CN110570030A - Wind power cluster power interval prediction method and system based on deep learning - Google Patents
Wind power cluster power interval prediction method and system based on deep learning Download PDFInfo
- Publication number
- CN110570030A CN110570030A CN201910779807.2A CN201910779807A CN110570030A CN 110570030 A CN110570030 A CN 110570030A CN 201910779807 A CN201910779807 A CN 201910779807A CN 110570030 A CN110570030 A CN 110570030A
- Authority
- CN
- China
- Prior art keywords
- wind power
- prediction
- deep learning
- variables
- data
- 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 46
- 238000013135 deep learning Methods 0.000 title claims abstract description 42
- 238000005457 optimization Methods 0.000 claims abstract description 27
- 239000002245 particle Substances 0.000 claims abstract description 18
- 230000009467 reduction Effects 0.000 claims abstract description 11
- 238000012847 principal component analysis method Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 12
- 238000012545 processing Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 9
- 238000000513 principal component analysis Methods 0.000 claims description 7
- 238000005259 measurement Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 13
- 238000004590 computer program Methods 0.000 description 7
- 239000011159 matrix material Substances 0.000 description 6
- 238000011156 evaluation Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000001186 cumulative effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 230000002596 correlated effect Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000010219 correlation analysis Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000000875 corresponding effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000002803 fossil fuel Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
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
- 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
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
the invention provides a method and a system for wind power cluster power interval prediction based on deep learning, wherein numerical weather forecast and historical wind power of each wind power station are obtained as original input data, mutual information between explanatory variables and target variables in an area is extracted by calculating mutual information of the explanatory variables to extract correlation information, the explanatory variables meeting the correlation degree are selected, a principal component analysis method is used for data reconstruction and dimensionality reduction, an interval constraint condition is built, a prediction model is built by using the deep learning, the reconstructed and dimensionality reduction data input model is trained, model optimization is performed by combining a particle swarm optimization method, a final prediction model is determined, power interval prediction is performed by using the final prediction model, and the method and the system have high accuracy.
Description
Technical Field
The disclosure belongs to the field of wind power prediction, and particularly relates to a method and a system for wind power cluster power interval prediction based on deep learning.
background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
Environmental problems caused by the combustion of a large amount of fossil fuels, energy depletion and other problems are receiving more and more global attention, and the rapid development of renewable clean energy is becoming a common consensus of various countries. However, the method is not used for the characteristic of strong controllability of traditional energy, and wind power has intermittence and randomness, so that the access of high-proportion wind power to a power grid brings a serious challenge to the economic, safe and stable operation of a power system. An accurate and reliable wind power prediction result is one of important means for solving the problem.
the wind power cluster mainly refers to a set of a plurality of wind power stations in an area. In recent years, most of researches mainly focus on the prediction of the output of a single wind power station, and the prediction of the power of a wind power cluster is relatively less. In fact, research on wind power prediction has been continued for many years, and it can be mainly classified into single-value prediction and probability prediction according to the expression form of results. Some single-value prediction methods are applied to the field of wind power prediction. Although some of the methods can obtain more accurate prediction results, the single-value prediction has a considerable problem, namely: due to the data loss and the wind power fluctuation characteristics, prediction errors are inevitably introduced into single value prediction, and the determined prediction result cannot provide uncertainty information about the wind power. The wind power prediction result is utilized to be limited in the decision making process based on random optimization or risk assessment.
in order to describe the randomness and the variability of wind power, in the past time, a wind power probability prediction technology is rapidly developed, and various scholars propose various probability prediction methods, such as quantile point regression, conditional kernel density estimation, interval prediction, a sparse Bayesian learning method and the like. Compared with a deterministic method, the probabilistic prediction can provide more information about wind power uncertainty for meeting the requirements of different decision objectives. Up to now, wind power probability prediction technology has been applied to the aspects of power generation planning, standby configuration, optimal unit combination, power market and the like and achieves better effects. However, as the inventor knows, most of the current probability prediction research only focuses on a single wind power station, and only uses numerical weather forecast (NWP) and historical data of a local wind power station to predict wind power, in fact, a certain relevance necessarily exists among a plurality of wind power stations in a region, and the wind power cluster power probability prediction result can be remarkably improved by effectively using the relevance.
disclosure of Invention
The method and the system directly utilize original data of each station to predict the wind power cluster, extract associated information by calculating mutual information between interpretation variables and target variables in a region on the basis of the original data, select highly correlated interpretation variables, and then perform data reconstruction and dimension reduction by using a principal component analysis method, thereby improving probability prediction efficiency. And finally, constructing an interval constraint condition, constructing a prediction model by using deep learning, and performing model optimization by using a particle swarm optimization method, so that the method has certain advancement, accuracy and effectiveness.
According to some embodiments, the following technical scheme is adopted in the disclosure:
A wind power cluster power interval prediction method based on deep learning comprises the following steps:
Obtaining numerical weather forecast and historical wind power of each wind power station as original input data, extracting mutual information between an explanatory variable and a target variable in a region by calculating mutual information of the explanatory variable to extract associated information, selecting the explanatory variable conforming to the degree of correlation, performing data reconstruction and dimensionality reduction by using a principal component analysis method, constructing an interval constraint condition, constructing a prediction model by using deep learning, training the reconstructed and dimensionality reduced data input model, performing model optimization by combining a particle swarm optimization method, determining a final prediction model, and performing power interval prediction by using the final prediction model.
Based on the obtained interval prediction result, the configuration of the standby units in the wind power cluster can be carried out, and the configuration specifically comprises capacity configuration and position configuration.
and the optimal set combination can be determined, so that the wind power cluster can be designed or constructed, and the order, safety and high efficiency of power utilization can be ensured.
As an alternative implementation mode, the total power of the wind power cluster is used as a target variable, NWP data and historical measurement data of each wind power station in the cluster are used as interpretation variables, mutual information between the interpretation variables and the target variable is calculated, the mutual information is calculated from a sample by using a law of large numbers, and a group of interpretation variables most relevant to the target variable is selected by calculating the mutual information between the interpretation variables and the target variable.
As an alternative embodiment, unified normalization processing is carried out on the data, the data are enabled to be between [0,1], mutual information is calculated through selection of the interpretation variables, and historical wind power, irradiance, temperature and humidity variables are selected as the interpretation variables according to the size of the mutual information.
as an alternative implementation mode, the wind power cluster covers a plurality of stations, the principal component analysis is utilized to reduce the dimension of the explanatory variables, and meanwhile, the key features of the explanatory variables are extracted, so that the input data of deep learning are independent.
As an alternative implementation mode, a prediction model is built by utilizing deep learning, interval prediction is converted into a multi-objective optimization problem, optimization is carried out by utilizing the deep learning, and the optimal weight is sought.
As an alternative embodiment, the objective of the multi-objective optimization is to minimize the width of the prediction interval at a given prediction interval coverage.
as an alternative embodiment, the fitness value of each particle is calculated according to a fitness function, and for each particle, its fitness value is compared with the fitness value for which its history is optimal, if better, it is taken as historical optimal, its fitness value is compared with the fitness value for the best location experienced by the population, if better, it is taken as group optimal.
A wind power cluster power interval prediction system based on deep learning comprises:
the data processing module is used for acquiring numerical weather forecast and historical wind power of each wind power station as original input data, and extracting mutual information between the interpretation variables and the target variables in the region to extract associated information by calculating the mutual information of the interpretation variables;
The dimension reduction module is configured to select the interpretation variables according with the correlation degree, and data reconstruction and dimension reduction are carried out by using a principal component analysis method;
The model building module is configured to build an interval constraint condition, build a prediction model by using deep learning, input the reconstructed and dimensionality-reduced data into the model for training, perform model optimization by combining a particle swarm optimization method, determine a final prediction model, and perform power interval prediction by using the final prediction model.
a computer readable storage medium having stored therein a plurality of instructions adapted to be loaded by a processor of a terminal device and to execute the method for deep learning based wind power cluster power interval prediction.
a terminal device comprising a processor and a computer readable storage medium, the processor being configured to implement instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method for wind power cluster power interval prediction based on deep learning.
Compared with the prior art, the beneficial effect of this disclosure is:
Compared with the prediction of the output of a single wind power station, the wind power cluster power prediction can directly provide information for a power system decision maker, so that a reasonable power generation plan is made, and a standby plan is made. Meanwhile, the dependence on station prediction is reduced, and therefore the generation of abandoned wind is avoided. According to the characteristics of more explanation variables, large data volume and complexity of wind power cluster power prediction, mutual information and principal component analysis are utilized to carry out dimensionality reduction on initial data, probabilistic interval prediction is converted into an optimization problem with constraints, deep learning is utilized to mine a nonlinear relation, and an accurate prediction result can be obtained.
drawings
the accompanying drawings, which are included to provide a further understanding of the disclosure, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure and are not to limit the disclosure.
FIG. 1 is a schematic diagram of a deep learning basic structure;
FIG. 2 is a flow chart of interval prediction;
FIG. 3 is a schematic diagram of mutual information arrangement of wind power stations in a cluster;
FIG. 4 is a diagram illustrating key explanatory variable selection;
FIG. 5 is a graph of predicted interval width versus variation;
FIG. 6 is a diagram illustrating a prediction result of an interval based on deep learning;
fig. 7 is a schematic diagram of the interval prediction result of the raw data input.
The specific implementation mode is as follows:
The present disclosure is further described with reference to the following drawings and examples.
It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.
it is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
A short-term wind power cluster power interval prediction method based on deep learning directly utilizes original data of each station to predict wind power cluster power. Firstly, extracting mutual information between an interpretation variable and a target variable in a region by calculating the mutual information of the interpretation variable on the basis of initial data to extract associated information, selecting a highly-relevant interpretation variable, and then performing data reconstruction and dimension reduction by using a principal component analysis method to improve probability prediction efficiency. And finally, constructing an interval constraint condition, constructing a prediction model by using deep learning, and performing model optimization by using a particle swarm optimization method.
Wind power cluster power prediction is different from a prediction technology of a single station, and the wind power cluster power prediction comprises initial data of a plurality of wind power stations, so that the data volume is large, the interpretation variables are many and complex, and the stations are necessarily related. Therefore, the following two factors must be considered to improve the prediction accuracy of the wind power cluster power:
First, the data quality is obtained. The detailed geographic information, meteorological conditions and historical measurement data of the wind power cluster station can improve the prediction precision. In fact, most wind power stations have incomplete data integrity, and a plurality of wind power stations have to correspond to each other one by one at the data acquisition time point, so the acquisition quality of wind power cluster data is more difficult to guarantee. Thus, the data type should not be overly dependent in the correlation analysis and predictive modeling process. The numerical weather forecast and the historical wind power are data which must be stored, and the correlation between the stations can be analyzed through the two types of data. Therefore, the numerical weather forecast and the historical wind power of each wind power station in the cluster are selected as the most reasonable original input data in the embodiment.
Second, the input data dimension. In theory, sufficient input data is beneficial to improve the prediction accuracy, but also brings about the problems of increasing the calculation pressure and difficulty in model estimation. As the data dimensionality increases, the computational efficiency of most algorithms decreases and the computation time increases. To solve this problem, first, correlation analysis is performed on all input data using mutual information, and highly correlated explanatory variables are selected. And then reconstructing and reducing the dimension of the data again by using a principal component analysis method. Finally, a correlation data input model is extracted.
the information theory proposed by shannon covers the concepts of information entropy and mutual information. The information entropy of any random variable is the amount of information contained in this variable. The information entropy of the random variable X can be represented by equation (1):
H(X)＝∫-fX·log(fX) (1)
Wherein f isXIs a probability density function of the variable X.
Entropy is often used to measure the information content of a physical or artificial system. Mutual information is based on entropy and is a measure of useful information, which can be understood as the amount of information contained in one random variable about another random variable, namely: the reduced uncertainty of one random variable due to the knowledge of the other random variable. So we can use mutual information to measure the degree of association between two random variables. The average mutual information for random variables X and Y can be represented by equation (2):
Wherein f isYIs a probability density function of a random variable Y, fX,YIs a joint probability density function of the random variables X and Y.
as can be seen from equation (2), when fX,Y＝fX·fYThis means that the random variables X and Y are independent of each other,i.e. I (X; Y) equals zero. Conversely, if I (X; Y) is greater than zero, it indicates that an association exists between the two variables. The higher the mutual information value, the stronger the correlation between variables.
The correlation coefficient method can be used for calculating the correlation, however, the correlation coefficient can only reflect the linear relation among variables, the wind power station is a complex artificial system, the nonlinear relation among data of the wind power station is particularly prominent, mutual information can not only reflect the linear relation, but also reflect the nonlinear relation of the mutual information, and therefore the correlation of the mutual information among the reaction variables is more comprehensive than the correlation coefficient.
Taking the total power of the wind power cluster as a target variable, taking NWP data and historical measurement data of each wind power station in the cluster as an interpretation variable, and then calculating mutual information between the interpretation variable and the target variable, wherein for simplifying calculation, the mutual information can be calculated from a sample by using a law of large numbers:
By calculating mutual information between the explanatory variables and the target variables, a set of explanatory variables highly correlated with the target variables is selected.
The selected interpretation variables of the wind power cluster power prediction are relatively more, and the data dimensionality is higher. Meanwhile, information contained in some variables is redundant in the interpretation variables extracted through mutual information, and the principal component analysis method can effectively extract key interpretation variables and main features, reduce data dimensionality and enable the key interpretation variables to be independent and reflect more information as much as possible.
Principal component analysis first normalizes the explanatory variables, assuming m explanatory variables X1,X2,X3,…,XmTo represent the respective characteristics of the target variable, the number of samples being N, which can be represented by an N × m matrix, i.e.
Then its center is normalized to:
Wherein,And sjTo explain the variable xjMean value and square ofAnd (4) poor.
Calculating an autocorrelation matrix of the explanatory variables by the obtained central normalization matrix:
Here, theIs a center normalized N x m sample matrix, N is the number of samples, m is the number of variables, and R is the autocorrelation matrix. Calculating m eigenvalues λ of the autocorrelation matrix R1>λ2>,…,>λmand the corresponding feature vector P.
calculating the variance contribution rate and the cumulative variance contribution rate of each feature vector:
Where i is 1,2, …, m.
If the cumulative variance contribution ratio of the first p feature vectors is greater than 85% -95%, the number of principal components is determined to be p. The principal component selected at this point already contains most of the information that the original variable can provide.
generally, the evaluation of the section prediction result mainly uses a prediction section coverage (PICP) and a section width (PINRW) as evaluation indexes. In the embodiment, the evaluation index is used as an optimization condition, and the target training is performed by using deep learning.
Evaluation index of section prediction
(1) Prediction Interval Coverage (PICP)
Generally, the prediction section coverage is used for evaluating the reliability of the model, and is one of important evaluation indexes for section prediction. The predicted interval coverage can be expressed as:
Where N is the number of samples, εtIs a boolean variable used to represent the relationship between the prediction interval and the target value.
The relationship is specifically expressed as:
Wherein L istand Utand respectively predicting a lower interval limit and an upper interval limit.
Therefore, ideally, the target value should be covered by the prediction interval in its entirety, i.e. PICP is 100%.
(2) Width of Prediction Interval (PINRW)
In addition to the PICP being able to evaluate the quality of the prediction interval, the width of the prediction interval must also be taken into account. The minimum value and the maximum value of the target variable are assumed to be used as the upper limit and the lower limit of the prediction interval, and although the target variable can be wrapped in the target variable well, the prediction interval is too wide, so that the reference significance to a decision maker is not large. Therefore, the prediction interval width is an important index for measuring the interval prediction acuity. The expression form is as follows:
Wherein R represents the difference between the maximum value and the minimum value of the target variable. The smaller the PINAW, the more sensitive the predictive model.
The deep learning method has the characteristics of distributed parallel processing, self-adaptive learning, nonlinear mapping and generalization capability, and has strong adaptability to wind power cluster prediction. Deep learning is a supervised learning model. The structure is generally shown in fig. 1.
In the inter-prediction we always pursue the largest PICP and the smallest PINAW, i.e. pursue higher reliability and better acuity. Therefore, the embodiment converts the interval prediction into a multi-objective optimization problem, optimizes the interval prediction by using deep learning, and seeks the optimal weight.
If the PICP is given in advance, equation (12) can be converted to a single target optimization problem:
And performing target optimization by using a Particle Swarm Optimization (PSO). The flow chart of the interval prediction is shown in fig. 2.
According to the flow chart of fig. 2, the flow of the prediction model of the present embodiment can be summarized as follows:
1) And (4) preprocessing data. Firstly, the data is subjected to unified normalization processing, so that the data is between [0,1 ]. The selection of the interpretation variables mainly calculates mutual information, and the variables such as historical wind power, irradiance, temperature, humidity and the like are selected as the interpretation variables according to the size of the mutual information. And meanwhile, the correlation information is calculated by utilizing the mutual information.
2) And (5) reducing the dimension of the data. A plurality of stations are covered in the wind power cluster, so that the dimensionality of input data is overlarge, and the precondition for modeling by utilizing deep learning is that the input data are mutually independent. Therefore, the principal component analysis can be used for not only reducing the dimension of the explanatory variables, but also extracting the key features of the explanatory variables, so that the input data of deep learning are independent.
3) And constructing a deep learning model. Optimizing the structure of the neural network, and determining the number of hidden layers and nodes. An optimization objective is constructed, giving PICP, such that the PINAW value is minimal.
4) And initializing the neural network weight and the particle swarm algorithm parameters. The PSO algorithm parameter initialization includes particle position and velocity. The particle position is represented by the weight of the neural network and the velocity is initialized randomly.
5) And calculating the fitness value of each particle according to the fitness function. For each particle, comparing its fitness value with its historical record-optimal fitness value, and if better, taking it as historical optimal (pbest); its fitness value is compared to the fitness value of the best location experienced by the population, and if better, it is considered as the best (gbest) of the population.
6) and (4) finishing training: the training end criterion may be set to the maximum number of iterations, otherwise the training process will continue and return to step 5.
7) And (6) testing and evaluating.
As verification, data of 10 wind power stations in a certain area in China are mainly used for predicting the total power of the wind power cluster within 72 hours in the future and with the time resolution of 15min, and the prediction intervals are 80% and 90%. The data set is divided into a training set and a validation set. The training set is used for establishing an interval prediction model for deep learning, and the verification set is used for testing the performance of the model.
Processing the respective power of the wind power station into cluster power, and calculating mutual information between the target variable and the initial interpretation variable from the sample data, wherein the mutual information is shown in fig. 3. And selecting an explanation variable with mutual information larger than 0.45 as a key explanation variable. The size order is shown in figure 4.
After the key interpretation variables are selected through the mutual information, the principal component analysis is used for reducing the dimension of the key interpretation variables, and key features are extracted to enable the key interpretation variables to be mutually independent. The contribution rate and the cumulative contribution rate of each feature were calculated, and principal components were extracted, as shown in table 1.
TABLE 1 principal Components analysis results
In general, when the variance cumulative contribution rate reaches 80% to 95%, we consider it as a main component. Therefore, the first 12 feature variables are selected as the key features in the present embodiment.
In the embodiment, the key features obtained through mutual information and principal component analysis are used as input data of the neural network. And (5) carrying out sample training. In the comparative example, original data without data preprocessing is used for sample training, and 80% and 90% prediction intervals are obtained respectively. The predicted time points of 80% and 90% interval widths are shown in fig. 5.
As can be seen from fig. 5, the method proposed in this embodiment has a narrower prediction interval compared to the result obtained by inputting the original data, regardless of the prediction interval of 80% or the prediction interval of 90%. This shows that the prediction result obtained by the method of this embodiment has better acuity, and can provide more reliable and comprehensive information for the decision maker. In addition, as the prediction time increases, the prediction interval width significantly increases, because as the time scale extends, the obtained data such as weather becomes more and more unreliable, and the uncertainty information increases, so that the prediction effect becomes worse.
Fig. 6 and 7 show the prediction results of 90% and 80% confidence intervals of the total power of the wind power cluster on certain three days obtained by raw data input and feature extraction, respectively. Comparing fig. 6 and fig. 7, it can be seen that the acuity of the prediction result after feature extraction is better, and the uncertain information contained is more comprehensive. The confidence interval band gets wider as the prediction time scale increases, just demonstrating the results of the PINAW change in fig. 5. Meanwhile, we can also see that the prediction effect of fig. 6 is significantly better than that of fig. 7, which illustrates that the method proposed in this embodiment has good applicability.
As will be appreciated by one skilled in the art, embodiments of the present disclosure may be provided as a method, system, or computer program product. Accordingly, the present disclosure may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present disclosure may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present disclosure is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the disclosure. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may 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, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure, and various modifications and changes may be made to the present disclosure by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present disclosure should be included in the protection scope of the present disclosure.
although the present disclosure has been described with reference to specific embodiments, it should be understood that the scope of the present disclosure is not limited thereto, and those skilled in the art will appreciate that various modifications and changes can be made without departing from the spirit and scope of the present disclosure.
Claims (10)
1. A wind power cluster power interval prediction method based on deep learning is characterized by comprising the following steps: the method comprises the following steps:
obtaining numerical weather forecast and historical wind power of each wind power station as original input data, extracting mutual information between an explanatory variable and a target variable in a region by calculating mutual information of the explanatory variable to extract associated information, selecting the explanatory variable conforming to the degree of correlation, performing data reconstruction and dimensionality reduction by using a principal component analysis method, constructing an interval constraint condition, constructing a prediction model by using deep learning, training the reconstructed and dimensionality reduced data input model, performing model optimization by combining a particle swarm optimization method, determining a final prediction model, and performing power interval prediction by using the final prediction model.
2. The method for wind power cluster power interval prediction based on deep learning as claimed in claim 1, characterized by: the total power of the wind power cluster is used as a target variable, weather forecast data and historical measurement data of each wind power station in the cluster are used as interpretation variables, mutual information between the interpretation variables and the target variable is calculated, the mutual information is calculated from a sample by using a law of large numbers, and a group of interpretation variables most related to the target variable is selected by calculating the mutual information between the interpretation variables and the target variable.
3. the method for wind power cluster power interval prediction based on deep learning as claimed in claim 1, characterized by: and carrying out unified normalization processing on the data to enable the data to be between [0,1], selecting and calculating mutual information of the interpretation variables, and selecting historical wind power, irradiance, temperature and humidity variables as the interpretation variables according to the size of the mutual information.
4. The method for wind power cluster power interval prediction based on deep learning as claimed in claim 1, characterized by: a plurality of stations are covered in the wind power cluster, the principal component analysis is utilized to reduce the dimension of the explanatory variables, and meanwhile, the key features of the explanatory variables are extracted, so that the input data of deep learning are independent.
5. The method for wind power cluster power interval prediction based on deep learning as claimed in claim 1, characterized by: and (3) constructing a prediction model by utilizing deep learning, converting interval prediction into a multi-objective optimization problem, optimizing by utilizing the deep learning, and seeking optimal weight.
6. The method for wind power cluster power interval prediction based on deep learning as claimed in claim 1, characterized by: the objective of the multi-objective optimization is that the width value of the prediction interval is minimum under the given coverage rate of the prediction interval.
7. The method for wind power cluster power interval prediction based on deep learning as claimed in claim 1, characterized by: and calculating the fitness value of each particle according to the fitness function, comparing the fitness value of each particle with the fitness value with the optimal historical record, if the fitness value is better, taking the fitness value as the historical optimal fitness, comparing the fitness value of each particle with the fitness value of the best position experienced by the group, and if the fitness value is better, taking the fitness value as the group optimal.
8. A wind power cluster power interval prediction system based on deep learning is characterized in that: the method comprises the following steps:
the data processing module is used for acquiring numerical weather forecast and historical wind power of each wind power station as original input data, and extracting mutual information between the interpretation variables and the target variables in the region to extract associated information by calculating the mutual information of the interpretation variables;
the dimension reduction module is configured to select the interpretation variables according with the correlation degree, and data reconstruction and dimension reduction are carried out by using a principal component analysis method;
The model building module is configured to build an interval constraint condition, build a prediction model by using deep learning, input the reconstructed and dimensionality-reduced data into the model for training, perform model optimization by combining a particle swarm optimization method, determine a final prediction model, and perform power interval prediction by using the final prediction model.
9. A computer-readable storage medium characterized by: a plurality of instructions are stored, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the method for wind power cluster power interval prediction based on deep learning in any one of claims 1 to 7.
10. A terminal device is characterized in that: the system comprises a processor and a computer readable storage medium, wherein the processor is used for realizing instructions; the computer readable storage medium is used for storing a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing the method for wind power cluster power interval prediction based on deep learning of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910779807.2A CN110570030A (en) | 2019-08-22 | 2019-08-22 | Wind power cluster power interval prediction method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910779807.2A CN110570030A (en) | 2019-08-22 | 2019-08-22 | Wind power cluster power interval prediction method and system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110570030A true CN110570030A (en) | 2019-12-13 |
Family
ID=68774382
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910779807.2A Pending CN110570030A (en) | 2019-08-22 | 2019-08-22 | Wind power cluster power interval prediction method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110570030A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111353651A (en) * | 2020-03-12 | 2020-06-30 | 广西电网有限责任公司 | Regional power prediction method, device, equipment and storage medium |
| CN111414717A (en) * | 2020-03-02 | 2020-07-14 | 浙江大学 | XGboost-L ightGBM-based unit power prediction method |
| CN111428766A (en) * | 2020-03-17 | 2020-07-17 | 深圳供电局有限公司 | Power consumption mode classification method for high-dimensional mass measurement data |
| CN111667093A (en) * | 2020-04-22 | 2020-09-15 | 华北电力科学研究院有限责任公司 | Medium-and-long-term wind power generation calculation method and device |
| CN111783999A (en) * | 2020-07-01 | 2020-10-16 | 北京知因智慧科技有限公司 | Data processing method and device |
| CN111815039A (en) * | 2020-06-28 | 2020-10-23 | 山东大学 | Weekly scale wind power probability prediction method and system based on weather classification |
| CN112785033A (en) * | 2020-12-13 | 2021-05-11 | 国网辽宁省电力有限公司经济技术研究院 | Wind power disorder index and prediction effect evaluation method based on information entropy theory |
| CN112801350A (en) * | 2021-01-13 | 2021-05-14 | 中国电建集团华东勘测设计研究院有限公司 | Deep learning ultra-short-time wind power prediction method based on uncertainty |
| CN113516315A (en) * | 2021-07-23 | 2021-10-19 | 中南大学 | Wind power generation power interval prediction method, device and medium |
| CN113887843A (en) * | 2021-12-07 | 2022-01-04 | 中国电力科学研究院有限公司 | Group prediction method and system for wind power plant output power |
| CN114118638A (en) * | 2022-01-28 | 2022-03-01 | 华控清交信息科技(北京)有限公司 | Wind power plant power prediction method, GBDT model transverse training method and device |
| CN115619170A (en) * | 2022-10-28 | 2023-01-17 | 北京国电通网络技术有限公司 | Method, device, equipment, computer medium and program product for adjusting electric quantity load |
| CN118629265A (en) * | 2024-08-13 | 2024-09-10 | 中国水产科学研究院南海水产研究所 | Fishing boat anti-collision early warning method and system based on system safety theory |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105046374A (en) * | 2015-08-25 | 2015-11-11 | 华北电力大学 | Power interval predication method based on nucleus limit learning machine model |
| CN106650982A (en) * | 2016-08-30 | 2017-05-10 | 华北电力大学 | Depth learning power prediction method based on multi-point NWP |
| CN106875033A (en) * | 2016-12-26 | 2017-06-20 | 华中科技大学 | A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting |
| 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 |
| CN108667069A (en) * | 2018-04-19 | 2018-10-16 | 河海大学 | A kind of short-term wind power forecast method returned based on Partial Least Squares |
| CN109214605A (en) * | 2018-11-12 | 2019-01-15 | 国网山东省电力公司电力科学研究院 | Power-system short-term Load Probability prediction technique, apparatus and system |
| CN109242143A (en) * | 2018-07-31 | 2019-01-18 | 中国电力科学研究院有限公司 | A kind of neural network wind power forecasting method and system |
| US20190219994A1 (en) * | 2018-01-18 | 2019-07-18 | General Electric Company | Feature extractions to model large-scale complex control systems |
-
2019
- 2019-08-22 CN CN201910779807.2A patent/CN110570030A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105046374A (en) * | 2015-08-25 | 2015-11-11 | 华北电力大学 | Power interval predication method based on nucleus limit learning machine model |
| CN106650982A (en) * | 2016-08-30 | 2017-05-10 | 华北电力大学 | Depth learning power prediction method based on multi-point NWP |
| CN106875033A (en) * | 2016-12-26 | 2017-06-20 | 华中科技大学 | A kind of wind-powered electricity generation cluster power forecasting method based on dynamic self-adapting |
| 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 |
| US20190219994A1 (en) * | 2018-01-18 | 2019-07-18 | General Electric Company | Feature extractions to model large-scale complex control systems |
| CN108667069A (en) * | 2018-04-19 | 2018-10-16 | 河海大学 | A kind of short-term wind power forecast method returned based on Partial Least Squares |
| CN109242143A (en) * | 2018-07-31 | 2019-01-18 | 中国电力科学研究院有限公司 | A kind of neural network wind power forecasting method and system |
| CN109214605A (en) * | 2018-11-12 | 2019-01-15 | 国网山东省电力公司电力科学研究院 | Power-system short-term Load Probability prediction technique, apparatus and system |
Non-Patent Citations (2)
| Title |
|---|
| 熊磊: "基于动态自适应技术的风电集群短期功率预测研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 高志强等: "《深度学习 从入门到实战》", 30 June 2018, 中国铁道出版社 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111414717A (en) * | 2020-03-02 | 2020-07-14 | 浙江大学 | XGboost-L ightGBM-based unit power prediction method |
| CN111353651A (en) * | 2020-03-12 | 2020-06-30 | 广西电网有限责任公司 | Regional power prediction method, device, equipment and storage medium |
| CN111428766A (en) * | 2020-03-17 | 2020-07-17 | 深圳供电局有限公司 | Power consumption mode classification method for high-dimensional mass measurement data |
| CN111428766B (en) * | 2020-03-17 | 2024-01-19 | 深圳供电局有限公司 | Power consumption mode classification method for high-dimensional mass measurement data |
| CN111667093A (en) * | 2020-04-22 | 2020-09-15 | 华北电力科学研究院有限责任公司 | Medium-and-long-term wind power generation calculation method and device |
| CN111667093B (en) * | 2020-04-22 | 2023-10-24 | 华北电力科学研究院有限责任公司 | Medium-and-long-term wind power generation calculation method and device |
| CN111815039A (en) * | 2020-06-28 | 2020-10-23 | 山东大学 | Weekly scale wind power probability prediction method and system based on weather classification |
| CN111783999A (en) * | 2020-07-01 | 2020-10-16 | 北京知因智慧科技有限公司 | Data processing method and device |
| CN112785033A (en) * | 2020-12-13 | 2021-05-11 | 国网辽宁省电力有限公司经济技术研究院 | Wind power disorder index and prediction effect evaluation method based on information entropy theory |
| CN112801350B (en) * | 2021-01-13 | 2022-09-06 | 中国电建集团华东勘测设计研究院有限公司 | Uncertainty-based deep learning ultra-short-time wind power prediction system and method |
| CN112801350A (en) * | 2021-01-13 | 2021-05-14 | 中国电建集团华东勘测设计研究院有限公司 | Deep learning ultra-short-time wind power prediction method based on uncertainty |
| CN113516315B (en) * | 2021-07-23 | 2023-02-03 | 中南大学 | Wind power generation power interval prediction method, device and medium |
| CN113516315A (en) * | 2021-07-23 | 2021-10-19 | 中南大学 | Wind power generation power interval prediction method, device and medium |
| CN113887843A (en) * | 2021-12-07 | 2022-01-04 | 中国电力科学研究院有限公司 | Group prediction method and system for wind power plant output power |
| CN114118638A (en) * | 2022-01-28 | 2022-03-01 | 华控清交信息科技(北京)有限公司 | Wind power plant power prediction method, GBDT model transverse training method and device |
| CN114118638B (en) * | 2022-01-28 | 2022-04-19 | 华控清交信息科技(北京)有限公司 | Wind power plant power prediction method, GBDT model transverse training method and device |
| CN115619170A (en) * | 2022-10-28 | 2023-01-17 | 北京国电通网络技术有限公司 | Method, device, equipment, computer medium and program product for adjusting electric quantity load |
| CN118629265A (en) * | 2024-08-13 | 2024-09-10 | 中国水产科学研究院南海水产研究所 | Fishing boat anti-collision early warning method and system based on system safety theory |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110570030A (en) | Wind power cluster power interval prediction method and system based on deep learning | |
| CN101982843B (en) | Method for selecting state vector in nonparametric regression short-time traffic flow prediction | |
| CN111242351A (en) | Tropical cyclone track prediction method based on self-encoder and GRU neural network | |
| CN116644970A (en) | Photovoltaic power prediction method based on VMD decomposition and lamination deep learning | |
| CN116316617B (en) | Multi-station intelligent fusion new energy generation power region prediction method and system | |
| CN117394306B (en) | Wind power prediction model establishment method based on new energy grid connection and application thereof | |
| CN113052469A (en) | Method for calculating wind-solar-water-load complementary characteristic of small hydropower area lacking measurement runoff | |
| CN113807605A (en) | Power consumption prediction model training method, prediction method and prediction device | |
| CN113554361A (en) | Comprehensive energy system data processing and calculating method and processing system | |
| Liu et al. | State of health estimation of lithium-ion batteries based on multi-feature extraction and temporal convolutional network | |
| CN117951577A (en) | Virtual power plant energy state sensing method | |
| Zhu et al. | Short-term wind speed prediction based on FEEMD-PE-SSA-BP | |
| CN113762591B (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM countermeasure learning | |
| CN117578434B (en) | Power distribution network flexibility evaluation method and device considering flexible resource adjustability | |
| CN114186412A (en) | Hydropower station water turbine top cover long sequence water level prediction system and method based on self-attention mechanism | |
| CN117113202A (en) | Power loop energy consumption detection method and equipment based on joint error stacking model | |
| Li et al. | Short-term load forecasting using support vector regression-based local predictor | |
| CN116561569A (en) | Industrial power load identification method based on EO feature selection and AdaBoost algorithm | |
| Hasan et al. | Solar Radiation Prediction Using Satin Bowerbird Optimization with Modified Deep Learning. | |
| Sarp et al. | Analysis of false data injection impact on AI based solar photovoltaic power generation forecasting | |
| Parrado-Duque et al. | A comparative analysis of machine learning methods for short-term load forecasting systems | |
| CN115217152A (en) | Method and device for predicting opening and closing deformation of immersed tunnel pipe joint | |
| CN112256735B (en) | Power consumption monitoring method and device, computer equipment and storage medium | |
| Katranji et al. | Short-Term Wind Speed Prediction for Saudi Arabia via 1D-CNN | |
| Gu et al. | Improved similarity-based residual life prediction method based on grey Markov model |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191213 |