CN112862166A - Wind power interval prediction combination method based on signal decomposition - Google Patents
Wind power interval prediction combination method based on signal decomposition Download PDFInfo
- Publication number
- CN112862166A CN112862166A CN202110112003.4A CN202110112003A CN112862166A CN 112862166 A CN112862166 A CN 112862166A CN 202110112003 A CN202110112003 A CN 202110112003A CN 112862166 A CN112862166 A CN 112862166A
- Authority
- CN
- China
- Prior art keywords
- prediction
- wind power
- signal
- sub
- interval
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 33
- 230000002194 synthesizing effect Effects 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 230000007547 defect Effects 0.000 abstract description 7
- 238000010248 power generation Methods 0.000 abstract description 4
- 230000015572 biosynthetic process Effects 0.000 abstract description 2
- 238000003786 synthesis reaction Methods 0.000 abstract 1
- 230000006870 function Effects 0.000 description 16
- 239000011159 matrix material Substances 0.000 description 13
- 238000011156 evaluation Methods 0.000 description 8
- 238000013507 mapping Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 239000000126 substance Substances 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000000053 physical method Methods 0.000 description 2
- 240000004760 Pimpinella anisum Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000010977 unit operation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Tourism & Hospitality (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Probability & Statistics with Applications (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a wind power interval prediction combination method based on signal decomposition, and belongs to the technical field of wind power generation output prediction. The method realizes interval prediction combination through three steps of data preprocessing, prediction interval and subinterval synthesis. The prediction model firstly carries out naive Bayes point prediction on the subcomponent signals, establishes a kernel limit learning machine and a kernel density estimation combination model weighted by an entropy weight method, and inputs point prediction errors into the combination model to obtain a prediction subinterval. And finally, synthesizing all subintervals to obtain a final prediction result. According to the method, the non-stationarity of the wind power signal is considered, and signal decomposition is carried out before prediction. The influence of the unstable signal on the prediction accuracy is reduced. Aiming at the inherent defects of a single model, the invention considers the combination of different models to form complementation to a certain extent and improve the prediction precision.
Description
Technical Field
The invention belongs to the technical field of wind power generation output prediction, and particularly relates to a wind power interval prediction combination method based on signal decomposition.
Background
In order to cope with the gradual depletion of fossil energy and the problem of environmental pollution caused by the combustion of fossil energy, new energy industries are vigorously developed in countries in the world. Wind power generation has been vigorously developed in various countries in the world due to its advantages of low cost, cleanness and reproducibility. However, due to the uncertainty and the inverse peak regulation characteristic of wind power generation, scheduling is difficult, and serious impact is brought to a power grid when large-scale wind power is connected into a power system. Therefore, efficient consumption of wind power with large-scale network access is a preoccupation in power development. And aiming at the serious asymmetry of the power production and consumption layout in China, the wind power base is mainly concentrated in the three-north area, the power consumption is mainly concentrated in the southeast coastal area, and the local consumption is difficult, so that the serious wind abandon phenomenon caused by land is difficult to avoid. At present, one main approach for solving the problem is to improve the wind power prediction precision, help the power grid to carry out reasonable economic dispatching, arrange unit operation and equipment maintenance, and realize safe, reliable and economic network access of large-scale wind power.
The traditional wind power prediction mainly adopts a deterministic prediction method, namely, a determined value which may appear in future wind power is provided. Deterministic prediction methods can be largely classified into statistical learning methods and physical methods according to the source of data used. In the existing research, no matter a statistical method or a physical method, due to the uncertainty of wind resources and the inherent defects of a prediction model, deterministic prediction errors cannot be avoided, and the result of the deterministic prediction errors cannot quantitatively describe the uncertainty of wind power. Therefore, researchers at home and abroad seek out uncertainty prediction to quantitatively reflect the uncertainty of the wind power.
The wind power probability prediction is a wind power prediction type which is used for establishing a prediction model aiming at the uncertainty of wind power according to meteorological data, historical wind power actual measurement data and prediction data and providing a fluctuation region or distribution (density) function of the wind power at a future moment. There are two main forms: interval prediction and density prediction.
Meanwhile, the research of the combination of the prediction models obtains better prediction results. The single prediction model has inherent advantages and defects, and whether the prediction method is suitable or not depends on the prediction scale, the wind power plant geographic factors and the prediction purpose. And a plurality of methods are adopted for combined prediction, so that a better prediction result can be expected.
Wind power signals are typically non-stationary signals. The wind power at different moments contains significantly different frequency components and carried energy, and a unified model is only used for predicting the non-stationary signal characteristic of the wind power, which obviously causes prediction errors. Therefore, before the wind power is predicted, signal decomposition is carried out firstly, and therefore the method has important significance for improving the prediction effect. The original time series signal is decomposed into a plurality of components by methods such as Wavelet Transform (WT), Empirical Mode Decomposition (EMD), Variational Mode Decomposition (VMD) and the like, each component is predicted, and then recombined to be used as a final prediction result. Researches show that the wind power prediction precision is obviously improved after signal decomposition.
At present, the improvement of the wind power prediction precision meets a new bottleneck period. On one hand, errors are inevitably generated due to the algorithm defects of various prediction models; on the other hand, when the wind power signal is used as a non-stationary signal and a single model is used for prediction, the non-stationary characteristic of the signal obviously reduces the prediction precision.
Aiming at the defects, the invention provides a wind power interval prediction combination model based on a signal decomposition method. Firstly, the wind power signal is decomposed by adopting Empirical Mode Decomposition (EMD) and then is independently modeled, so that the influence of the non-stationarity of the wind power signal on the prediction precision is reduced; secondly, the probability distribution of the wind power prediction error estimated by a parametric method and a nonparametric method has advantages and disadvantages, and the interval prediction combination model is adopted to combine the two methods, so that complementation can be formed to a certain extent, and the inherent defect of a single prediction model is overcome.
Disclosure of Invention
The invention aims to provide a wind power interval prediction combination method based on signal decomposition, which reduces the generated error and improves the non-stationarity prediction precision of a signal.
In order to achieve the purpose, the invention adopts the following technical scheme:
a wind power interval prediction combination method based on signal decomposition comprises the following steps:
1) performing data preprocessing on historical wind power data, and decomposing the historical data into a plurality of sub-signal components by adopting an EMD signal decomposition method;
2) inputting each sub-signal component into the combined model of the interval prediction algorithm independently for prediction to obtain a prediction sub-interval of each sub-signal component;
3) and synthesizing all the subintervals to obtain a final wind power interval prediction result.
Further, a wind power certainty prediction model based on a naive Bayes algorithm is established in the step 2).
Further, the sub-signal components are predicted separately in the step 1).
Further, an interval prediction combination model of the kernel density estimation and kernel limit learning machine based on the entropy weight method is established in the step 2).
Further, the deterministic prediction error of each sub-signal obtained in the step 2) is input into a combination model, so as to obtain a prediction sub-interval of each sub-signal component.
Further, the naive Bayes algorithm prediction error comprises a kernel density estimation interval prediction and a kernel limit learning machine interval prediction.
The invention has the beneficial effects that:
1) according to the method, wind power signal decomposition and wind power interval prediction of different prediction model combinations are considered at the same time. Points to be protected: and predicting the combined model of the wind power interval based on EMD signal decomposition. EMD decomposition is carried out on the wind power original signal, and each sub-component is independently predicted.
2) The prediction model firstly carries out naive Bayes point prediction on the subcomponent signals, establishes a kernel limit learning machine and a kernel density estimation combination model weighted by an entropy weight method, and inputs point prediction errors into the combination model to obtain a prediction subinterval. And finally, synthesizing all subintervals to obtain a final prediction result.
3) According to the method, the non-stationarity of the wind power signal is considered, and signal decomposition is carried out before prediction. The influence of the unstable signal on the prediction accuracy is reduced.
4) Aiming at the inherent defects of a single model, the invention considers the combination of different models to form complementation to a certain extent and improve the prediction precision.
Drawings
FIG. 1 is a wind power prediction model diagram based on EMD signal decomposition.
FIG. 2 is a schematic diagram of a wind power interval prediction combination model based on an entropy weight method.
Fig. 3 is a flow chart of EMD signal decomposition.
FIG. 4 is a schematic diagram of a wind power interval prediction model based on a kernel limit learning machine.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
A wind power interval prediction combination method based on signal decomposition comprises the following steps:
1) performing data preprocessing on historical wind power data, and decomposing the historical data into a plurality of sub-signal components by adopting an EMD signal decomposition method; the sub-signal components are predicted separately.
2) Inputting each sub-signal component into the combined model of the interval prediction algorithm independently for prediction to obtain a prediction sub-interval of each sub-signal component; and establishing a wind power certainty prediction model based on a naive Bayes algorithm. And establishing an interval prediction combination model of the kernel density estimation and kernel limit learning machine based on an entropy weight method. And inputting the obtained deterministic prediction error of each sub-signal into a combined model to obtain a prediction sub-interval of each sub-signal component. The naive Bayes algorithm prediction error comprises kernel density estimation interval prediction and kernel limit learning machine interval prediction.
3) And synthesizing all the subintervals to obtain a final wind power interval prediction result.
And performing point prediction (deterministic prediction) on the sub-signal component obtained by the EMD signal decomposition method, inputting point prediction errors into different interval prediction algorithm models respectively, and weighting a plurality of prediction results by using an entropy weight method to obtain a prediction sub-interval of the sub-signal component. The design algorithm in the interval prediction combination model is as follows:
empirical mode decomposition of EMD
Empirical Mode Decomposition (EMD) is a processing method proposed by the scholars Huang in 1988 for complex nonlinear non-stationary signals. The EMD can be used for decomposing a complex original signal into a series of intrinsic mode components (IMF) and a residual component, and the fluctuation characteristics of the original signal under different scales and the variation trend of the whole time sequence are respectively represented, so that the real physical significance and the local features of the signal can be extracted. These IMF signals have two characteristics:
the number of extreme points (maxima and minima) is equal to or at most one different from the number of zero crossing points.
The average of the upper envelope composed of local maxima and the lower envelope composed of local minima is zero. Assuming that the analyzed non-stationary time signal is X (t), the specific algorithm flow of EMD is shown in the figure according to the meaning of IMF.
In the figure, r (t) is the residual component, and C (t) is the finite number of IMFs obtained by decomposition. The threshold for stopping iteration is generally the standard deviation SdSuch as:
typical values are between 0.2 and 0.3.
Through the above process, the non-stationary time signal X (t) will eventually be decomposed into n IMFs (labeled C)1,C2,…Cn) And a residual component (denoted as r)nThe residual component may be an average trend or constant), i.e.
2. Naive Bayes point prediction
Bayesian theory does not start from linear or nonlinear mapping relations among historical data, and infers future states of the data by counting probability information of the historical data. Naive Bayes Classifier (NBC) is one of the most widely used models in Bayes classifiers. The naive bayes classifier model structure is shown in figure 1.
Assume a set of variables U ═ { a, C }, where a ═ a }, where1,A2,…,AnContains n condition attributes, C ═ C1,C2…CmContains m types of labels. Naive Bayes classifier modelType hypothesis Condition Attribute Ai(i-1, 2, … n) are all child nodes of the class variable c, given a sample X-a1,a2,…anIs designated as Ci(1. ltoreq. i.ltoreq.m), and only if:
where P (X) is the unconditional probability (also known as the prior probability) of the sample X to be sorted, P (C)i| X) is given class C in the case of a given classiConditional probability (also called posterior probability)
If the probability of the data set is unpredictable, the following maximization P (C) can be used, assuming that the probability of each class is equali|X):
P(Ci)=P(Cj)(Ci,Cj∈C,i≠j)
Otherwise, maximize P (C)i)P(X|Ci). Since all categories are well-known, there are:
with a naive bayes classifier algorithm, the conditional attributes are independent of each other:
wherein the content of the first and second substances,Siis class C in the training sampleiAnd S is the total number of training samples. Thus, the naive bayes model formula expression is:
wherein, P (a)i|Ci),P(a2|Ci),L,P(an|Ci) The probability of (c) can be estimated by training the samples. According to this formula, the sample C to be classified belongs to the class Ci。
3. Nuclear limit learning machine
ELM is a single-hidden-layer forward neural network proposed by scholars in 2006, and has excellent nonlinear fitting performance and fast training speed. Let Xi=[Xi1,Xi2,…,XiN]∈RnTo input data, ti=[ti1,ti2,…tim]∈RmFor a target output value, the ELM algorithm matrix expression is:
in the formula: y is formed by RnIs the output value of the network; n is the number of samples; omegaiIs the weight connecting the ith hidden node and the input node; biA bias for the ith hidden layer of the network; beta is an output weight between the hidden layer node and the output layer; gi(ωii×Xj+bi) Is the activation function of the first hidden node.
When the activation function g (-) is able to zero-error approximate arbitrary N samples, i.e.Then (c) is performed. Equation (5) represents a matrix form of N equations, and has
Wherein, the hidden layer output matrix of the extreme learning machine is
In the formula: k is the number of hidden layer neurons; y (X) is the ELM network output.
The beta optimum can be obtained by Moore-Penrose generalized inverse matrix solution:
H+=H+(HH+)-1
in the formula: t is the sample target value vector, i.e. the desired output vector.
An improved algorithm of KeLM algorithm type ELM of kernel limit learning machine is characterized by introducing kernel function into original ELM, adding coefficient I/C into matrix HHTOn the main diagonal of (A), the ELM is more generalized and stable.
The method comprises the following steps:
constructing a kernel function using Mercer conditions:
ΩELM=HHT:Ωi,j=h(Xi)×h(Xj)=K(Xi,Xj)
kernel matrix omega instead of HHTAll input samples are mapped from the dimensional input controls to the high dimensional hidden layer feature space. h (X) is a hidden node output function; there are various types of kernel functions K (μ, ν), and the RBF kernel is selected, namely:
K(μ,υ)=exp[-(μ-υ2)/σ]
and after the setting of the kernel parameters mu, upsilon and sigma is finished, the mapping value of the kernel matrix omega is a fixed value.
Introduction of parameters I/C into diagonal matrix HH using ridge regression principleTSolving a regularized least square solution:
β=HT(I/C+HHT)-1T
in the formula: i and C are the diagonal matrix and the regularization parameters, respectively.
The output of ELM can be obtained from the above formula as
Compared with ELM, KELM replaces random mapping in ELM with stable kernel mapping, so that the stable kernel generalization capability of the model is enhanced, and the KELM has the advantages of less adjusting parameters, high convergence rate and the like. This is due to the use of K (X, X)i) The kernel function in the form does not need to preset the number of hidden layer nodes and bias parameters.
According to the method, wind speed at a moment to be predicted is selected as input, output data U (X) and output data L (X) are respectively an upper limit and a lower limit of a wind power interval with prediction, and a wind power interval prediction model basic structure based on a KELM is established as shown in the figure.
4. Nuclear density estimation
Kernel Density Estimation (KDE) is a simulation of the true probability distribution using a smooth peaking function (Kernel function) to fit the observed data points. In the field of machine learning, KDE is an unsupervised learning method.
Let x1,x2,…,xnSetting the probability density function of n sample points of independent same distribution F as F, and estimating the kernel density as follows:
wherein, K (·) is not kernel function, h > 0 is a smooth parameter called bandwidth;to scale the kernel function.
In the process of implementing the kernel density estimation, since the implementation principle is to traverse each point of the output curve and compute the kernel density estimation, if n is large, resulting in too many points on the curve, each point needs to perform n accumulated addition operations, and most of the n accumulated addition operations are +0, thus resulting in redundant computation.
The solution is as follows: an index may be established, and then the index may be used to search for nearby points when calculating the kernel density estimate for a point, and then the kernel functions for these points may be accumulated to obtain the result. The above objective is achieved with a nearest neighbor search algorithm.
In the kernel density estimation process, the selection of bandwidth is the most critical, and the kernel function estimation results under different bandwidths can be very different. The bandwidth reflects the overall flatness of the KDE curve, i.e., the proportion of observed data points in the KDE curve formation process. The larger the bandwidth is, the smaller the proportion of the observed data points in the finally formed curve shape is, and the flatter the KDE overall curve is; the smaller the bandwidth, the greater the proportion of observed data points in the resulting curve shape, and the steeper the KDE overall curve.
The choice of bandwidth depends largely on subjective judgments: if the real probability distribution curve is considered to be relatively flat, selecting a larger bandwidth; conversely, if the true probability distribution curve is considered steeper, a smaller bandwidth is selected. However, the subjective judgment cannot be used as an objective basis to study the subsequent results, and the present subject uses a minimum Error method based on Mean Interpolated Squared Error (MISE) to measure the quality of the bandwidth h. MISE is defined as follows:
wherein the content of the first and second substances,is an estimate of bandwidth, anIs estimated for the evaluation as mean square errorA point-by-point criterion of quality; the mise (h) can be seen as the accumulation of local mean square error at each point x, i.e. a global criterion.
wherein the content of the first and second substances,called progressive mean square integral error, in order to minimize the mise (x), it translates into a pole-solving problem, namely:
the optimal bandwidth h can be obtainedAMISEAs shown in the above formula. The above formula has h ═ O (n)-1/5) In this case, MISE ═ O (n)-4/5)。
5. Entropy weight method
Any single method predictive model has its own limitations due to different mechanisms and different information involved. The combined prediction model combines different prediction methods, comprehensively utilizes information provided by various prediction methods, and can improve the prediction accuracy of the system. On the basis of the basic wind power prediction information provided by the single prediction method, the entropy weight method-based wind power prediction combination model obtains the combination weight of each single prediction method through the information entropy theory, and the photovoltaic power can be predicted more reasonably and accurately. The method comprises the following steps:
the entropy weight method determines the weight of the combined prediction, and the selection of the evaluation index and the evaluation object is very key. The absolute value of the error between the predicted value and the actual value of the 2 single prediction methods is selected as 2 evaluation indexes, and 96 prediction points are used as 96 evaluation objects every day.
(1) Calculating the predicted value and the actual value of 2 single prediction methodsThe absolute value of the error of each single prediction method is used as a column vector, an evaluation matrix is established and is marked as A ═ aij]m×nWherein m is 96 and n is 2.
(2) Normalizing the evaluation matrix to obtain a normalized evaluation matrix R ═ Rij]m×nThe normalized formula is:
(3) normalized matrix R ═ Rij]m×nAnd calculating the proportion of the absolute value of the error of the ith prediction point under the jth method:
(4) entropy values corresponding to 2 methods are calculated:
(5) the combining weights for 3 methods are calculated:
(6) finding the predicted value of the combined prediction, P1,P2Predicted for 2 single methods, P3For the combined model prediction values:
P3=ω1P1+ω2P2
finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments or portions thereof without departing from the spirit and scope of the invention.
Claims (6)
1. A wind power interval prediction combination method based on signal decomposition is characterized by comprising the following steps:
1) performing data preprocessing on historical wind power data, and decomposing the historical data into a plurality of sub-signal components by adopting an EMD signal decomposition method;
2) inputting each sub-signal component into the combined model of the interval prediction algorithm independently for prediction to obtain a prediction sub-interval of each sub-signal component;
3) and synthesizing all the subintervals to obtain a final wind power interval prediction result.
2. The wind power interval prediction combination method based on signal decomposition according to claim 1, characterized in that a wind power certainty prediction model based on a naive Bayes algorithm is established in the step 2).
3. The wind power interval prediction combination method based on signal decomposition as claimed in claim 1, wherein the sub-signal components are predicted separately in step 1).
4. The wind power interval prediction combination method based on signal decomposition according to claim 1, characterized in that an interval prediction combination model of kernel density estimation and kernel limit learning machine based on an entropy weight method is established in the step 2).
5. The wind power interval prediction combination method based on signal decomposition according to claim 1, characterized in that deterministic prediction errors of the sub-signals obtained in step 2) are input into a combination model to obtain prediction sub-intervals of the sub-signal components.
6. The wind power interval prediction combination method based on signal decomposition as claimed in claim 2, wherein the naive bayes algorithm prediction error comprises kernel density estimation interval prediction and kernel limit learning machine interval prediction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110112003.4A CN112862166A (en) | 2021-01-27 | 2021-01-27 | Wind power interval prediction combination method based on signal decomposition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110112003.4A CN112862166A (en) | 2021-01-27 | 2021-01-27 | Wind power interval prediction combination method based on signal decomposition |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112862166A true CN112862166A (en) | 2021-05-28 |
Family
ID=75985967
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110112003.4A Pending CN112862166A (en) | 2021-01-27 | 2021-01-27 | Wind power interval prediction combination method based on signal decomposition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112862166A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016101182A1 (en) * | 2014-12-23 | 2016-06-30 | 清华大学 | Interval type indicator forecasting method based on bayesian network and extreme learning machine |
CN107798426A (en) * | 2017-10-16 | 2018-03-13 | 武汉大学 | Wind power interval Forecasting Methodology based on Atomic Decomposition and interactive fuzzy satisfying method |
CN110334875A (en) * | 2019-07-10 | 2019-10-15 | 东北电力大学 | Meter and the wind power combined probability prediction technique of evaluation index conflict |
CN111062512A (en) * | 2019-11-14 | 2020-04-24 | 广东电网有限责任公司 | Wind power prediction method |
-
2021
- 2021-01-27 CN CN202110112003.4A patent/CN112862166A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016101182A1 (en) * | 2014-12-23 | 2016-06-30 | 清华大学 | Interval type indicator forecasting method based on bayesian network and extreme learning machine |
CN107798426A (en) * | 2017-10-16 | 2018-03-13 | 武汉大学 | Wind power interval Forecasting Methodology based on Atomic Decomposition and interactive fuzzy satisfying method |
CN110334875A (en) * | 2019-07-10 | 2019-10-15 | 东北电力大学 | Meter and the wind power combined probability prediction technique of evaluation index conflict |
CN111062512A (en) * | 2019-11-14 | 2020-04-24 | 广东电网有限责任公司 | Wind power prediction method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Tian et al. | Multi-step short-term wind speed prediction based on integrated multi-model fusion | |
CN111091233B (en) | Short-term wind power prediction modeling method for wind power plant | |
Hou et al. | Review of load forecasting based on artificial intelligence methodologies, models, and challenges | |
Pei et al. | Short-term regional wind power forecasting based on spatial–temporal correlation and dynamic clustering model | |
CN111506868B (en) | Ultra-short-term wind speed prediction method based on HHT weight optimization | |
CN114781723A (en) | Short-term photovoltaic output prediction method based on multi-model fusion | |
CN114169251A (en) | Ultra-short-term wind power prediction method | |
CN116341613A (en) | Ultra-short-term photovoltaic power prediction method based on Informar encoder and LSTM | |
CN117633446A (en) | Wind speed prediction method and system based on CEEMD-SE-EWT double decomposition and ARIMA combined prediction | |
CN116151464A (en) | Photovoltaic power generation power prediction method, system and storable medium | |
CN110675278A (en) | Photovoltaic power short-term prediction method based on RBF neural network | |
CN114611757A (en) | Electric power system short-term load prediction method based on genetic algorithm and improved depth residual error network | |
CN117688362A (en) | Photovoltaic power interval prediction method and device based on multivariate data feature enhancement | |
CN113627674A (en) | Distributed photovoltaic power station output prediction method and device and storage medium | |
CN117132132A (en) | Photovoltaic power generation power prediction method based on meteorological data | |
Juan et al. | Multi-step prediction of dissolved oxygen in river based on random forest missing value imputation and attention mechanism coupled with recurrent neural network | |
Yang et al. | Investigating the predictability of photovoltaic power using approximate entropy | |
Xu et al. | A novel hybrid wind speed interval prediction model based on mode decomposition and gated recursive neural network | |
CN112862166A (en) | Wind power interval prediction combination method based on signal decomposition | |
Xia et al. | A Novel Hybrid Model for Short‐Term Wind Speed Forecasting Based on Twice Decomposition, PSR, and IMVO‐ELM | |
Dai et al. | Multimodal deep learning water level forecasting model for multiscale drought alert in Feiyun River basin | |
Bin et al. | A short-term power load forecasting method based on eemd-abgru | |
Zhang et al. | A Novel Combined Model Based on Hybrid Data Decomposition, MSWOA and ENN for Short-Term Wind Speed Forecasting | |
Jin et al. | A hybrid prediction framework based on deep learning for wind power | |
Yang et al. | A hybrid vmd-based ARIMA-LSTM model for day-ahead pv prediction and uncertainty analysis |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |