or more computer usable storage media (including but not limited to disk storage and optical storage) containing computer usable program codes.

[0193] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a means for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 The function of one flow or multiple flows and / or blocks Figure 1 The function of one block or multiple blocks.

[0194] These computer program instructions can also be stored in a computer readable storage medium capable of guiding a computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer readable storage medium produce a product including instruction means, which implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The function of one flow or multiple flows and / or blocks Figure 1the function specified in the one or more blocks.

[0195] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide operational steps for implementing the processes in the flowcharts Figure 1 one or more flows and / or blocks Figure 1 the function specified in the one or more blocks.

[0196] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by computer programs instructing related hardware, and the programs can be stored in a computer-readable storage medium. When the programs are executed, the processes of the above-mentioned embodiment methods can be included. The storage medium can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM), etc.

[0197] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for generating intraday scheduling schemes considering source-load cross-correlation, characterized in that, Includes the following steps: Based on historical wind speed and load data, considering the cross-correlation of source and load, random wind speed sequence samples are obtained. Combining the autocorrelation of wind speed time series, the random wind speed sequence samples are transformed into wind energy sequences to obtain a set of wind energy output scenarios. Based on the wind energy output scenario set, a rolling optimization framework is established. Based on the typical wind power and load scenarios of the current period and the conventional unit start-up and shutdown scheme specified a day before, the scheduling plan for the future set time is obtained by combining the intraday scheduling optimization model. Then, the prediction interval is rolled to the next time period and optimized again by the intraday scheduling optimization model until the optimal scheduling scheme for each time period of the total time interval is generated. The construction process of the intraday scheduling optimization model includes: Establish an intraday rolling scheduling model that incorporates reliability constraints and multiple uncertainties. According to the Benders decomposition method, the intraday rolling scheduling model is decomposed into a main problem and several comprehensive typical scenario sub-problems. The main problem solves the start-up and shutdown plan of the fast-start unit, and the sub-problems solve the output plan of the conventional unit and the output plan of the fast-start unit. By using power loss indicators from typical scenarios as constraints, the start-up and shutdown plans of the units obtained from solving the main problem are substituted into the subproblems to generate Benders cuts. Convergence criteria are set for the lower bounds obtained from the main problem and the upper bounds obtained from the subproblems to limit the degree of iteration. It is determined whether multiple subproblems have reached the convergence stage. If not, the Benders cut is added as a constraint to the main problem. The main problem and subproblems are solved iteratively and interactively until all subproblems meet the convergence requirement. The iteration ends, and the optimized value obtained at this time is taken as the optimal intraday scheduling scheme for the current period.

2. The method for generating intraday scheduling schemes considering source-load cross-correlation as described in claim 1, characterized in that, The random wind speed sequence sample obtained based on historical wind speed data and load data, considering the cross-correlation of source and load, includes: Based on historical wind speed data and combined with statistical Weibull probability distribution parameters, a cumulative wind speed distribution function is established. Based on historical wind speed and load data, the rank correlation coefficient of the source load sequence is calculated, and a random marginal distribution value of the wind speed sequence with the rank correlation coefficient of the source load sequence is constructed. Combining the inverse function of the cumulative wind speed distribution function, the marginal distribution value is transformed into a random wind speed sequence sample.

3. The method for generating intraday scheduling schemes considering source-load cross-correlation as described in claim 1, characterized in that, The source load cross-correlation matrix is ​​as follows: Where, ρ xy K pairs of samples (x) are random variables i ,y i The correlation coefficient of the product moments, or The step of converting random wind speed sequence samples into wind energy sequences includes: given a known wind speed scene sequence, projecting the wind speed scene onto a wind energy sequence scene to indirectly obtain a wind energy sequence scene set. The specific projection relationship is as follows: Among them, P W,t For wind power output, v t The wind speed at time t, v in It is the fan cut-in wind speed, v rate It is the rated wind speed, v out It's about cutting off the wind speed. This refers to the rated wind power output. or The autocorrelation of the wind speed time series is: Where, ζ i,i+k For any two time values ​​x i and x i+k The correlation coefficient between the sequences is μ, where μ is the mean of the time series and N is the length of the wind speed time series.

4. The method for generating intraday scheduling schemes considering source-load cross-correlation as described in claim 1, characterized in that, The intraday rolling scheduling model includes an objective function and constraints: The objective function is: The constraints are: The power flow constraints in the prediction scenario are: Current constraints in typical scenarios: Power balance constraints in prediction scenarios: Power balance constraints in various typical scenarios: Output constraints for fast-start and conventional units in predicted scenarios: Output constraints for fast-start units and conventional units in typical scenarios: Reliability constraints: Where t0 is the current time period during the optimization process, and h roll N represents the total number of windows, ΔH is the time per unit window, and N is the total number of windows. G N represents the total number of conventional units in the system. S N represents the total number of comprehensive typical scenarios generated. GF π represents the total number of fast-start units in the system. S The probability of it occurring in the current scene. These are conventional unit m and fast-start unit m, respectively. F The contribution at time t1 in a typical scenario These represent m in the intraday plan at time t1. F The start-up and shutdown status of the No. 1 fast-start unit For the start-up and shutdown status of conventional units, C operation1 C operation2 C represents the cost per unit of electricity generated by conventional units and fast-start units, respectively. up C down S represents the cost of starting and shutting down the unit quickly, respectively. N represents the maximum power flow value of line l. b Let ψ be the total number of nodes in the system. bl The power transfer distribution factor in the system, These represent the predicted output value and the comprehensive typical output value of unit w at time t1 in the scheduling plan, respectively. These represent the load of node b in the system at time t1, representing the predicted demand value and the comprehensive typical demand value, respectively. and The power output of conventional units and fast-start units under the predicted scenarios are respectively; This is the minimum output of a conventional unit. P is the maximum output of a conventional unit. S This indicates the probability of this typical scenario occurring. This represents the load shedding amount at node b at time t1 in this typical scenario. EENS represents the expected power shortage during the window period containing t1. max This represents the maximum power loss index during each window period.

5. The method for generating intraday scheduling schemes considering source-load cross-correlation as described in claim 1, characterized in that, In the process of solving the main problem, the original objective function of the intraday rolling scheduling model is transformed into a solution that only includes integer problems, with the rest represented by the parameter α. The constraints include 0-1 constraints on the fast-start units. In the process of solving the sub-problems, the objective function only includes four variables: the output of the conventional units and the fast-start units under the comprehensive typical scenario and the predicted scenario.

6. The method for generating intraday scheduling schemes considering source-load cross-correlation as described in claim 1, characterized in that, The step of substituting the start-up and shutdown plan of the fast-start unit obtained from solving the main problem into the subproblem to generate Benders cut includes: The optimal start-up state of the unit obtained from the main problem is passed to the constraints of the subproblem. The optimal value of the subproblem is solved, and the state of the intraday rolling scheduling model is determined. If the model has no solution at this time, the polar ray value of the model is obtained, and a Benders feasible cut is created. If the model has the optimal solution at this time, the extreme value of the model is obtained, and a Benders optimization cut is created.

7. A device for generating intraday scheduling schemes considering source-load cross-correlation, characterized in that, include: The wind energy scene generation module is used to obtain random wind speed sequence samples based on historical wind speed data and load data, taking into account the cross-correlation of source loads. Combining the autocorrelation of wind speed time series, the random wind speed sequence samples are transformed into wind energy sequences to obtain a wind energy output scene set. The intraday scheduling scheme generation module is used to establish a rolling optimization framework based on the wind energy output scenario set. Based on the typical wind power and load scenarios of the current period and the conventional unit start-up and shutdown scheme specified a day before, it combines the intraday scheduling optimization model to optimize the scheduling plan for the future set time. Then, the prediction interval is rolled to the next time period and optimized again by the intraday scheduling optimization model until the optimal scheduling scheme for each time period of the total time interval is generated. The construction process of the intraday scheduling optimization model includes: Establish an intraday rolling scheduling model that incorporates reliability constraints and multiple uncertainties. According to the Benders decomposition method, the intraday rolling scheduling model is decomposed into a main problem and several comprehensive typical scenario sub-problems. The main problem solves the start-up and shutdown plan of the fast-start unit, and the sub-problems solve the output plan of the conventional unit and the output plan of the fast-start unit. By using power loss indicators from typical scenarios as constraints, the start-up and shutdown plans of the units obtained from solving the main problem are substituted into the subproblems to generate Benders cuts. Convergence criteria are set for the lower bounds obtained from the main problem and the upper bounds obtained from the subproblems to limit the degree of iteration. It is determined whether multiple subproblems have reached the convergence stage. If not, the Benders cut is added as a constraint to the main problem. The main problem and subproblems are solved iteratively and interactively until all subproblems meet the convergence requirement. The iteration ends, and the optimized value obtained at this time is taken as the optimal intraday scheduling scheme for the current period.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the intraday scheduling scheme generation method that considers source-load cross-correlation as described in any one of claims 1-6.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the intraday scheduling scheme generation method considering source-load cross-correlation as described in any one of claims 1-6.

10. A program product, said program product being a computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps in the intraday scheduling scheme generation method considering source-load cross-correlation as described in any one of claims 1-6.

Images