Index-driven long-term wind power curve closed-loop feedback generation method

Abstract

The application discloses a kind of index-driven medium and long-term wind power curve closed-loop feedback generation method, constructs Markov chain-autoregressive wind power curve generation model, determines basic parameter and hyperparameter;Index-driven medium and long-term wind power curve closed-loop feedback generation model is constructed, on the basis of Markov chain-autoregressive double-layer wind power curve generation model, calculate what model generates specified index of power curve data;According to the statistical characteristic index of model generation curve data, feedback correction is carried out to model parameter, and the model parameter is obtained again on the basis parameter and hyperparameter obtained, to obtain the medium and long-term wind power curve data satisfying demand.The application can give full play to dynamic boundary information update and expert experience and the like information when curve is generated, realize the flexible correction and control to the statistical characteristics of generated curve.

Description

Technical Field

[0001] This invention belongs to the field of power system operation planning technology, specifically relating to an index-driven method for generating a closed-loop feedback of medium- and long-term wind power curves. Background Technology

[0002] Due to the inherent randomness, intermittency, and volatility of wind power, its widespread integration into the power system can have significant negative impacts on many aspects, including system power balance and the stable operation of the electricity market. Therefore, boundary conditions for wind power generation scenarios are often essential information for various system analyses. Specifically, in medium- to long-term power system analysis scenarios, the wind power curve is usually the only boundary condition for wind power generation; thus, the generation of medium- to long-term wind power curves has been a focus of attention in recent years.

[0003] Currently, wind power curve generation models mainly include wind power sequence or wind speed time series models based on autoregressive processes, Markov chain models, stochastic differential equations, and deep learning models. Autoregressive models and their variants can effectively characterize the autocorrelation characteristics of wind power curves, but these models often have strict requirements for data stationarity, and most can only accurately model data that follows a normal distribution. Markov chain models can accurately describe the probability distribution characteristics of wind power curves, but they cannot characterize the unique long-range autocorrelation of wind power curves, resulting in significant deviations from the actual characteristics of the curves, and therefore cannot model wind power curves with high time resolution. Stochastic differential equation models are typically used to model data at the second level and higher resolution, and can be used to study the dynamic processes of systems. Deep learning models require a large amount of data as a training set, but the actual amount of annual wind power curve data often falls short of this requirement, thus making them unsuitable for direct application in generating medium- to long-term curves such as monthly or yearly curves. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide an index-driven closed-loop feedback generation method for medium- and long-term wind power curves, which can give full play to the dynamic boundary information update and expert experience information during curve generation, and realize flexible correction and control of the statistical characteristics of the generated curves, thereby solving the technical problem of flexible and high-precision simulation of single-objective medium- and long-term wind power curves.

[0005] The present invention adopts the following technical solution:

[0006] An indicator-driven closed-loop feedback method for generating medium- and long-term wind power curves includes the following steps:

[0007] S1. Construct a Markov chain-autoregressive wind power curve generation model and determine the basic parameters and hyperparameters;

[0008] S2. Construct an index-driven medium- and long-term wind power curve closed-loop feedback generation model. Based on the Markov chain-autoregressive two-layer wind power curve generation model obtained in step S1, calculate the specified index of the power curve data generated by the Markov chain-autoregressive two-layer wind power curve generation model.

[0009] S3. Based on the statistical characteristic index of the generated curve data of the Markov chain-autoregressive two-layer wind power curve generation model obtained in step S2, the model parameters are corrected by feedback. Based on the basic parameters and hyperparameters obtained in step S1, the corrected model parameters are obtained again to obtain medium- and long-term wind power generation curve data that meet the requirements.

[0010] Preferably, in step S1, a Markov chain-autoregressive wind power curve generation model is constructed based on the static data of the wind power generation system to be modeled and the long-term wind power curve data of the wind power generation system. The Markov chain-autoregressive wind power curve generation model includes a daily average power model based on Markov chain and a daily fluctuating power curve model based on autoregression. Initial parameters and hyperparameters are estimated based on the collected raw data.

[0011] More preferably, the parameter estimation of the Markov chain-autoregressive two-layer wind power curve generation model aims to maximize the likelihood function of the original data, and the parameters include:

[0012] Markov chain parameters: state transition matrix, initial state probability;

[0013] Relevant parameters for autoregressive models: regression coefficients, noise variance;

[0014] Hyperparameters: number of Markov chain states, order of autoregressive model.

[0015] More preferably, the static data of the wind power generation system to be modeled includes: the installed capacity information of the wind power generation system to be modeled, the current installed capacity or the installed capacity and change data of the wind power generation system within the modeling period;

[0016] The long-term wind power curve data of the wind power generation system includes: the power generation data of the wind power generation system to be modeled; based on the modeling period, the data is in units of years or months, with a time resolution of fifteen minutes or one hour.

[0017] More preferably, the daily average power model based on Markov chains is:

[0018] p(X0=S i ) = p i

[0019] p(X d =S j |X d-1 =S i ) = pji

[0020] Where i,j=1,2,…,M,p i The first state X0 is S i The probability of; p ji The previous day's status was S. i Under the condition that the state is S the next day j The probability; M is the number of states in the Markov model; X d The state on day d is obtained by discretizing the wind power output.

[0021] The autoregressive intraday volatility power curve model is as follows:

[0022]

[0023] in, ε represents the coefficients of the autoregressive model; r represents the order of the autoregressive model; ε t The noise in the autoregressive model follows a normal distribution N(0,σ). ε 2 ), w t The relative power at time t is the daily wind power output minus the daily average power output; w t ′ represents the intraday relative corrected power of wind power that approximately follows a normal distribution after Z-Score standardization.

[0024] Preferably, in step S2, the specified index is:

[0025] Monthly power generation I1: A measure of the reliable power generation capacity of a wind power system on a monthly time scale;

[0026] Daily power generation fluctuation I2: measures the relative change in the average power of a wind power system between two consecutive days;

[0027] Low output duration I3: Characterizes the degree of sustained low daily average power generation capacity of the wind power system, for a specified threshold ε;

[0028] Intraday Anti-Peak Shaving Intensity I4: The characteristic of wind power generation output being lower during the day and higher at night is opposite to the peak shaving demand of the power grid operation. Its essence is the linear correlation between wind power generation and load electricity consumption.

[0029] Daily peak and off-peak power generation I5: Reflects the ability of the wind power system to support the system load during peak and off-peak periods relative to the average of the day;

[0030] Intraday peak-valley power output volatility I6: Reflects the ability of the wind power generation system to stably support the system load during peak and valley periods; it is measured by the variance of the relative power during peak and valley periods through the metering system.

[0031] More preferably, monthly power generation I1:

[0032]

[0033] wherein, W m is a set of wind power curves of the m-th month, m=1,2,…,12, w d is the daily average value of wind power;

[0034] Daily power generation fluctuation I2:

[0035]

[0036]

[0037] wherein, I 2,1 and I 2,2 are calculated values of the indicator relative to the previous day or the next day;

[0038] Duration of low output I3:

[0039]

[0040] wherein, d1<d2 are any two days within the range, ε3 is a low output threshold constant between 0 and 1 set by the user;

[0041] Intraday anti-peak shaving intensity I4:

[0042] I4=Cov(w t ,l t )

[0043] wherein, Cov(w t ,l t ) is the covariance of the intraday wind power curve and the load power curve;

[0044] Power generation in intraday peak and valley periods I5:

[0045]

[0046]

[0047]

[0048] wherein, W v is the valley load period; W p1 is the morning peak load period; W p2 is the evening peak load period;

[0049] Output fluctuation in intraday peak and valley periods I6:

[0050]

[0051]

[0052]

[0053] in, It is the time of Guhe (a type of lotus). This is the peak lotus season in the morning. This is the peak load period in the evening.

[0054] Preferably, in step S3, the correction of the model parameters specifically involves:

[0055] Monthly power generation I1:

[0056] In the Markov chain model, let the power distribution on day d be p. d =[p d (1),…,p d (M)] T The state transition matrix is ​​A = (p ij ) M×M Then, for any date d and day interval Δd, the Markov model has one and only one steady-state distribution π = [π1, π2, ..., π]. M ] T The condition Aπ = π is satisfied, and for any initial state distribution p0, when t → ∞, A t p0 = π; Suppose the current state probability distribution has reached the steady-state distribution π of the state transition matrix A, but the desired data distribution is p. Construct A' through an iterative format so that p is the steady-state distribution of A', and π smoothly transitions to p through A'; The iterative process converges and satisfies linear convergence, and A is constrained. (k+1) ∈[0,1] N×N And normalize the matrix column-wise to ensure the non-negativity and regularity of the discrete probability distribution;

[0057] Daily power generation fluctuation I2:

[0058] When the Markov chain reaches a stationary distribution, this probability will be time-invariant (d omitted), and the expectation will be p(I). 2,1 ≥ξ 2,1 The target probability is ζ. 2,1 Corrected joint probability distribution p t (j,i), and through p ji =p d (j,i) / p d-1 (i) The state transition matrix is ​​obtained by reconstruction;

[0059] Low output duration I3:

[0060] The basic properties of the outer layer's daily average power Markov chain model are obtained from The I3 index of the generated data follows a discrete exponential distribution. The steady-state distribution of the outer model is adjusted to control the duration of low output I3.

[0061] Intraday reverse peak intensity I4:

[0062] Since the distribution transformation is a bijective transformation and is an order-preserving mapping, that is... Therefore, w' can be used. t Equivalent replacement of w t The calculated index I'4 is used to replace I4 in the analysis, and is denoted as l'. t =l t -El t ,

[0063] In the inner autoregressive model, the power simulation data w' for a given day after distribution transformation... t , t=1,…,24 follow a multivariate normal distribution N(μ,Σ),

[0064] While keeping the daily average power constant, adjust the simulated mean μ = μ + Adjust μ + Mean is 0: μ + =μ + (k) -1 T μ + (k) / 24, after mean adjustment, let u and v be μ respectively. + The first and last two elements are used to correct the initial value of the intraday fluctuation power curve simulation of the inner AR model to w. d t +w d -w d+1 +vu;

[0065] Daily peak and off-peak power generation I5:

[0066] Marginal distribution of w' v = [w',…], t∈W v It still follows a multivariate Gaussian distribution, therefore I' 5,1 ~N(1 T μ v ,1 T Σ v 1), μ v For W v μ within the time period + , Σ v W of power during the time period v Let the expected value be EI' 5,1 The target value is ξ 5,1 Make corrections;

[0067] Intraday peak-to-trough power output volatility I6:

[0068] Use w' t Replace w t The calculated index I'6 is used to calculate the dispersion of the population for a multivariate normal distribution, using the determinant of the matrix I'. 6,1 =|Σ v |, let the expectation be I' 6,1 The target value is ξ 6,1 Adjust the variance matrix Σ of the power curve. + ΣΣ T + , Σ + =diag(σ +24 ,σ +23 ,…,σ +1 ), and make corrections.

[0069] More preferably, in the monthly power generation I1, the probability distributions of the states at times d and d+Δd satisfy p d+Δd =A Δd p d The iteration format is as follows:

[0070] A (k+1) =A (k) -(A (k) -I)pπ T ;

[0071] In the daily power generation fluctuation I2, the corrected joint probability distribution p t (j,i):

[0072]

[0073] Among them, I 2,1 The value of indicator I2 is calculated relative to the previous day, and j represents state S. j i represents state S i ∈2 is a non-negative fluctuation threshold constant set by the user, ζ 2,1 For p(I) 2,1 ≥ξ 2,1 The target probability;

[0074] During the low-output duration I3, the I3 index follows a discrete exponential distribution and satisfies:

[0075]

[0076] Where p represents probability. X is the set of states with power below ∈3. d denoted as the state of the wind power sequence on day d, where ∈3 is a non-negative fluctuation threshold constant set by the user;

[0077] Use w' t Equivalent replacement of w t The calculated index I'4 replaces the intraday anti-peak intensity I4, specifically as follows:

[0078]

[0079] Where C is a constant determined by the length of a certain sequence. For the average value operator, l′ is the relative power vector after deducting the average value of the daily load, and w′ is the relative corrected power vector of wind power during the day;

[0080] The daily peak-valley power generation I5 is corrected using the following formula:

[0081]

[0082] Where, ξ 5,1 Expected value of EI' for the valley-load period sub-indicator 5,1 The target value, Let be the mean parameters of the inner autoregressive model at time t. Let be the mean parameters of the inner autoregressive model during the valley loading period after the k-th iteration. The valley-load period is represented by t, where t is the time of day.

[0083] The power fluctuation I6 during the intraday peak and trough periods is corrected using the following formula:

[0084]

[0085] in, Let be the variance parameter of the inner autoregressive model during the valley loading period after the k-th iteration. Let ξ be the standard deviation parameter of the inner autoregressive model at time t after the k-th iteration. 6,1 Sub-indicator I' for the valley-load period 6,1 The target value.

[0086] Secondly, embodiments of the present invention provide an index-driven closed-loop feedback generation system for medium- and long-term wind power curves, comprising:

[0087] The parameter module constructs a Markov chain-autoregressive wind power curve generation model to determine the basic parameters and hyperparameters.

[0088] The indicator module constructs an indicator-driven closed-loop feedback generation model for medium- and long-term wind power curves. Based on the Markov chain-autoregressive two-layer wind power curve generation model obtained from the parameter module, it calculates the specified indicators for the power curve data generated by the Markov chain-autoregressive two-layer wind power curve generation model.

[0089] The output module generates curve data statistical characteristic indicators based on the Markov chain-autoregressive two-layer wind power curve generation model obtained from the indicator module, and performs feedback correction on the model parameters. Based on the basic parameters and hyperparameters obtained from the parameter module, the corrected model parameters are obtained again to obtain medium- and long-term wind power generation curve data that meet the requirements.

[0090] Thirdly, a chip includes 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 steps of the aforementioned indicator-driven medium- and long-term wind power curve closed-loop feedback generation method.

[0091] Fourthly, embodiments of the present invention provide an electronic device, including a computer program, wherein when the computer program is executed by the electronic device, it implements the steps of the above-mentioned indicator-driven medium- and long-term wind power curve closed-loop feedback generation method.

[0092] Compared with the prior art, the present invention has at least the following beneficial effects:

[0093] A closed-loop feedback method for generating medium- to long-term wind power curves, driven by indicators, is crucial for power system analysis on a medium- to long-term timescale, given the increasing proportion of renewable energy in the power system. Since wind power curves are often the only boundary conditions for wind power generation, their accurate generation is particularly important. Effective feedback correction from key indicators can fully leverage historical data, real-time data, and existing expert experience to improve the accuracy of generated wind power curve data. This allows subsequent methods for power system optimization and medium- to long-term power balance analysis to fully utilize their high accuracy advantages.

[0094] Furthermore, the hyperparameters of the model are determined based on model selection methods such as the Bayesian information criterion, and the model parameters are determined by parameter estimation methods such as maximum likelihood estimation and Monte Carlo simulation. This minimizes the influence of subjective factors of the modeler and provides a scientific and objective basis for generating and simulating medium- and long-term wind power curves.

[0095] Furthermore, maximum likelihood estimation is an application of probability theory in statistics and is one of the methods of parameter estimation. Maximum likelihood estimation aims to maximize the probability of observed data occurrences, using statistical results to infer the probability of events, thus accurately estimating the optimal parameters of the model. It also possesses many excellent properties such as consistency, efficiency, and invariance.

[0096] Furthermore, modeling static data and long-term wind power curve data of wind power generation systems can provide important basic initial information such as the installed capacity and resource endowment characteristics of wind power generation systems. This information is used to preliminarily determine the hyperparameters and parameters of the model, avoiding blind subjective speculation and providing a fundamental guarantee for the accuracy and rationality of model application.

[0097] Furthermore, a high-precision single-objective wind power curve generation model is provided by combining a daily average power model based on Markov chains and an intraday fluctuating power curve model based on autoregression. By fusing a daily-scale Markov model and an hourly-scale autoregressive model to construct a two-layer structure, the model fully leverages the accuracy advantages of Markov chains in modeling probability distributions and autoregressive models in modeling time-series correlations, while also considering the accuracy of the generated curve probability distribution and autocorrelation function. This effectively improves the feature accuracy of medium- and long-term wind power curve simulation.

[0098] Furthermore, considering typical medium- and long-term application scenarios in the power system, based on the actual needs and operational experience of the power sector, and using mathematical statistics as a carrier, representative features of wind power curves are selected to construct a multi-dimensional feature index system. The index values ​​of the curves are generated through calculation models and compared with historical data, which can effectively measure the accuracy of the generated curves and assess whether the generated curves can meet the accuracy requirements.

[0099] Furthermore, based on the indicator-based model parameter correction, the statistical characteristics such as the distribution and correlation of the model-generated data can be adjusted according to the requirements, which can be used for sensitivity analysis dependent on boundary conditions; the generated curve can be effectively and quantitatively adjusted according to a small number of samples to meet the needs of dynamically updating the uncertainty boundary caused by the actual operating conditions of the system with frequently changing uncertainty boundary conditions; and it can integrate expert experience to correct the statistical characteristics of the generated curve, so that the boundary conditions that meet the system analysis requirements can still be obtained when the amount of historical data is insufficient.

[0100] It is understandable that the beneficial effects of the second aspect mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0101] In summary, this invention achieves flexible adjustment of the statistical characteristics of the generated curve by introducing a Markov chain-autoregressive two-layer wind power curve generation model and forming a closed loop from the model evaluation module to the parameter estimation module.

[0102] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0103] Figure 1 Flowchart of the simulation process for generating Markov chain-autoregressive two-layer wind power curves;

[0104] Figure 2 To compare the probability distribution and autocorrelation function of the generated curve with historical data, where (a) is the annual wind power distribution and (b) is the autocorrelation function of the annual wind power curve;

[0105] Figure 3 For comparison of different generation curves under each evaluation index, (a) is the monthly power generation I1, (b) is the annual daily power generation fluctuation I2 distribution, (c) is the annual low output duration I3 distribution with ∈3=0.2, (d) is the annual intraday anti-peak intensity I4 distribution, (e) is the annual intraday evening peak power generation I5 distribution, and (f) is the annual intraday evening peak power fluctuation I6 distribution.

[0106] Figure 4 For the comparison of continuous low-output scenarios of historical and generated data, (a) is the curve of historical and generated data, and (b) is the distribution map of historical and generated data.

[0107] Figure 5 A comparison of the anti-peak shaving scenarios of historical and generated data is provided, where (a) is the historical and generated data curves for the first week, and (b) is the average daily power curve of historical and generated data.

[0108] Figure 6 Wiring diagram for NREL-118 system;

[0109] Figure 7 Comparison of simulation deviations for the NREL-118 system based on different generation curves;

[0110] Figure 8 A schematic diagram of a computer device provided in an embodiment of the present invention;

[0111] Figure 9 This is a block diagram of a chip provided according to an embodiment of the present invention. Detailed Implementation

[0112] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0113] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0114] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0115] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this invention generally indicates that the preceding and following objects have an "or" relationship.

[0116] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0117] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0118] The accompanying drawings illustrate various structural schematic diagrams according to embodiments disclosed in this invention. These drawings are not to scale, and some details have been enlarged for clarity, and some details may have been omitted. The shapes of the various regions and layers shown in the drawings, as well as their relative sizes and positional relationships, are merely exemplary and may deviate from reality due to manufacturing tolerances or technical limitations. Furthermore, those skilled in the art can design regions / layers with different shapes, sizes, and relative positions as needed.

[0119] This invention provides an index-driven closed-loop feedback method for generating medium- and long-term wind power curves. By effectively correcting the feedback from key indicators, it fully leverages historical data, real-time data, and existing expert experience to improve the accuracy of generated wind power curve data. This allows subsequent methods for power system optimization and medium- and long-term power balance analysis to fully utilize their existing high accuracy advantages.

[0120] To fully utilize dynamic changes in conditions and expert feedback during curve generation, adjustments can be made based on feedback during curve generation to correct the statistical characteristics of the model-generated data, such as distribution and correlation, and then used for sensitivity analysis dependent on boundary conditions. Alternatively, when the boundary changes dynamically, the statistical characteristics of the generated curve can be effectively adjusted based on a small sample size. Furthermore, when the amount of historical data is insufficient to meet the model parameter estimation requirements, the statistical characteristics of the generated curve can be corrected to obtain a boundary condition curve that meets the system analysis requirements.

[0121] Please see Figure 1 The present invention discloses an index-driven closed-loop feedback method for generating medium- and long-term wind power curves, comprising the following steps:

[0122] S1. Obtain static data of the wind power generation system to be modeled and long-term wind power curve data of the wind power generation system from relevant departments, and construct a Markov chain-autoregressive wind power curve generation model, including a daily average power model based on Markov chain and an intraday fluctuation power curve model based on autoregression.

[0123] Static data of the wind power generation system to be modeled: Installed capacity information of the wind power generation system to be modeled, which can be the current installed capacity or the installed capacity and changes of the wind power generation system during the modeling period.

[0124] Long-term wind power curve data of wind power generation system: power generation data of the wind power generation system to be modeled; depending on the modeling period, the data can be flexibly combined in terms of year or month, and time resolution of fifteen minutes or one hour.

[0125] Daily average power model based on Markov chain

[0126] The daily average power curve of wind power generation has weak correlation and irregular distribution, making it difficult to characterize using a deterministic parametric distribution. However, the first-order discrete Markov model does not include prior assumptions about the data distribution and can consider the conditional correlation between adjacent data, meeting the modeling requirements. Outer Markov chain model:

[0127]

[0128] Where, p i The first state X0 is S i The probability of; p ji The previous day's status was S. i Under the given conditions, the state on the following day is S. j The probability; M is the number of states in the Markov model; X d The state on day d is obtained by discretizing the wind power output, and it is related to the wind power output w. d The conversion relationship between them is taken as:

[0129]

[0130] The parameters of a first-order discrete Markov model include: the number of states M, and the initial state probability p. i State transition probability p ji Among them, the number of states M is a hyperparameter, which is specified in advance based on the ramp-up conditions of the controllable units within the system:

[0131]

[0132] Where C represents the set of controllable units; RD i RU i The maximum downhill and maximum uphill rates of controllable unit i are divided by the installed capacity G of the wind power system. s The conversion result;

[0133] Initial state probability p i and state transition probability p ji The estimated value is obtained from the data using maximum likelihood estimation, and its expression is:

[0134]

[0135]

[0136] Where N is the number of historical wind power output sequence samples, f ji n The states of the sequence in the nth state sequence sample are determined by S. i Transfer to S j frequency, f i n Let S be the indicator function for the initial state in the nth state sequence sample. i When it is 1, it is 0.

[0137] Autoregressive intraday volatility power curve model

[0138] The power values ​​at each moment of the medium- and long-term wind power curve are reduced by the daily average power w at that moment. d The intraday fluctuating power sequence w of wind power generation can be obtained. t Compared to the daily average power W d On the one hand, intraday fluctuation power w t It will inherit the long-range correlation and daily periodicity characteristics of the original power data; on the other hand, due to the intraday fluctuation sequence w t The average power w of the day was deducted. d Therefore, w tThis can be considered as a sequence with an approximate mean of 0. Considering the aforementioned properties of intraday volatility power, the inner model can be modeled using an autoregressive model capable of characterizing strong correlations in the data:

[0139]

[0140] in, ε represents the coefficients of the autoregressive model; r represents the order of the autoregressive model; ε t The noise in the autoregressive model follows a normal distribution N(0,σ). ε 2 ).

[0141] Since the data generated by the autoregressive model will follow a normal distribution, it is necessary to process the original intraday fluctuation series w. t Make corrections, and record w' t To correct the fluctuation component of wind power at time t within the day, which follows a normal distribution, the specific calculation method is as follows:

[0142]

[0143] Among them, w t f represents the output of wind power generation at time t within a day; S f is the cumulative probability density function of the intraday fluctuation component of wind power generation; N It is the cumulative probability density function of the standard normal distribution.

[0144] The parameters of an autoregressive model include: the order r of the autoregressive model, and the autoregressive model coefficients. and noise variance parameter σ ε r is determined based on the partial autocorrelation function of the data; while and σ ε The maximum likelihood estimation yields the following:

[0145]

[0146]

[0147] in, This is the coefficient vector of the autoregressive model. W represents the corrected intraday fluctuation component of wind power output [w'] t-1 ,…,w' t-r The matrix consists of an N×2 dimensional predictor variable matrix; w is the corrected intraday fluctuation component of wind power output w'. t The resulting N×1 dimensional response variable matrix.

[0148] S2, constructing an index-driven closed-loop feedback generation model for medium- and long-term wind power curves, and calculating specified indicators of the power curve data generated by the model based on the Markov chain-autoregressive two-layer wind power curve generation model obtained in step S1; the system includes indicators reflecting overall power generation, power and electricity fluctuations between adjacent days, intra-day peak shaving and fluctuation characteristics;

[0149] 1) Monthly power generation I1

[0150] Monthly power generation indicator I1: measures the reliable power generation capacity of a wind power generation system on a monthly time scale, and can be measured by calculating the total wind power generation in each month:

[0151]

[0152] wherein, W m is the set of wind power curves for the m-th month, where m = 1,2,…,12.

[0153] 2) Daily power generation fluctuation I2

[0154] Daily power generation fluctuation indicator I2: measures the relative change degree of the average power of the wind power generation system between two adjacent days, and can be measured by calculating the relative change of the average power of two adjacent days:

[0155]

[0156]

[0157] wherein, I 2,1 and I 2,2 are calculated values of the indicator relative to the previous day or the next day, which can be selected specifically according to requirements.

[0158] 3) Low output duration I3

[0159] Low output duration indicator I3: characterizes the degree of持续 low daily average power generation capacity of the wind power generation system. For a specified threshold ∈3, it can be measured by counting the duration during which the daily average power is continuously lower than the specified value:

[0160]

[0161] wherein, d1<d2 are any two days within the range.

[0162] 4) Intra-day anti-peak shaving intensity I4

[0163] The intraday anti-peak shaving intensity index originates from phenomena observed by power system operators in practice. "Anti-peak shaving" originally referred to the characteristic of wind power output being lower during the day and higher at night, contrary to the peak-shaving demand of the power grid. Its essence is the linear correlation between wind power generation and load consumption. Therefore, this invention defines the intraday anti-peak shaving intensity index I4 as the wind power daily power curve w t Type-2 load daily curve l t Correlation, and measured using the Pearson correlation coefficient:

[0164] I4=Cov(w t ,l t )

[0165] 5) Daily peak and off-peak power generation I5

[0166] Peak-valley power generation index I5: Reflects the wind power system's ability to support system load during peak and valley periods relative to the daily average. It is measured by the metering system through the relative power generation during peak and valley periods.

[0167]

[0168]

[0169]

[0170] Among them, W v It is the valley and lotus period; W p1 The peak load period in the morning; W p2 This is the peak load period in the evening.

[0171] 6) Intraday peak and trough power output volatility I6

[0172] Intraday peak-valley power generation index I6: This reflects the ability of the wind power system to stably support the system load during peak and valley periods. It is measured by the variance of the relative power during peak and valley periods using the metering system.

[0173]

[0174]

[0175]

[0176] Among them, W v It is the valley and lotus period; W p1 The peak load period in the morning; W p2 This is the peak load period in the evening.

[0177] S3. Based on the statistical characteristic indicators of the model-generated curve data obtained in step S2, the model parameters are corrected by feedback. Based on the basic parameters and hyperparameters obtained in step S2, the corrected model parameters are obtained again to obtain medium- and long-term wind power generation curve data that meet the requirements.

[0178] 1) Monthly power generation I1

[0179] In the Markov chain model, let the power distribution on day d be p. d =[p d (1),…,p d (M)] T The state transition matrix is ​​A = (p ij ) M×M Then, for any date d and day interval Δd, the probability distributions of the states at times d and d+Δd satisfy the following equation:

[0180] p d+Δd =A Δd p d .

[0181] Since wind power output varies continuously within the installed capacity range, meaning all states are interconnected, the aforementioned Markov chain wind power sequence model satisfies irreducibility, and all states are recurrent states. Furthermore, given the finite number of states in the model, the aforementioned Markov model has one and only one stationary distribution π = [π1, π2, ..., π]. M ] T The condition Aπ = π is satisfied, and for any initial state distribution p0, when t → ∞, A t p0 = π.

[0182] Suppose that the current state probability distribution has reached the steady-state distribution π of the state transition matrix A, but the desired data distribution is p. Then, we construct A' using the following iterative format, such that p is the steady-state distribution of A', and π smoothly transitions to p through A':

[0183] A (k+1) =A (k) -(A (k) -I)pp T

[0184] The above iterative process converges and satisfies linear convergence. However, in actual iterations, some elements may be less than 0, therefore, it is necessary to restrict A. (k+1) ∈[0,1] N×N The matrix is ​​normalized column-wise to ensure the non-negativity and regularity of the discrete probability distribution.

[0185] 2) Daily power generation fluctuation I2

[0186] Take I 2,1 For example, for a given threshold ξ2, the probability of the index crossing the line ∈2 is:

[0187]

[0188] Where, p d (j,i)=p ji p d-1 (i) is the state of sequence d-1 day S i And the state on day d is S j The probability of.

[0189] When the Markov chain reaches a stationary distribution, this probability becomes time-invariant, and d can be omitted. Let the expectation be p(I). 2,1 ≥ξ 2,1 The target probability is ζ. 2,1 Then the joint probability distribution p is corrected by the following formula. t (j,i):

[0190]

[0191] And through p ji =p d (j,i) / p d-1 (i) The state transition matrix is ​​obtained by restoring the original state.

[0192] 3) Duration of low output I3

[0193] From the basic properties of the outer layer's daily average power Markov chain model, we know that: The I3 index of the generated data follows a discrete exponential distribution and satisfies:

[0194]

[0195] Where p represents probability. X is the set of states with power below ∈3. d denoted as the state of the wind power sequence on day d, and ∈3 is a non-negative fluctuation threshold constant set by the user.

[0196] Therefore, by adjusting the steady-state distribution of the outer model, index I3 can be controlled, and the specific correction method is the same as that for index I1.

[0197] 4) Intraday reverse peak intensity I4

[0198] Since the distribution transformation is a bijective transformation and is an order-preserving mapping, that is... Therefore, use w' t Equivalent replacement of w tThe calculated index I'4 is used to replace I4 in the analysis, and is denoted as l'. t =l t -El t Then it can be written as:

[0199]

[0200] Where C is a certain constant.

[0201] Taking hourly wind power data as an example, in the inner autoregressive model, the power simulation data w' for a certain day after distribution transformation is... t The t = 1, ..., 24 follows a multivariate normal distribution N(μ, Σ), where μ = 0. For a multivariate normal distribution, it is easy to see that:

[0202]

[0203] Therefore, theoretically, the expected daily correlation coefficient between the wind power curve generated by the MC-AR model and the load is always 0. Thus, it is necessary to adjust the simulated mean μ = μ while keeping the daily average power constant. + , where μ + (0) Let it be Ew', and iterate according to the following formula:

[0204]

[0205] Where, ρ lw' The Pearson correlation coefficient between the set or actual load l and w'.

[0206] Finally, adjust μ + Mean is 0: μ + =μ + (k) -1 T μ + (k) / 24. After mean adjustment, to ensure a smooth connection between the curves of the two days, it is also necessary to... Figure 2 The simulation process of the MC-AR model in the middle is modified. Let u and v be μ, respectively. + The first and last elements only need to be... Figure 2 The initial simulated value of the intraday fluctuation power curve of the inner AR model in box ④ is corrected to w. d t +w d -w d+1 Simply add +vu.

[0207] 5) Daily peak and off-peak power generation I5

[0208] Since the distribution transformation is an order-preserving mapping, w' t Replace w tThe calculated index I'5 shows a strict positive correlation with I5, which can be used for analysis. Using I'... 5,1 For example, the marginal distribution of w' v = [w',…], t∈W v It still follows a multivariate Gaussian distribution, therefore I' 5,1 ~N(1 T μ v ,1 T Σ v 1), μ v For W v μ within the time period + , Σ v W of power during the time period v Let the expected value be EI' 5,1 The target value is ξ 5,1 The following formula can be used to correct this:

[0209]

[0210] 6) Intraday peak and trough power output fluctuations I6

[0211] Since the distribution transformation is an order-preserving mapping, w' t With w t The volatility is positively correlated, therefore w' can be used t Replace w t The calculated index I'6 is then used for further calculations. Using I'... 6,1 For example, with wind power generation data on an hourly timescale, the inner autoregressive model is written as:

[0212]

[0213] Where, ε t =[ε t ,ε t-1 ,…] T ~N(0,Σ ε ), Σ ε =σ ε 2 I;w t+1 =[w t+1 ,w t ,…] T ~N(μ) + ,Σ); Φ is the square matrix of autoregressive coefficients:

[0214]

[0215] Then at this time Φ=(σ i,j ) 24×24 For (ΦΣ) ε Φ) -1The matrix is ​​composed of the first 24 rows and columns. This is due to the coefficients of each row in the autoregressive coefficient matrix Φ and the standard deviation σ of the noise term in the autoregressive model at each time step. εt They are all the same, therefore:

[0216] Σ v =Σ p1 =Σ p2

[0217] For a multivariate normal distribution, with I' 6,1 For example, the determinant of a matrix can be used to describe the degree of dispersion of the population. 6,1 =|Σ v | is called the generalized variance. Let the expectation be I'. 6,1 The target value is ξ 6,1 Then the variance matrix Σ of the power curve can be adjusted. + ΣΣ T + , Σ + =diag(σ +24 ,σ +23 ,…,σ +1 The following formula can be used to correct this:

[0218]

[0219]

[0220] Where, v = Card(W) v );σ (0) +(t) The initial value can be chosen as Var(w' t ).

[0221] In another embodiment of the present invention, an index-driven medium- and long-term wind power curve closed-loop feedback generation system is provided. This system can be used to implement the above-mentioned index-driven medium- and long-term wind power curve closed-loop feedback generation method. Specifically, the index-driven medium- and long-term wind power curve closed-loop feedback generation system includes a parameter module, an index module, and an output module.

[0222] Among them, the parameter module constructs a Markov chain-autoregressive wind power curve generation model to determine the basic parameters and hyperparameters;

[0223] The indicator module constructs an indicator-driven closed-loop feedback generation model for medium- and long-term wind power curves. Based on the Markov chain-autoregressive two-layer wind power curve generation model obtained from the parameter module, it calculates the specified indicators for the power curve data generated by the Markov chain-autoregressive two-layer wind power curve generation model.

[0224] The output module generates curve data statistical characteristic indicators based on the Markov chain-autoregressive two-layer wind power curve generation model obtained from the indicator module, and performs feedback correction on the model parameters. Based on the basic parameters and hyperparameters obtained from the parameter module, the corrected model parameters are obtained again to obtain medium- and long-term wind power generation curve data that meet the requirements.

[0225] In another embodiment of the present invention, a terminal device is provided, comprising a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve a corresponding method flow or corresponding function. The processor described in this embodiment can be used in the operation of an index-driven medium- and long-term wind power curve closed-loop feedback generation method, including:

[0226] A Markov chain-autoregressive wind power curve generation model is constructed to determine the basic parameters and hyperparameters. An index-driven closed-loop feedback generation model for medium- and long-term wind power curves is also constructed. Based on the Markov chain-autoregressive two-layer wind power curve generation model, specified indices for the power curve data generated by the model are calculated. According to the statistical characteristic indices of the curve data generated by the Markov chain-autoregressive two-layer wind power curve generation model, the model parameters are corrected through feedback. Based on the obtained basic and hyperparameters, the corrected model parameters are re-obtained to acquire medium- and long-term wind power curve data that meets the requirements.

[0227] In another embodiment of the present invention, a storage medium is also provided, specifically a computer-readable storage medium (memory). This computer-readable storage medium is a memory device in a terminal device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device.

[0228] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the indicator-driven closed-loop feedback generation method for medium- and long-term wind power curves in the above embodiments; one or more instructions in the computer-readable storage medium are loaded and executed by the processor to perform the following steps:

[0229] A Markov chain-autoregressive wind power curve generation model is constructed to determine the basic parameters and hyperparameters. An index-driven closed-loop feedback generation model for medium- and long-term wind power curves is also constructed. Based on the Markov chain-autoregressive two-layer wind power curve generation model, specified indices for the power curve data generated by the model are calculated. According to the statistical characteristic indices of the curve data generated by the Markov chain-autoregressive two-layer wind power curve generation model, the model parameters are corrected through feedback. Based on the obtained basic and hyperparameters, the corrected model parameters are re-obtained to acquire medium- and long-term wind power curve data that meets the requirements.

[0230] Please see Figure 8 The terminal device is a computer device. In this embodiment, the computer device 60 includes a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When executed by the processor 61, the computer program 63 implements the fluid composition calculation method in the reservoir stimulation wellbore of this embodiment. To avoid repetition, these details are not elaborated here. Alternatively, when executed by the processor 61, the computer program 63 implements the functions of each model / unit in the fluid composition calculation system in the reservoir stimulation wellbore of this embodiment. To avoid repetition, these details are not elaborated here.

[0231] Computer device 60 can be a desktop computer, laptop, handheld computer, cloud server, or other computing device. Computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will understand that... Figure 8 This is merely an example of computer device 60 and does not constitute a limitation on computer device 60. It may include more or fewer components than shown, or combine certain components, or different components. For example, computer device may also include input / output devices, network access devices, buses, etc.

[0232] The processor 61 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0233] The memory 62 can be an internal storage unit of the computer device 60, such as a hard disk or RAM of the computer device 60. The memory 62 can also be an external storage device of the computer device 60, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the computer device 60.

[0234] Furthermore, the memory 62 may include both internal storage units of the computer device 60 and external storage devices. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 can also be used to temporarily store data that has been output or will be output.

[0235] Please see Figure 9 The terminal device is a chip. In this embodiment, the chip 600 includes a processor 622, which may be one or more, and a memory 632 for storing computer programs executable by the processor 622. The computer program stored in the memory 632 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processor 622 may be configured to execute the computer program to perform the generalizable monocular absolute depth map estimation method described above.

[0236] Additionally, chip 600 may also include a power supply component 626 and a communication component 650. The power supply component 626 can be configured to perform power management of chip 600, and the communication component 650 can be configured to enable communication of chip 600, such as wired or wireless communication. Furthermore, chip 600 may also include an input / output interface 658. Chip 600 can operate on an operating system stored in memory 632.

[0237] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0238] Example

[0239] The test examples consist of annual wind power output data from a province in my country, as well as a modified NREL-118 node system and an open-source wind power dataset from the British power company Elexon. The NREL-118 system is shown below. Figure 6 As shown, since the original NREL-118 system had a small installed capacity of wind power, it had little impact on the system operation. Therefore, under the condition that the location of the NREL-118 system wind farm remained unchanged, the wind power installed capacity of the modified NREL-118 system was analyzed to be three times that of the original example.

[0240] To verify the effectiveness of the indicator-driven closed-loop feedback method for generating medium- and long-term wind power curves, a case study was conducted, selecting 2+6 statistical indicators, 2 different historical typical scenarios, and 1 typical application scenario of a medium- and long-term time series curve. The method was compared with the classic Markov chain model (MC), the non-homogeneous Markov chain model (NHMC), and the restricted moving average autoregressive model (LARIMA) to analyze the curve generation method.

[0241] The 2+6 statistical indicators are: autocorrelation function and probability distribution (2 indicators) plus monthly power generation, daily power generation fluctuation, duration of low output, intraday counter-peak intensity, intraday peak-valley power generation, and intraday peak-valley power output fluctuation (6 indicators).

[0242] Two typical historical scenarios are "a typical week of sustained low power output in summer" and "a typical week of day-night peak shaving characteristics in winter";

[0243] A typical application scenario for a medium- to long-term time series curve is the medium- to long-term power balance analysis of the NREL-118 system based on Elexon wind power curve data from the 8760 time series production simulation.

[0244] Please see Figure 2 The distribution of data generated by the MC model is completely consistent with historical data, followed by the MCAR model. The LARIMA and NHMC models show the largest deviations. Among the autocorrelation functions, the MC model shows the largest deviation from historical autocorrelation functions. The NHMC model accurately simulates the autocorrelation function for the first two days by introducing a non-homogeneous state transition matrix, but the autocorrelation function of the generated curve begins to deviate from the historical autocorrelation function after two days. The LARIMA model can basically follow the trend of historical autocorrelation function changes, maintaining an autocorrelation value above 0.5 on the fifth day. The MC-AR model combines the advantages of both models well, modeling daily periodicity while maintaining an autocorrelation value with minimal deviation from historical data on the fifth day. Meanwhile, according to... Figure 3 Among the six indicators, the MC-AR model showed certain advantages in five indicators: monthly power generation I1, daily power generation fluctuation I2, duration of low output I3, evening peak power generation I5, and fluctuation I6.

[0245] Please see Figure 4 and Figure 5 The generated wind power curve data and the target typical historical scenario data are highly consistent in terms of time series characteristics and numerical probability distribution.

[0246] Please see Figure 7 The graph shows the relative deviations of production simulation results across different time series; the smaller the deviation, the smaller the area covered. Comparing the various lines in the graph, it can be seen that the blue line of the proposed MC-AR method exhibits the smallest overall relative deviation in power generation from thermal power, hydropower, photovoltaic power, run-of-river hydropower units, and wind power curtailment.

[0247] The above examples demonstrate that the present invention can accurately simulate medium- and long-term wind power curves.

[0248] In summary, this invention provides an index-driven closed-loop feedback method for generating medium- and long-term wind power curves. From the perspective of grid operators, it obtains more accurate statistical characteristics of medium- and long-term wind power curve simulation data, which can effectively improve the accuracy of medium- and long-term system analysis results such as medium- and long-term power balance calculations. Furthermore, it can utilize existing data to evaluate and analyze wind power generation systems under construction with similar environmental and resource conditions, and rationally plan system power source construction. From the perspective of power plant operators, by obtaining more accurate statistical characteristics of wind power curve simulation data, it improves the simulation accuracy for participating in the medium- and long-term electricity market, thereby enabling the formulation of more reasonable bidding strategies and enhancing competitiveness.

[0249] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0250] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0251] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0252] In the embodiments provided by this invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0253] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0254] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0255] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0256] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations 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, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0257] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0258] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0259] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A closed-loop feedback method for generating medium- and long-term wind power curves driven by indicators, characterized in that, Includes the following steps: S1. Construct a Markov chain-autoregressive two-layer wind power curve generation model and determine the basic parameters and hyperparameters. The Markov chain-autoregressive two-layer wind power curve generation model takes static data of the wind power generation system and medium- and long-term wind power curve data as inputs and outputs medium- and long-term wind power curve data that meet statistical characteristics. Based on the static data of the wind power generation system to be modeled and the medium- and long-term wind power curve data of the wind power generation system, the Markov chain-autoregressive two-layer wind power curve generation model is constructed. The Markov chain-autoregressive two-layer wind power curve generation model includes a daily average power model based on Markov chains and an intraday fluctuation power curve model based on autoregression. The basic parameters and hyperparameters are estimated based on the collected static data of the wind power generation system to be modeled and the medium- and long-term wind power curve data of the wind power generation system. The Markov chain-based daily average power model is used to characterize the probability distribution of daily wind power average power and the state transition characteristics between adjacent days. The parameters include the state transition matrix, the initial state probability, and the state number hyperparameter. The autoregressive intraday fluctuating power curve model is used to characterize the temporal correlation of intraday hourly wind power, and the parameters include regression coefficients, noise variance, and model order hyperparameters. The statistical characteristic indicators of the wind power curve include monthly power generation, daily power generation fluctuation, duration of low output, intraday anti-peak shaving intensity, intraday peak-valley power generation, and intraday peak-valley output fluctuation. S2. Based on the Markov chain-autoregressive two-layer wind power curve generation model obtained in step S1, calculate the statistical characteristic index of the wind power curve generated by the Markov chain-autoregressive two-layer wind power curve generation model. S3. Based on the Markov chain-autoregressive two-layer wind power curve generation model obtained in step S2, generate statistical characteristic indicators of wind power curves, and perform feedback correction on the basic parameters and hyperparameters. On the basic parameters and hyperparameters obtained in step S1, re-obtain the corrected model parameters to obtain medium- and long-term wind power generation curve data that meet the requirements.

2. The index-driven closed-loop feedback method for generating medium- and long-term wind power curves according to claim 1, characterized in that, The static data of the wind power generation system to be modeled includes: the installed capacity information of the wind power generation system to be modeled, the current installed capacity or the installed capacity and change data of the wind power generation system within the modeling period; The long-term wind power curve data of the wind power generation system includes: the power generation data of the wind power generation system to be modeled; based on the modeling period, the data is in units of years or months, with a time resolution of fifteen minutes or one hour.

3. The index-driven closed-loop feedback method for generating medium- and long-term wind power curves according to claim 1, characterized in that, The daily average power model based on Markov chains is as follows: in, , p i For the first state X 0 is S i The probability of; p ji The status of the previous day was S i Under these conditions, the status of the following day is S j The probability of; M The number of states in the Markov model; X d The first discretization of wind power output d Daily status; The autoregressive intraday volatility power curve model is as follows: in, φ i These are the coefficients of the autoregressive model; r The order of the autoregressive model; ε t The noise in the autoregressive model follows a normal distribution N(0, σ ε 2 ), w t The daily wind power output minus the daily average power output. t Relative power at any given time; The intraday relative correction power of wind power, after Z-Score standardization, approximately follows a normal distribution.

4. The index-driven closed-loop feedback method for generating medium- and long-term wind power curves according to claim 1, characterized in that, In step S2, the specific statistical characteristic indicators of the wind power curve are as follows: Monthly power generation I 1: Measure the reliable power generation capacity of a wind power system on a monthly timescale; Daily power generation fluctuation I 2: To measure the relative change in the average power output of wind power systems over two consecutive days; Low output duration I 3: To characterize the degree of sustained low daily average power generation capacity of a wind power system, based on a specified threshold. ε ; Intraday counter-peak intensity I 4: The characteristic of wind power generation being lower during the day and higher at night is opposite to the peak-shaving demand of the power grid. This is essentially a problem of the linear correlation between wind power generation and load electricity consumption. Electricity generation during peak and off-peak hours of the day I 5: Reflects the ability of the wind power system to support the system load during peak and off-peak periods relative to the average of the day; Fluctuations in output during peak and trough periods throughout the day I 6: Reflects the ability of the wind power generation system to stably support the system load during peak and valley periods; measures the variance of the relative power during peak and valley periods through the metering system.

5. The index-driven closed-loop feedback method for generating medium- and long-term wind power curves according to claim 4, characterized in that, Monthly power generation I 1: Among them, W m For the first m Collection of monthly wind power curves For the first m The number of days in a month m =1,2,…,12, For the first No. of the month Daily average wind power output; Daily power generation fluctuation I 2: in, I 2,1 and I 2,2 The value of the indicator is calculated relative to the previous or next day; Low output duration I 3: in, d 1< d 2 represents any two consecutive days. A low output threshold constant between 0 and 1 set by the user; Intraday counter-peak intensity I 4: in, This represents the covariance between the intraday wind power curve and the load power curve. Electricity generation during peak and off-peak hours of the day I 5: Among them, W v It is the valley and lotus period; W p1 The peak load period in the morning; W p2 This is the peak load period in the evening; Fluctuations in output during peak and trough periods throughout the day I 6: in, It is the time of Guhe (a type of lotus). This is the peak lotus season in the morning. This is the peak load period in the evening.

6. The index-driven closed-loop feedback method for generating medium- and long-term wind power curves according to claim 1, characterized in that, In step S3, the specific steps for correcting the model parameters are as follows: Monthly power generation I 1: In the Markov chain model, let the th chain be... d Daily power distribution is p d =[ p d (1),…, p d ( M )] T The state transition matrix is A =( p ij ) M×M Then for any date d Interval with day Δ d Markov models have one and only one steady-state distribution. π =[ π 1, π 2,…, π M ] T ,satisfy Aπ = π And for any initial state distribution p 0, as t→∞, A t p 0= π Suppose that the probability distribution of the current state has reached the state transition matrix. A steady-state distribution π However, the expected distribution of the data is p Constructed through iterative format A' ,make p for A' steady-state distribution π through A' Smooth transition to p ; The iterative process converges and satisfies linear convergence, with constraints A (k+1) ∈[0,1] N×N And normalize the matrix column-wise to ensure the non-negativity and regularity of the discrete probability distribution; Daily power generation fluctuation I 2: When the Markov chain reaches a stationary distribution, this probability will be a time-invariant, omitted. d Let the expectation be... p ( I 2,1 ≥ ξ 2,1 The target probability is ζ 2,1 Corrected joint probability distribution p t ( j , i ), and through p ji = p d ( j , i ) / p d-1 ( i The state transition matrix is ​​obtained by reconstructing the state. Low output duration I 3: The basic properties of the outer layer's daily average power Markov chain model are obtained from 3∈(0,1), generating data I The three indicators follow a discrete exponential distribution. Adjusting the steady-state distribution of the outer model helps to account for low-output durations. I 3. Implement control measures; Intraday counter-peak intensity I 4: Since the distribution transformation is a bijective transformation and is an order-preserving mapping, that is... w 1> w 2 w' 1> w' 2, therefore available w' t Equivalent substitution w t Calculated indicators I' 4 Replacement I 4. Conduct analysis and record. l' t = l t -E l t , In the inner autoregressive model, the power simulation data for a given day after distribution transformation... w' t , t =1,…,24 follow a multivariate normal distribution N( μ , Σ ), While keeping the daily average power constant, adjust the simulated mean. μ = μ + ,Adjustment μ + The mean is 0: μ + = μ + (k) -1 T μ + (k) / 24, after mean adjustment, record u , v They are respectively μ + The first and last two elements are used to correct the initial values ​​of the intraday fluctuation power curve simulation of the inner AR model. w d t + w d - w d+1 + v - u ; Electricity generation during peak and off-peak hours of the day I 5: w' marginal distribution w' v ,t∈W v It still follows a multivariate Gaussian distribution, therefore I' 5,1 ~N(1 T μ v ,1 T Σ v 1), μ v For W v During the time period μ + , Σ v W of power during the time period v Let the expectation be E I' 5,1 The target value is ξ 5,1 Make corrections; Fluctuations in output during peak and trough periods throughout the day I 6: use w' t replace w t Calculated indicators I' 6. Perform calculations to describe the dispersion of the population using the determinant of a matrix for a multivariate normal distribution. I' 6,1 =| Σ v |, set expectation I' 6,1 The target value is ξ 6,1 Adjust the variance matrix of the power curve Σ=Σ + ΣΣ T + , Σ + =diag( σ +24 , σ +23 , … , σ +1 ), and make corrections.

7. The index-driven closed-loop feedback method for generating medium- and long-term wind power curves according to claim 6, characterized in that, Monthly power generation I In 1, d and d +Δ d The probability distributions of the states at time t satisfy the following relationship: p d+Δd =A Δd p d The iteration format is as follows: ； Daily power generation fluctuation I In section 2, the modified joint probability distribution p t ( j , i ): in, As an indicator I 2. Compared to the calculated value of the previous day, j Representing state S j , i Representing state S i , A non-negative fluctuation threshold constant set by the user. for p ( I 2,1 ≥ ξ 2,1 The target probability; Low output duration I 3 in, I The three indicators follow a discrete exponential distribution and satisfy the following conditions: in, p Represents probability. For power lower than The set of states, X d The wind power sequence is in state on day d. A non-negative fluctuation threshold constant set by the user; use w' t Equivalent substitution w t Calculated indicators I' 4 replaces intraday reverse peak intensity I 4. Specifically: in, C A constant determined by the length of a given sequence. For the average value operator, This is the relative power vector after deducting the mean from the intraday load. This is the intraday relative power vector correction for wind power. Electricity generation during peak and off-peak hours of the day I In step 5, the following formula can be used for correction: in, The expected value of the sub-indices during the valley-load period, E I' 5,1 The target value, For inner autoregressive models t The mean parameter at time, For the first k The mean parameters of the inner autoregressive model during the valley loading period after rounds of iterations. It is the time of Guhe (a type of lotus). For a specific moment; Fluctuations in power output during peak and off-peak hours throughout the day I In step 6, the following formula is used for correction: in, For the first k The variance parameters of the inner autoregressive model during the valley loading period after rounds of iterations. For the first k After rounds of iterations, the inner autoregressive model is at t The standard deviation parameter at time, Sub-indicators for the valley-heat period I' 6,1 The target value.

8. A closed-loop feedback generation system for medium- and long-term wind power curves driven by indicators, characterized in that, include: The parameter module constructs a Markov chain-autoregressive two-layer wind power curve generation model to determine the basic parameters and hyperparameters. The Markov chain-autoregressive two-layer wind power curve generation model takes static data of the wind power generation system and medium- and long-term wind power curve data as inputs and outputs medium- and long-term wind power curve data that meet statistical characteristics. Based on the static data of the wind power generation system to be modeled and the medium- and long-term wind power curve data of the wind power generation system, the Markov chain-autoregressive two-layer wind power curve generation model is constructed. The Markov chain-autoregressive two-layer wind power curve generation model includes a daily average power model based on Markov chains and an intraday fluctuation power curve model based on autoregression. The basic parameters and hyperparameters are estimated based on the collected static data of the wind power generation system to be modeled and the medium- and long-term wind power curve data of the wind power generation system. The Markov chain-based daily average power model is used to characterize the probability distribution of daily wind power average power and the state transition characteristics between adjacent days. The parameters include the state transition matrix, the initial state probability, and the state number hyperparameter. The autoregressive intraday fluctuating power curve model is used to characterize the temporal correlation of intraday hourly wind power, and the parameters include regression coefficients, noise variance, and model order hyperparameters. The statistical characteristic indicators of the wind power curve include monthly power generation, daily power generation fluctuation, duration of low output, intraday anti-peak shaving intensity, intraday peak-valley power generation, and intraday peak-valley output fluctuation. The indicator module calculates the statistical characteristic indicators of the wind power curve generated by the Markov chain-autoregressive two-layer wind power curve generation model based on the Markov chain-autoregressive two-layer wind power curve generation model obtained from the parameter module. The output module generates statistical characteristic indicators of wind power curves based on the Markov chain-autoregressive two-layer wind power curve generation model obtained from the indicator module. It then performs feedback correction on the basic parameters and hyperparameters. Based on the basic parameters and hyperparameters obtained from the parameter module, it re-obtains the corrected model parameters to acquire medium- and long-term wind power generation curve data that meet the requirements.

Images