CN112186761A - Wind power scene generation method and system based on probability distribution - Google Patents
Wind power scene generation method and system based on probability distribution Download PDFInfo
- Publication number
- CN112186761A CN112186761A CN202011064892.3A CN202011064892A CN112186761A CN 112186761 A CN112186761 A CN 112186761A CN 202011064892 A CN202011064892 A CN 202011064892A CN 112186761 A CN112186761 A CN 112186761A
- Authority
- CN
- China
- Prior art keywords
- distribution
- wind power
- wind
- error
- prediction
- 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.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/06—Wind turbines or wind farms
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
- H02J2300/20—The dispersed energy generation being of renewable origin
- H02J2300/28—The renewable source being wind energy
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Power Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The utility model provides a wind power scene generation method and system based on probability distribution, comprising: acquiring historical data of a predicted value and an actual measurement value of the wind power, and extracting a difference value between the predicted value and the actual measurement value of the wind power, namely a prediction error; selecting wind speed, wind direction and fitting errors which directly affect wind power, and respectively fitting the distribution of prediction errors under the influence of the wind speed, the wind direction and the fitting errors; combining the prediction error distributions under the three types of influence factors according to corresponding weight parameters, and solving the optimal solution of the weight by taking the minimum sum of the squares of the residuals of the probability density function and the distribution histogram as a target function to obtain a prediction error probability density function; discretizing the probability density function to obtain a large number of random scenes and selecting typical scenes. By adopting mixed distribution and setting weights for error distribution under different influence factors, the influence of each influence factor on the error distribution is comprehensively considered, and the obtained fitting result is more accurate compared with single error distribution.
Description
Technical Field
The disclosure belongs to the technical field of wind power generation, and particularly relates to a wind power scene generation method and system based on probability distribution.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
Wind energy is inexhaustible as a large-scale pollution-free renewable energy source; and the construction period of the wind power plant is short, the installed scale is flexible, the operation cost is low, and the wind power generation technology is widely favored based on the wind power generation technology. However, wind speed has the characteristic of intermittence, so that wind power generation has strong randomness and instability, the generated power quality and grid connection reliability are reduced, and great challenges are brought to the wind power generation technology. Aiming at the influence brought by the randomness of wind power fluctuation, the wind power prediction technology enables a wind power plant to predict the condition of wind power in a future period of time, and scheduling is convenient to arrange, so that the wind power plant becomes a mainstream technology in the field of current wind power generation. However, wind power is influenced by various natural and technical factors, strong irregularity is presented on a time scale, it is extremely difficult to directly describe volatility by greatly improving single-point prediction precision under the current technical level, probability analysis of prediction errors can make up for the deficiency of wind power prediction, and generating a wind power scene according to probability distribution is an important method for processing volatility problem in wind power generation.
The wind power scene generation technology mainly comprises the following two parts, namely volatility probability analysis and representative scene generation. The volatility probability analysis is based on a statistical method, and is used for counting error values generated by prediction of the wind power day ahead, fitting according to an error distribution histogram obtained by statistics, and finally obtaining a probability density function of error distribution. According to the probability distribution of the errors, the probability distribution condition of the wind power at each moment of the future moment can be obtained by combining the prediction of the day ahead, and random numbers of power values at each moment can be generated on the basis, so that a plurality of wind power curves, namely scene sets, are obtained.
The inventor finds that the existing wind power scene generation technology has the following problems in research. In the probability fitting, a single distribution function is often selected to fit the wind power prediction error at each moment, for example, normal distribution, and the single distribution function hardly shows the characteristic of 'tail after peak' of the wind power prediction error distribution and the difference of the error distribution at each moment. In the past, the influence of wind speed on prediction error is only considered in the fitting process, and the influence of other natural and technical factors is not comprehensively considered, so that the fitting result is inaccurate. In addition, a large number of wind power curves can be generated in the scene generation process, so that the efficiency of wind power plant scheduling optimization is reduced, and how to quickly select a representative typical scene is to be researched.
Disclosure of Invention
In order to overcome the defects of the prior art, the wind power scene generation method based on probability distribution is provided, and the problems that wind power prediction error probability analysis is inaccurate, the number of generated scenes is too large, the generated scenes are unrepresentative and the like in the existing research are solved.
In order to achieve the above object, one or more embodiments of the present disclosure provide the following technical solutions:
in a first aspect, a wind power scene generation method based on probability distribution is disclosed, which includes:
acquiring historical data of a predicted value and an actual measurement value of the wind power, and extracting a difference value between the predicted value and the actual measurement value of the wind power, namely a prediction error;
selecting wind speed, wind direction and fitting errors which directly affect wind power, and respectively fitting the distribution of prediction errors under the influence of the wind speed, the wind direction and the fitting errors;
combining the prediction error distributions under the three types of influence factors according to corresponding weight parameters, and solving the optimal solution of the weight by taking the minimum sum of the squares of the residuals of the probability density function and the distribution histogram as a target function to obtain a prediction error probability density function;
discretizing the probability density function to obtain a large number of random scenes and selecting typical scenes.
In a second aspect, a wind power scene generation system based on probability distribution is disclosed, which includes:
a prediction error acquisition module configured to: acquiring historical data of a predicted value and an actual measurement value of the wind power, and extracting a difference value between the predicted value and the actual measurement value of the wind power, namely a prediction error;
a prediction error probability density function acquisition module configured to: selecting wind speed, wind direction and fitting errors which directly affect wind power, and respectively fitting the distribution of prediction errors under the influence of the wind speed, the wind direction and the fitting errors;
combining the prediction error distributions under the three types of influence factors according to corresponding weight parameters, and solving the optimal solution of the weight by taking the minimum sum of the squares of the residuals of the probability density function and the distribution histogram as a target function to obtain a prediction error probability density function;
a representative scene acquisition module configured to: discretizing the probability density function to obtain a large number of random scenes and selecting typical scenes.
The above one or more technical solutions have the following beneficial effects:
aiming at the problem that the influence of a plurality of wind speeds on error distribution is counted in the prior art, the adopted distribution is too single, the technical scheme disclosed by the invention considers factors such as wind directions and fitting errors, and simultaneously adopts non-parameter estimation to express the nonstandard performance of the error distribution under the influence of the wind directions.
According to the technical scheme, mixed distribution is adopted, weights are set for error distribution under different influence factors, the influence of each influence factor on the error distribution is comprehensively considered, and the obtained fitting result is more accurate compared with single error distribution.
Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
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 graph of the results of statistical prediction errors for each wind speed segment according to an embodiment of the present disclosure;
FIG. 2 is a fitting graph of error distributions at various wind speed segments according to an embodiment of the present disclosure;
FIG. 3 is a wind power prediction error distribution diagram for four seasons according to an embodiment of the present disclosure;
FIG. 4 is a graph of predicted error distribution under power fitting error in accordance with an embodiment of the present disclosure;
FIG. 5 is a mixed distribution diagram of wind power prediction error fitted by taking spring wind speed of 6-7 m/s as an example in the embodiment of the disclosure;
FIG. 6 is a wind power scene set diagram corresponding to a matrix in an embodiment of the disclosure;
FIG. 7 is a graph showing a variation curve of a relative entropy index and a sum of squared errors index in a wind power curve clustering process according to an embodiment of the present disclosure;
fig. 8 is a scene set diagram of the clustered wind power curve in the embodiment of the present disclosure.
Detailed Description
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.
The embodiments and features of the embodiments in the present disclosure may be combined with each other without conflict.
The embodiment discloses a wind power scene generation method based on probability distribution, which comprises the following steps:
the method comprises the following steps: determining a prediction error according to historical data;
step two: fitting a distribution function of prediction errors under various influence factors;
step three: establishing a wind power prediction error mixed distribution model;
step four: according to the probability distribution, a Monte Carlo simulation method is applied to produce wind power scenes;
step five: and improving k-means clustering analysis to select typical scenes.
Specifically, the first step: a prediction error is determined from the historical data.
The method is based on probability analysis of historical data, so that certain probability characteristics are found, and future data can be predicted more accurately from the probability perspective. The traditional energy scheduling is to plan the output of equipment according to a wind power predicted value, and the day-ahead predicted values of other output equipment and multi-load of the energy system; however, the predicted data and the real-time data have a certain difference, which is more obvious in the aspect of wind power because the wind power generation is influenced by various natural and technical factors. The probability distribution of the difference value can be obtained through historical prediction and real-time data probability analysis, and a typical scene generated according to the distribution is equivalent to representing the uncertain difference value, so that the typical scene is closer to the real-time data compared with a predicted value.
The historical data comprises a predicted value and an actual measured value of the wind power at the historical moment, and the predicted value of the wind power is obtained by substituting the predicted value of the wind speed into the output curve of the fan in the prediction of the wind power at the day ahead. And fitting an air outlet machine output characteristic curve according to the wind speed measured value of the anemometer tower near the fan in one year and by referring to the relation between the single machine power and the cut-in wind speed and the rated wind speed. And calculating the difference value between the wind power predicted value and the local measured value at each moment.
In a specific implementation example, there are many methods for wind speed sequence prediction, and in the context of current machine learning, a more common method is a statistical model, such as establishing a wind speed time sequence model, performing deep learning by using an artificial neural network, and performing a least squares support vector machine model; no matter how the time series prediction method is improved, the obtained predicted value and the actual value have errors and are not considered, and the method is developed based on error analysis.
The rated output 16KW of a single fan is fit to the fan power curve according to the annual output condition of the fan and the corresponding moment as shown in the following formula.
And substituting the predicted value of the wind speed into a power curve of the fan, calculating the predicted value of the power at each moment, and extracting the difference value between the predicted value and the actual power.
And the subsequent steps are to count two pairs of difference values, the difference value distribution conditions under different wind speeds, the difference value distribution conditions under different seasons and the like. Through the series of analysis, the distribution rule of the difference values in the historical data is obtained and used for predicting the future data to obtain a typical scene.
In the above-mentioned formula,
the actual data of the wind power at the historical moment,
is prediction data of historical time, difference value delta PWT,tThe difference between the actual power and the predicted power is used for statistical analysis.
Step two: fitting distribution function of prediction error under various influence factors
Wind power prediction errors are influenced by various natural and technical factors to present irregularity, indirect influence factors such as temperature and humidity are ignored in the research method, wind speed, wind direction and fitting errors which directly influence wind power are selected, and the distribution conditions of the prediction errors under the influence of the three are respectively fitted, specifically as follows:
the wind speed is divided into 5 sections of 0-3 m/s, 3-6 m/s, 6-9 m/s, 9-12 m/s and 12-15 m/s, and the result of statistical prediction error of each wind speed section is shown in figure 1.
And fitting the error distribution under each wind speed section, wherein two sections with small head-to-tail error fluctuation of the wind speed are ignored, and the distribution characteristics of the errors under 3 wind speed sections of 3-6 m/s, 6-9 m/s and 9-12 m/s can be well described by adopting normal distribution with different standard deviations according to the past experience and the histogram of the current statistical result, and the effect is shown in the following figure 2.
Since the wind direction changes with the change of seasons and keeps constant in one day, the wind direction difference can be represented by seasons, such as the southeast wind prevailing in spring and the northwest wind prevailing in winter. The wind power prediction error histograms are counted according to different seasons as follows, but it can be seen that the error distributions in different seasons are highly irregular and cannot be represented by a determined probability density function. And (3) directly estimating the probability density of the selected sample point by using a K nearest neighbor method in non-parameter estimation in statistical learning without assuming the form of the probability density. The principle of the K-nearest neighbor method is simple and clear, assuming that n is the total number of sample points, taking a point x to be estimated as a center as an area Vn, and continuously expanding the area Vn to increase the number of sample points contained in the area until the area contains KnAfter each sample point, the formula is substituted:
the point probability density is calculated. And continuously substituting the error sample points to calculate to obtain an error distribution condition, and fitting to obtain the wind power prediction error distribution of four seasons as shown in the following figure 3.
In addition, the last part of the process of fitting the fan output curve has power characteristic fitting errors, the influence of the errors is relatively small and single, and the errors can be obtained through prediction error distribution under the actually measured wind speed. By extreme value distribution, the prediction error distribution under the power fitting error can be shown as shown in fig. 4.
Step three: establishing a wind power prediction error mixed distribution model
After the prediction error distributions of the three types of influence factors are determined, the three types of influence factors are comprehensively considered according to weight through mixed distribution. The intuitive understanding of the mixed distribution is to give a set { f (x1) … f (xi) } containing a plurality of random variable distributions, then generate a new complex random variable distribution f (x) based on the set, combine the simple distributions of the plurality of random variables in the generation process according to corresponding weights, and use a probability density function as follows, wherein f (x) isi) Is the probability distribution under each influence factor.
Combining prediction error distributions under three types of influence factors according to corresponding weight parameters, wherein when the weight parameters are determined, l1, l2 and l3 are respectively set as the weights of the three types of influence factors, e1, e2 and e3 are used as the errors of fitted distributions and a statistical histogram under the current seasonal wind speed, the minimum sum of squares of residuals of a probability density function and the distribution histogram is used as an objective function, the optimal solution of the weights is obtained as an equation, wherein i and j represent multiplication of error difference values of two differences, and the number i is 3.
Taking spring wind speed of 6-7 m/s as an example, fitting the mixed distribution of the wind power prediction errors as shown in fig. 5, and further determining the wind power prediction error distribution of each time according to the wind speed and wind direction conditions at each moment in the day, wherein the solution values of l1, l2 and l3 are 0.2, 0.65 and 0.15.
Step four: according to probability distribution, a Monte Carlo simulation method is applied to produce wind power scenes
According to a prediction error probability density function obtained by fitting a statistical model and a source load power point prediction value obtained by prediction in the day-ahead, wind power, photovoltaic and load output serving as uncertain random variables can be expressed as follows:
And discretizing the probability density function of the random variable by random sampling by adopting a Monte Carlo simulation method to obtain a large number of random scenes, thereby determining the random parameters in an uncertain way. The Monte Carlo simulation method generates n random numbers according to the probability distribution of the random parameters at the current moment, further generates a scene set containing n power curves according to the random numbers at all the moments in one day, and expresses the scene set in a matrix form.
The above analysis obtains the probability distribution of the prediction error, and further obtains the probability distribution of the future prediction value, and the probability distribution is a continuous random parameter. When the probability distribution of the random parameters is applied to energy scheduling management, specific data which obey the probability distribution needs to be obtained, and therefore the specific data, namely a typical scene, is obtained by discretizing the probability distribution by adopting a Monte Carlo sampling method.
W＝[w1,w2...wn]T
Taking the generated wind power scene set n × T dimensional matrix W as an example, W includes n generated wind power curves, each curve includes T ═ 24 power values in one day, and the wind power scene set corresponding to the matrix is as shown in fig. 6. The scene frequency histogram fitting experiment shows that when the scene number n in the matrix is 100, the probability density of the random variable can be reflected by the data set at each moment.
Step five: selecting typical scenes by improving k-means clustering analysis
The Monte Carlo simulation generates a large number of scenes at one time, the calculation efficiency is influenced, and the efficiency of the traditional single scene reduction algorithm is low. Therefore, the k-means clustering algorithm in machine learning is applied to realize efficient scene reduction. The k-means clustering algorithm is the most common one of the division clustering, the Euclidean distance is used as a clustering index, the number of the initial clustering centers needs to be manually specified, and the initial clustering centers are divided according to the distance from each sample to the clustering center; meanwhile, when the k-means clustering algorithm is applied to scene reduction, the retention degree of the clustered scene set to the distribution condition of the initial scene set is also considered in addition to the representativeness of the clustered scene. Therefore, the traditional k-means clustering algorithm needs to be improved, so that the clustering result reflects the distribution condition of the original scene set, and the method is specifically realized as follows:
combining the process of specifying the initial clustering center number with the retention degree of the distribution condition, introducing a relative entropy index to reflect the change of the distribution condition before and after clustering, and selecting and optimizing the clustering center from the distribution condition and the representativeness together with the conventional error square sum index. Relative entropy can be used to measure the difference between the probability distributions of two discrete random variables, and the formula for relative entropy is as follows:
the formula quantifies the uncertainty level under the distributions p (x) and q (x) for easy visual comparison, which can be directly understood as the uncertainty loss caused by using the distribution q (x) instead of p (x). The specific application here is: and the random variable is a power curve scene set, the reduced scene distribution is used as Q (x) in the formula, the scene distribution before reduction is used as P (x) in the formula, the difference degree of the scene set distribution before and after reduction is obtained after the scene distribution before and after reduction is substituted into a relative entropy formula, and the retention degree of the uncertainty of the original scene set after reduction is obtained.
Clustering the wind power curve scene set generated and integrated by the previous part, generating and integrating n-100 electric load daily random curves by the MCS, wherein each load curve comprises T-24 moment load data, and forming an n-T dimension electric load random curve matrix. The main steps of clustering and analyzing the load curve are as follows:
1) and (3) specifying the number of initial centroids from k to 3, randomly determining the initial positions of the centroids by MATLAB, and generating an n-T-dimensional centroid curve matrix C.
C＝[c1,c2...ck]T
2) Determining the centroid situation of the object by calculating the Euclidean distance from each power curve to each clustering centroid curve, wherein the high-dimensional Euclidean distance calculation formula of the curves is as follows, W is an original scene curve matrix, C is a clustering centroid curve matrix, i is a curve subscript in the j-th class, i belongs to n, and j belongs to k.
3) Updating the position of the centroid, and determining new centroid curves of various classes according to a mean value formula, wherein njThe number of curves in class j, and i is the subscript of the curve in class j.
4) And iterating back to the first step until the position of the centroid curve basically does not change any more, and finishing the kmeans clustering under the current centroid number.
5) And according to the current clustering result, carrying out validity check on the representativeness and the distribution degree of the clustering centroid curve, and further selecting the most appropriate clustering curve as a scene reduction result. Using the sum of squares of errors index ISSEA representative formula to verify centroid is as follows, where d (c)j,wi) Is the Euclidean distance from the class-like curve to the class-like centroid curve, ISSERepresenting the sum of squared euclidean distances from the subclasses to the cluster centroid of the cluster in which they are located:
the relative entropy index introduced above is adopted to represent the retention degree of the subtracted scene (cluster centroid) to the original scene set distribution. As the number of clusters increases, the error square sum index and the relative entropy index are reduced, which means that the representativeness and the distribution degree of the centroid curve are better, but when the two indexes are too small, the cluster is too large and meaningless, so the number near the inflection point is taken as the optimal clustering number, and the obtained clustering centroid curve is optimal. FIG. 7 is a graph showing a variation curve of a relative entropy index and an error square sum index in a wind power curve clustering process, wherein a clustering number 10 at an inflection point of an index curve is taken, a clustering result can be representative, and an original distribution condition can be reflected.
The clustered wind power curve scene set is shown in fig. 8.
Compared with the traditional single scene reduction algorithm, the k-means serving as the unsupervised learning algorithm for machine learning has higher calculation efficiency and more representativeness of the reserved random scene.
The k-means clustering algorithm is improved by introducing the relative entropy index, so that the reserved scene is representative, and the reservation degree of the original scene distribution can be reflected better.
Based on the same inventive concept, the present embodiment is directed to a computing device, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the specific steps of the method.
Based on the same inventive concept, the present embodiment is directed to a computer-readable storage medium.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method.
Based on the same inventive concept, the embodiment discloses a wind power scene generation system based on probability distribution, which comprises:
a prediction error acquisition module configured to: acquiring historical data of a predicted value and an actual measurement value of the wind power, and extracting a difference value between the predicted value and the actual measurement value of the wind power, namely a prediction error;
a prediction error probability density function acquisition module configured to: selecting wind speed, wind direction and fitting errors which directly affect wind power, and respectively fitting the distribution of prediction errors under the influence of the wind speed, the wind direction and the fitting errors;
combining the prediction error distributions under the three types of influence factors according to corresponding weight parameters, and solving the optimal solution of the weight by taking the minimum sum of the squares of the residuals of the probability density function and the distribution histogram as a target function to obtain a prediction error probability density function;
a representative scene acquisition module configured to: discretizing the probability density function to obtain a large number of random scenes and selecting typical scenes.
Aiming at the fact that the object oriented to the prior art is a plurality of wind power plants, the matrix provided by the prior art considers the influence factors among the wind power plants, the influence of the space is ignored, the single wind power plant is analyzed, and the influence factors of the nature and the technology are mainly considered. In addition, when probability analysis is carried out, the prediction error in historical data is expanded, and a typical scene at a future moment is obtained through describing the probability distribution of the prediction error.
The steps involved in the apparatus of the above embodiment correspond to the first embodiment of the method, and the detailed description thereof can be found in the relevant description of the first embodiment. The term "computer-readable storage medium" should be taken to include a single medium or multiple media containing one or more sets of instructions; it should also be understood to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by a processor and that cause the processor to perform any of the methods of the present disclosure.
Those skilled in the art will appreciate that the modules or steps of the present disclosure described above can be implemented using general purpose computer means, or alternatively, they can be implemented using program code executable by computing means, whereby the modules or steps may be stored in memory means for execution by the computing means, or separately fabricated into individual integrated circuit modules, or multiple modules or steps thereof may be fabricated into a single integrated circuit module. The present disclosure is not limited to any specific combination of hardware and software.
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 scene generation method based on probability distribution is characterized by comprising the following steps:
acquiring historical data of a predicted value and an actual measurement value of the wind power, and extracting a difference value between the predicted value and the actual measurement value of the wind power, namely a prediction error;
selecting wind speed, wind direction and fitting errors which directly affect wind power, and respectively fitting the distribution of prediction errors under the influence of the wind speed, the wind direction and the fitting errors;
combining the prediction error distributions under the three types of influence factors according to corresponding weight parameters, and solving the optimal solution of the weight by taking the minimum sum of the squares of the residuals of the probability density function and the distribution histogram as a target function to obtain a prediction error probability density function;
discretizing the probability density function to obtain a large number of random scenes and selecting typical scenes.
2. The method according to claim 1, wherein in the wind power prediction of the day ahead, the predicted value of the wind power is obtained by substituting the predicted value of the wind speed into the fan output curve.
3. The method according to claim 1, wherein the distribution of prediction errors under the influence of wind speed is fitted: and fitting the error distribution under each wind speed section, wherein two sections with small head-to-tail error fluctuation of the wind speed are ignored.
4. The method according to claim 1, wherein the distribution of prediction errors under the influence of wind direction is fitted: and (3) directly estimating the probability density of the selected sample point by using a K nearest neighbor method in non-parameter estimation without assuming the form of the probability density.
5. The method for generating the wind power scene based on the probability distribution as claimed in claim 1, wherein the distribution of the prediction error under the influence of the fitting error is as follows:
the process of obtaining the fitted fan output curve through the predicted error distribution under the actually measured wind speed has a power characteristic fitting error, and the predicted error distribution under the power fitting error is shown through extreme value distribution.
6. The method according to claim 1, wherein the probability distribution-based wind power scene generation method is characterized in that a prediction error probability density function is obtained by fitting, source load power point prediction values obtained by day-ahead prediction, wind power, photovoltaic and load output are used as uncertain random variables;
and discretizing the probability density function of the random variable by random sampling by adopting a Monte Carlo simulation method to obtain a large number of random scenes, thereby determining the random parameters in an uncertain way.
7. The method as claimed in claim 1, wherein a typical scene is selected from a large number of random scenes by using a k-means cluster analysis algorithm:
the k-means clustering algorithm is improved by introducing the relative entropy index, so that the reserved scene is representative, and the reservation degree of the original scene distribution can be reflected better.
8. A wind power scene generation system based on probability distribution is characterized by comprising the following components:
a prediction error acquisition module configured to: acquiring historical data of a predicted value and an actual measurement value of the wind power, and extracting a difference value between the predicted value and the actual measurement value of the wind power, namely a prediction error;
a prediction error probability density function acquisition module configured to: selecting wind speed, wind direction and fitting errors which directly affect wind power, and respectively fitting the distribution of prediction errors under the influence of the wind speed, the wind direction and the fitting errors;
combining the prediction error distributions under the three types of influence factors according to corresponding weight parameters, and solving the optimal solution of the weight by taking the minimum sum of the squares of the residuals of the probability density function and the distribution histogram as a target function to obtain a prediction error probability density function;
a representative scene acquisition module configured to: discretizing the probability density function to obtain a large number of random scenes and selecting typical scenes.
9. A computing device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program performs the steps of the method of any of claims 1 to 7.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, is adapted to carry out the steps of the method according to any one of the preceding claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011064892.3A CN112186761B (en) | 2020-09-30 | 2020-09-30 | Wind power scene generation method and system based on probability distribution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011064892.3A CN112186761B (en) | 2020-09-30 | 2020-09-30 | Wind power scene generation method and system based on probability distribution |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112186761A true CN112186761A (en) | 2021-01-05 |
| CN112186761B CN112186761B (en) | 2022-03-01 |
Family
ID=73948395
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011064892.3A Active CN112186761B (en) | 2020-09-30 | 2020-09-30 | Wind power scene generation method and system based on probability distribution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112186761B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113095542A (en) * | 2021-03-01 | 2021-07-09 | 华中科技大学 | Photovoltaic output power prediction error fitting method and system based on DPMM |
| CN113468811A (en) * | 2021-07-06 | 2021-10-01 | 国网陕西省电力公司 | Power grid reserve capacity probabilistic dynamic evaluation method, system, terminal and readable storage medium containing new energy unit |
| CN113610575A (en) * | 2021-08-06 | 2021-11-05 | 浙江吉利控股集团有限公司 | Product sales prediction method and prediction system |
| CN113890109A (en) * | 2021-09-05 | 2022-01-04 | 三峡大学 | Multi-wind farm power day scene generation method with time-space correlation |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130013233A1 (en) * | 2010-03-19 | 2013-01-10 | Yoshiki Murakami | Electric-power-generation level predicting apparatus, method and program |
| CN103955779A (en) * | 2014-05-23 | 2014-07-30 | 武汉大学 | Wind power climbing event probability scene prediction method |
| CN108133279A (en) * | 2017-08-29 | 2018-06-08 | 甘肃省电力公司风电技术中心 | Wind power probability forecasting method, storage medium and equipment |
| CN108233357A (en) * | 2016-12-15 | 2018-06-29 | 中国电力科学研究院 | Wind-powered electricity generation based on nonparametric probabilistic forecasting and risk expectation dissolves optimization method a few days ago |
| CN110807554A (en) * | 2019-10-31 | 2020-02-18 | 合肥工业大学 | Generation method and system based on wind power/photovoltaic classical scene set |
| CN111600300A (en) * | 2020-05-21 | 2020-08-28 | 云南电网有限责任公司大理供电局 | Robust optimization scheduling method considering wind power multivariate correlation ellipsoid set |
-
2020
- 2020-09-30 CN CN202011064892.3A patent/CN112186761B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130013233A1 (en) * | 2010-03-19 | 2013-01-10 | Yoshiki Murakami | Electric-power-generation level predicting apparatus, method and program |
| CN103955779A (en) * | 2014-05-23 | 2014-07-30 | 武汉大学 | Wind power climbing event probability scene prediction method |
| CN108233357A (en) * | 2016-12-15 | 2018-06-29 | 中国电力科学研究院 | Wind-powered electricity generation based on nonparametric probabilistic forecasting and risk expectation dissolves optimization method a few days ago |
| CN108133279A (en) * | 2017-08-29 | 2018-06-08 | 甘肃省电力公司风电技术中心 | Wind power probability forecasting method, storage medium and equipment |
| CN110807554A (en) * | 2019-10-31 | 2020-02-18 | 合肥工业大学 | Generation method and system based on wind power/photovoltaic classical scene set |
| CN111600300A (en) * | 2020-05-21 | 2020-08-28 | 云南电网有限责任公司大理供电局 | Robust optimization scheduling method considering wind power multivariate correlation ellipsoid set |
Non-Patent Citations (3)
| Title |
|---|
| LEIJIAO GE 等: "Modeling Daily Load Profiles of Distribution Network for Scenario Generation Using Flow-Based Generative Network", 《IEEE ACCESS》 * |
| 吴丽珍 等: "基于最优场景生成算法的主动配电网无功优化", 《电力系统保护与控制》 * |
| 白凯峰 等: "融合风光出力场景生成的多能互补微网系统优化配置", 《电力系统自动化》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113095542A (en) * | 2021-03-01 | 2021-07-09 | 华中科技大学 | Photovoltaic output power prediction error fitting method and system based on DPMM |
| CN113095542B (en) * | 2021-03-01 | 2023-11-14 | 华中科技大学 | Fitting method and system for photovoltaic output power prediction error based on DPMM |
| CN113468811A (en) * | 2021-07-06 | 2021-10-01 | 国网陕西省电力公司 | Power grid reserve capacity probabilistic dynamic evaluation method, system, terminal and readable storage medium containing new energy unit |
| CN113468811B (en) * | 2021-07-06 | 2024-03-08 | 国网陕西省电力公司 | Power grid reserve capacity probabilistic dynamic assessment method and system containing new energy unit |
| CN113610575A (en) * | 2021-08-06 | 2021-11-05 | 浙江吉利控股集团有限公司 | Product sales prediction method and prediction system |
| CN113610575B (en) * | 2021-08-06 | 2024-03-08 | 浙江吉利控股集团有限公司 | Product sales prediction method and prediction system |
| CN113890109A (en) * | 2021-09-05 | 2022-01-04 | 三峡大学 | Multi-wind farm power day scene generation method with time-space correlation |
| CN113890109B (en) * | 2021-09-05 | 2023-08-25 | 三峡大学 | Multi-wind power plant power daily scene generation method with time-space correlation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112186761B (en) | 2022-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112186761B (en) | Wind power scene generation method and system based on probability distribution | |
| CN107766990B (en) | Method for predicting power generation power of photovoltaic power station | |
| CN107909211B (en) | Wind field equivalent modeling and optimization control method based on fuzzy c-means clustering algorithm | |
| CN108376262B (en) | Analytical model construction method for typical characteristics of wind power output | |
| CN112381137B (en) | New energy power system reliability assessment method, device, equipment and storage medium | |
| CN105701572B (en) | Photovoltaic short-term output prediction method based on improved Gaussian process regression | |
| CN107679687A (en) | A kind of photovoltaic output modeling method and Generation System Reliability appraisal procedure | |
| CN116402203A (en) | Method, system and medium for predicting short-time photovoltaic power generation capacity considering weather conditions | |
| CN115173465A (en) | Wind, light, water, fire and storage integrated coupling mechanism analysis method based on Copula theory | |
| CN113033136B (en) | Simplified photovoltaic cell physical parameter extraction optimization method and system | |
| CN109272258B (en) | Regional wind and solar power generation resource evaluation method based on K-means clustering | |
| CN110298494A (en) | A kind of wind power forecasting method based on Segment Clustering and Combinatorial Optimization | |
| CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
| CN112651576A (en) | Long-term wind power prediction method and device | |
| CN112633565A (en) | Photovoltaic power aggregation interval prediction method | |
| CN111815039A (en) | Weekly scale wind power probability prediction method and system based on weather classification | |
| CN109858667A (en) | It is a kind of based on thunder and lightning weather to the short term clustering method of loading effects | |
| CN114298132A (en) | Wind power prediction method and device and electronic equipment | |
| CN117993611A (en) | Flexible heat source new energy consumption capability assessment method based on scene time sequence | |
| CN110991743B (en) | Wind power short-term combination prediction method based on cluster analysis and neural network optimization | |
| CN108694475A (en) | Short-term time scale photovoltaic cell capable of generating power amount prediction technique based on mixed model | |
| CN102539823A (en) | Method for forecasting wind speed distribution of WTG (wind turbine generator) | |
| CN110097243A (en) | Method and device for determining representative wind generating set in wind power plant | |
| CN116522800A (en) | Multi-target wind-power storage station site selection and volume determination method based on scene probability | |
| CN111178601A (en) | Wind turbine generator power prediction method based on meteorological data post-processing |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |