CN116128211A - Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene - Google Patents
Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene Download PDFInfo
- Publication number
- CN116128211A CN116128211A CN202211612301.0A CN202211612301A CN116128211A CN 116128211 A CN116128211 A CN 116128211A CN 202211612301 A CN202211612301 A CN 202211612301A CN 116128211 A CN116128211 A CN 116128211A
- Authority
- CN
- China
- Prior art keywords
- wind
- light
- scene
- value
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
Abstract
The invention discloses a wind-light-water combined short-term optimization scheduling method based on a wind-light uncertainty prediction scene, which comprises the steps of preprocessing historical wind-light output data and dividing the hierarchy; building a wind and light forecasting scene building model and a reduction model to generate a wind and light forecasting typical scene set; establishing a scene migration probability calculation model to obtain migration probabilities among different periods of the actual value of the wind-light output; according to a typical scene set of wind and light forecast, combining real-time wind and light output measured data and scene migration probability information at each time to update a forecast scene; and carrying out wind-light-water combined short-term optimized scheduling. Based on the prediction and actual measurement data of the historical wind-light output, the invention provides a new method and technology for optimizing and dispatching a multi-energy complementary system connected with new energy by constructing a wind-light output prediction scene and considering the influence of uncertainty on wind-light-water combined short-term optimizing and dispatching.
Description
Technical Field
The invention relates to a wind-light-water combined short-term optimization scheduling method considering wind-light uncertainty, in particular to a wind-light-water combined short-term optimization scheduling method based on a wind-light uncertainty forecasting scene.
Background
The renewable energy is green low-carbon energy, is an important component of a multi-wheel driving energy supply system in China, and has important significance for improving energy structure, protecting ecological environment, coping with climate change and realizing sustainable development of economy and society. In recent decades, renewable energy sources become the main body of new power generation and installation machines in China, the duty ratio of the increment of the renewable energy sources in the increment of the power consumption of the whole society in China is more than 50%, and the power generation of wind power and solar energy is doubled. However, the uncertainty exists in wind-light and photovoltaic power generation, the direct connection brings great impact to a power grid, but the characteristics of flexible power output of water and electricity are utilized, wind-light fluctuation can be stabilized to a certain extent by bundling and sending wind-light and water, and the influence brought by the uncertainty is reduced.
At present, for wind-light-water combined short-term optimization scheduling methods considering wind-light uncertainty, students at home and abroad mainly concentrate on modeling methods based on Bayesian theory and fuzzy mathematic theory. According to Liu Qiaobo (2018), fitting the distribution of the wind power and photovoltaic prediction errors under different output levels based on historical data, introducing the relevant coefficients of the sporman to describe the complementarity of the wind power prediction errors in seasons, and establishing a wind power and wind power combined prediction error distribution model; zhang Juntao (2020) based on a random programming theory, adopting coupling quantile regression to convert historical statistical information from a deterministic prediction sequence into a scene set; xu Yechi (2022) based on a wind power randomness analysis of the prediction error distribution, a random optimized scheduling model is built that considers the frequency response by generating a wind power output scenario describing wind power randomness. However, the above methods describe individual scenes or multiple independent scenes which are not related to each other based on historical data for making an optimal scheduling scheme, and cannot consider the correlation between scenes and the optimization of the actual measurement values updated at any time period to the scheduling system, so that the accuracy of describing the wind-solar uncertainty will affect the making of the optimal scheduling scheme, and finally cause the deviation of the scheduling targets.
Disclosure of Invention
The invention aims to: aiming at the problems and defects existing in the prior art, the invention provides a wind-light-water combined short-term optimization scheduling method based on a wind-light uncertainty prediction scene. Based on the prediction and actual measurement data of the historical wind-light output, a wind-light scene prediction model, a scene reduction model and a scene migration probability calculation model which are constructed based on scenes are established, and a wind-light prediction typical scene set and a migration probability coefficient thereof are updated time by time, so that an updating optimization scheduling decision is calculated and updated time by time, and wind-light-water combined short-term optimization scheduling is performed.
The technical scheme is as follows: a wind-light-water combined short-term optimization scheduling method based on a wind-light uncertainty prediction scene comprises the following steps:
s1, selecting historical wind-solar power output data, processing a missing value and an abnormal value, normalizing the processed data, and dividing sample levels;
s2, establishing a scene-based wind-light scene prediction model, analyzing sample characteristics in the layers by adopting sample data of each layer, and performing wind-light scene prediction by using analysis results based on wind-light output day-ahead prediction data to obtain a wind-light prediction scene set;
s3, establishing a scene reduction model, performing scene reduction on an initial wind and light forecasting scene set, and iteratively calculating and selecting a representative wind and light forecasting typical scene to form the wind and light forecasting typical scene set;
s4, establishing a scene migration probability calculation model based on the wind-solar power output history actual measurement value, dividing intervals of the actual measurement values, and calculating migration probability among the actual measurement values with time correlation;
and S5, updating the optimal scheduling decision time by time based on the measured data of wind and light output time by time according to the result of wind and light forecasting typical scene concentration, and carrying out wind and light and water combined short-term optimal scheduling according to the principle of best source load matching degree.
Further, the step S1 specifically includes: the historical wind and light output data mainly comprises day-ahead forecast wind output data and actual measurement wind output data with consistent time-space correlation, and day-ahead forecast photovoltaic output data and actual measurement photovoltaic output data with consistent time-space correlation; the processing method for the missing value comprises the following steps: performing linear interpolation according to two adjacent groups of data to perform data supplementation; the processing method for the abnormal value comprises the following steps: firstly, deleting an abnormal value, and supplementing the deleted data according to a method of deleting the value; sample normalization is carried out on the processed data, and the sample normalization method comprises the following steps:
s′＝(s-μ)/σ
wherein s' is a normalized sample value; s is sample original data; mu is the average value of the original set of samples; σ is the standard deviation of the original set of samples.
Further, the wind-light scene prediction model structure based on the scene construction established in the step S2 is as follows:
based on the forecast and actual measurement data set obtained after sample normalization processing, grading the data set according to the numerical value of the forecast value, and calculating the characteristic value of the data in each level, wherein the characteristic value calculating method comprises the following steps:
wherein ε' i,j A j-th sample relative error value within the i-th hierarchy; sp'. i,j Normalized values for the predicted samples; st' i,j Is a normalized value of the measured sample.
In the method, in the process of the invention,
is a first feature value of the hierarchy; n is n i Is the number of samples relative to the error value for the i-th level.
In the method, in the process of the invention,
is the second eigenvalue of the hierarchy.
Furthermore, based on two characteristic values in the sample level, a wind and light output forecast scene set is generated by adopting super Latin square sampling.
Further, the step S3 specifically includes: and (3) performing scene reduction by adopting a K-means random centroid clustering method, wherein the clustering result is controlled between 4 and 8.
Further, the step S4 specifically includes: dividing intervals according to the normalization result of the wind-light output actual measurement data, and establishing a scene migration probability calculation model, wherein the probability calculation method comprises the following steps:
st″＝(st′-μ′)/σ′
wherein st' represents a standard fraction of a standard normal distribution; mu' represents the average value of the normalized value of the measured sample; σ' represents the standard deviation of the normalized value of the measured sample.
Wherein Pst' (u,d) And (3) representing the probability of st' after normalization of the next measured value when the current measured value is located in the (u, d) interval after normalization.
Further, the step S5 specifically includes: based on the result in the wind-light forecast typical scene set, calculating the occurrence probability of each typical scene one by one, and continuously generating a decision scene for short-term optimization scheduling calculation, wherein the calculation method comprises the following steps:
wherein P' q And forecasting the migration probability coefficient of the q-th scene in the typical scene set for the wind and light.
Wherein x 'is' p The wind and light forecast scene value of the p-th hour in the optimal scheduling period is represented; x is x p,q Representing the value of the p hour in the w-th scene in the wind and light forecast typical scene values.
Furthermore, a wind-light-water combined short-term optimization scheduling model is constructed by taking the optimal matching degree of the source load as a target, and wind-light-water combined short-term optimization scheduling considering wind-light uncertainty is realized based on a continuously generated decision scene of short-term optimization scheduling, and the calculation method is as follows:
Nh p ＝N w +N s +N h
wherein N is v The matching rate is the source load; ny p Total system output for the p-th hour; nh (Nh) p System load for the p-th hour; n (N) w 、N s 、N h The output values of wind power, photoelectricity and hydropower are respectively;
the power generation flow rate is the j-th power generation flow rate of the hydropower station; h j Is the power generation water head at the j-th time of the hydropower station.
A wind-light-water combined short-term optimization scheduling system based on a wind-light uncertainty forecast scene comprises:
the data processing module is used for carrying out hierarchical division on the input historical wind-solar power output data, carrying out characteristic value calculation on the historical data of each hierarchical level and generating scenes;
the scene generation module is used for establishing a wind-solar scene prediction model and a scene reduction model, generating a scene set by adopting characteristic values of each level sample based on day-ahead prediction data, reducing the scene set by using iterative clustering, and outputting a wind-solar prediction typical scene set;
the scene migration probability calculation module is used for calculating migration probability among actual measurement values of the existing time correlation and outputting migration probability coefficients of all scenes in the wind-light forecast typical scene set;
the optimization scheduling module is used for making a wind-light-water combined optimization scheduling plan, and performing wind-light-water combined short-term optimization scheduling calculation according to the principle of best source load matching degree based on a wind-light prediction typical scene set and each scene migration probability coefficient.
The computer equipment comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the wind-light-water joint short-term optimization scheduling method based on the wind-light uncertainty forecast scene is realized when the processor executes the computer program.
A computer readable storage medium storing a computer program for executing a wind-light-water joint short-term optimization scheduling method based on a wind-light uncertainty prediction scenario as described above.
The beneficial effects are that: compared with the prior art, the invention has the technical effects that: (1) According to the method, based on the prediction and actual measurement data of historical wind-light output, a wind-light scene prediction model, a scene reduction model and a scene migration probability calculation model which are constructed based on scenes are established, a wind-light typical prediction scene set is generated, a wind-light-water joint scheduling plan is formulated, and a novel wind-light-water joint short-term optimization scheduling method considering uncertainty is provided accordingly; (2) By updating the wind and light forecast typical scene set and the migration probability coefficient thereof time by time, the optimal scheduling decision can be calculated and updated time by time. The data required by the uncertain information extraction part of the method are all historical data, and a PYTHON program can be written to cope with wind-light-water joint short-term optimization scheduling calculation under the condition of different input data of any scheduling period; and (3) the principle is simple, the operation is simple, convenient and flexible, and the implementation is easy. The technical method carries out optimal scheduling calculation based on scene construction and reduction and a scene migration probability model, and has high calculation speed and short response time; (4) Support can be provided for the access of other subsequent renewable energy sources; the technology provides a new method and technology for optimizing and dispatching the multi-energy complementary system connected with the new energy, and lays a good foundation for absorbing the new energy of the multi-energy complementary system, reducing the power grid impact and optimizing the energy structure.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
fig. 2 is a block diagram of the system architecture of the present invention.
Detailed Description
The present invention is further illustrated below in conjunction with specific embodiments, it being understood that these embodiments are meant to be illustrative of the invention only and not limiting the scope of the invention, and that modifications of the invention, which are equivalent to those skilled in the art to which the invention pertains, will fall within the scope of the invention as defined in the claims appended hereto.
According to the wind-light-water combined short-term optimization scheduling method, based on the prediction and actual measurement data of the historical wind-light output, a wind-light scene prediction model, a scene reduction model and a scene migration probability calculation model which are constructed based on scenes are established, and a wind-light prediction typical scene set and a migration probability coefficient thereof are updated time by time, so that an optimization scheduling decision is calculated and updated time by time, and wind-light-water combined short-term optimization scheduling is performed.
As shown in fig. 1, the wind-light-water combined short-term optimization scheduling method based on the wind-light uncertainty prediction scene comprises the following steps:
s1, selecting historical wind-light output data, processing a missing value and an abnormal value, carrying out sample normalization on the processed data, and dividing sample levels, wherein the method comprises the following specific steps of:
s1.1, firstly screening historical wind-solar power output data, selecting relevant wind power stations and atmospheric environment measuring point data near a photovoltaic power station, calculating, controlling the data precision to be 1-15min, and simultaneously guaranteeing the overall consistency of data samples so as to ensure the time-space correlation of the data samples.
S1.2, after historical wind and light output prediction and actual measurement data are obtained through screening, missing values and abnormal values are required to be processed, negative values and data exceeding the measuring instrument range cannot appear in wind and light output data, and particularly, the measured value of photovoltaic output data before rising of the sun is always zero, so that the selected historical data are required to be complete and relatively reasonable wind and light output prediction and actual measurement data.
S1.3, carrying out normalization processing on the processed sample, wherein the sample normalization method comprises the following steps:
s′＝(s-μ)/σ
wherein s' is a normalized sample value; s is sample original data; mu is the average value of the original set of samples; σ is the standard deviation of the original set of samples.
S2, establishing a scene-based wind-light scene prediction model, carrying out sample feature analysis in the layers by adopting sample data of each layer, and carrying out wind-light scene prediction by using analysis results based on wind-light output day-ahead prediction data to obtain a wind-light prediction scene set, wherein the specific steps are as follows:
s2.1, calculating the forecast error of the forecast value of each time point in the data set, carrying out hierarchical division according to the statistical result of the forecast error, and carrying out hierarchical division by adopting the error value or carrying out uniform division according to the number of error samples to determine the upper and lower boundaries of the samples in the hierarchy for generating scenes by subsequent sampling.
S2.2, calculating the characteristic value of the data in each level, wherein the characteristic value calculating method comprises the following steps:
wherein ε' i,j A j-th sample relative error value within the i-th hierarchy; sp'. i,j Normalized values for the predicted samples; st' i,j Is a normalized value of the measured sample.
In the method, in the process of the invention,
is a first feature value of the hierarchy; n is n i Sample phase for the ith hierarchyFor the number of error values.
In the method, in the process of the invention,
is the second eigenvalue of the hierarchy.
S2.3, based on two characteristic values in the sample level, sampling by using a super Latin square to generate a wind and light output forecast scene set. The super latin square sampling (LHS) is to accurately establish an input distribution by sampling with fewer iterations, and compared with monte carlo, the key is to layer the input probability and establish intervals with equal cumulative probability scale, and the sampling result is forcedly represented by the value of each interval by sampling from each interval or layer of the input distribution, so that the input probability distribution is forcedly reconstructed. The generation strategy of the wind-light and photovoltaic forecast scene set is as follows: the whole sampling process adopts the principle of 'sampling not replacing', and the layering number of the accumulated distribution is the same as the iteration number in the whole execution process. It should be noted that when sampling is performed using this method, it is necessary to maintain independence between variables, and the maintenance of independence is achieved by randomly selecting a sampling interval for each variable, so that unintentional correlations between variables can be effectively avoided.
S3, a scene reduction model is established, scene reduction is carried out on an initial wind and light forecasting scene set, and the most representative wind and light forecasting typical scene is selected through iterative calculation to form the wind and light forecasting typical scene set.
The K-means algorithm is used as a basic algorithm in clustering, belongs to an algorithm in unsupervised learning, and has the basic principle that K initial points are randomly determined as cluster centroids, then the calculated distance between each point in sample data and the cluster centroids is calculated, the samples are classified according to the calculated distance, and a better clustering result can be obtained by continuously iterating the positions of the cluster centroids.
The method comprises the following specific steps:
s3.1, according to an initial wind and light forecast scene set, 4-8 initial centroids are selected, namely the typical cluster centroids.
S3.2, calculating the distance between other elements in the whole scene set and the centroid of the typical cluster, wherein the distance is calculated by adopting Euclidean distance.
S3.3, selecting a new cluster centroid, calculating the classification condition of the whole scene set under the new cluster centroid, and guiding the convergence of the classification result of the whole scene set by continuously repeating S3.2-S3.3, so as to indicate the end of clustering.
S4, establishing a scene migration probability calculation model based on the wind-solar power historical actual measurement value, dividing intervals of the actual measurement values, and calculating migration probability among the actual measurement values with time correlation, wherein the method comprises the following specific steps of:
s4.1, calculating normal distribution standard scores of corresponding conditions of each scene in the typical scene set, wherein the calculation method is as follows:
st″＝(st′-μ′)/σ′
wherein st' represents a standard fraction of a standard normal distribution; mu' represents the average value of the normalized value of the measured sample; σ' represents the standard deviation of the normalized value of the measured sample.
S4.2, dividing intervals according to the normalization result of the wind-solar actual measurement data, and establishing a scene migration probability calculation model, wherein the probability calculation method comprises the following steps:
wherein Pst' (u,d) And (3) representing the probability of st' after normalization of the next measured value when the current measured value is located in the (u, d) interval after normalization.
S5, according to the result of wind and light forecasting typical scene concentration, based on wind and light output time-by-time measured data, updating an optimal scheduling decision time by time, and carrying out wind and light and water combined short-term optimal scheduling according to the principle of best source load matching degree, wherein the specific steps are as follows:
s5.1, calculating migration probability coefficients of each scene based on migration probability values of each scene in a wind-light prediction typical scene set, wherein the calculation method is as follows:
wherein P' q And forecasting the migration probability coefficient of the q-th scene in the typical scene set for the wind and light.
S5.2, generating a calculation scene for short-term optimization scheduling based on migration probability coefficients of all scenes, wherein the calculation method comprises the following steps:
wherein x 'is' p The wind and light forecast scene value of the p-th hour in the optimal scheduling period is represented; x is x p,q Representing the value of the p hour in the q-th scene in the typical scene values of the wind and light forecast.
S5.3, constructing a wind-light-water combined short-term optimization scheduling model by taking the optimal source-load matching degree as a target, and realizing wind-light-water combined short-term optimization scheduling by considering wind-light uncertainty based on a continuously generated decision scene of short-term optimization scheduling, wherein the calculation method comprises the following steps:
Nh p ＝N w +N s +N h
wherein N is v The matching rate is the source load; ny p Total system output for the p-th hour; nh (Nh) p System load for the p-th hour; n (N) w 、N s 、N h The output values of wind power, photoelectricity and hydropower are respectively;
the power generation flow rate is the j-th power generation flow rate of the hydropower station; h j Is the power generation water head at the j-th time of the hydropower station.
According to the wind-light-water combined short-term optimization scheduling method, based on the prediction and actual measurement data of the historical wind-light output, a wind-light scene prediction model, a scene reduction model and a scene migration probability calculation model which are constructed based on scenes are established, and a wind-light prediction typical scene set and a migration probability coefficient thereof are updated time by time, so that an optimization scheduling decision is calculated and updated time by time, and wind-light-water combined short-term optimization scheduling is performed. The technology provides a new method and technology for optimizing and dispatching the multi-energy complementary system connected with the new energy, and lays a good foundation for absorbing the new energy of the multi-energy complementary system, reducing the power grid impact and optimizing the energy structure.
As shown in fig. 2, the wind-light-water combined short-term optimization scheduling system based on the wind-light uncertainty prediction scene comprises:
the data processing module is used for carrying out hierarchical division on input data, carrying out characteristic value calculation on historical data of each hierarchy and generating scenes;
the scene generation module is used for establishing a wind-solar scene prediction model and a scene reduction model, generating a scene set by adopting characteristic values of each level sample based on day-ahead prediction data, reducing the scene set by using iterative clustering, and outputting a wind-solar prediction typical scene set;
the scene migration probability calculation module is used for calculating migration probability among actual measurement values of the existing time correlation and outputting migration probability coefficients of all scenes in the wind-light forecast typical scene set;
the optimization scheduling module is used for making a wind-light-water combined optimization scheduling plan, and performing wind-light-water combined short-term optimization scheduling calculation according to the principle of best source load matching degree based on a wind-light prediction typical scene set and each scene migration probability coefficient.
The implementation process of the system is the same as the method and will not be described in detail.
It will be apparent to those skilled in the art that the steps of the wind-light-water joint short-term optimization scheduling method based on a wind-light uncertainty prediction scenario according to the above embodiments of the present invention may be implemented by general purpose computing devices, they may be centralized on a single computing device, or distributed over a network composed of multiple computing devices, or they may alternatively be implemented by program codes executable by computing devices, so that they may be stored in a storage device and executed by the computing devices, and in some cases, the steps shown or described may be executed in a different order from that herein, or they may be manufactured separately as individual integrated circuit modules, or a plurality of modules or steps in them may be manufactured as a single integrated circuit module. Thus, embodiments of the invention are not limited to any specific combination of hardware and software.
Claims (10)
1. A wind-light-water combined short-term optimization scheduling method based on a wind-light uncertainty prediction scene is characterized by comprising the following steps:
s1, selecting historical wind-solar power output data, processing a missing value and an abnormal value, normalizing the processed data, and dividing sample levels;
s2, establishing a scene-based wind-light scene prediction model, analyzing sample characteristics in the layers by adopting sample data of each layer, and performing wind-light scene prediction by using analysis results based on wind-light output day-ahead prediction data to obtain a wind-light prediction scene set;
s3, establishing a scene reduction model, performing scene reduction on an initial wind and light forecasting scene set, and iteratively calculating and selecting a representative wind and light forecasting typical scene to form the wind and light forecasting typical scene set;
s4, establishing a scene migration probability calculation model based on the wind-solar history actual measurement values, dividing intervals of the actual measurement values, and calculating migration probability among the actual measurement values with time correlation;
and S5, updating the optimal scheduling decision time by time based on the measured data of wind and light output time by time according to the result of wind and light forecasting typical scene concentration, and carrying out wind and light and water combined short-term optimal scheduling according to the principle of best source load matching degree.
2. The wind-light-water combined short-term optimization scheduling method based on the wind-light uncertainty prediction scene of claim 1, wherein in S1, the historical wind-light output data comprises: the solar forecast wind output and the actually measured wind output data with consistent time-space correlation, and the solar forecast photovoltaic output data and the actually measured photovoltaic output data with consistent time-space correlation; the processing method for the missing value comprises the following steps: performing linear interpolation according to two adjacent groups of data to perform data supplementation; the processing method for the abnormal value comprises the following steps: firstly, deleting an abnormal value, and supplementing the deleted data according to a method of deleting the value; the sample normalization method comprises the following steps:
s′＝(s-μ)/σ
wherein s' is a normalized sample value; s is sample original data; mu is the average value of the original set of samples; σ is the standard deviation of the original set of samples.
3. The wind-light-water joint short-term optimization scheduling method based on the wind-light uncertainty prediction scene according to claim 1, wherein in the step S2, the wind-light scene prediction model structure built based on the scene is:
based on the forecast and actual measurement data set obtained after sample normalization processing, grading the data set according to the numerical value of the forecast value, and calculating the characteristic value of the data in each level, wherein the characteristic value calculating method comprises the following steps:
wherein ε' i,j A j-th sample relative error value within the i-th hierarchy; sp'. i,j Normalized values for the predicted samples; st' i,j Normalization value of actual measurement sample;
in the method, in the process of the invention,
is a first feature value of the hierarchy; n is n i Is the number of samples relative to the error value for the i-th level.
In the method, in the process of the invention,
a second feature value for the hierarchy;
and based on two characteristic values in the sample level, sampling by using a super Latin square to generate a wind and light output forecast scene set.
4. The wind-light-water combined short-term optimization scheduling method based on the wind-light uncertainty prediction scene according to claim 1, wherein the scene reduction method adopted in the step S3 is a K-means clustering method, clustering is carried out by adopting a random centroid, and the clustering result is controlled between 4 and 8.
5. The wind-light-water combined short-term optimization scheduling method based on the wind-light uncertainty prediction scene according to claim 1, wherein in the step S4, interval division is performed according to a normalization result of wind-light actual measurement data, a scene migration probability calculation model is established, and the probability calculation method is as follows:
st″＝(st′-μ′)/σ′
wherein st' represents a standard fraction of a standard normal distribution; mu' represents the average value of the normalized value of the measured sample; σ' represents the standard deviation of the normalized value of the measured sample.
Wherein Pst' (u,d) And (3) representing the probability of st' after normalization of the next measured value when the current measured value is located in the (u, d) interval after normalization.
6. The wind-light-water joint short-term optimization scheduling method based on the wind-light uncertainty prediction scene according to claim 1, wherein in the step S5, based on the result in the wind-light prediction typical scene set, the probability of each typical scene occurrence is calculated each time, and the decision scene for short-term optimization scheduling calculation is continuously generated, and the calculation method is as follows:
wherein P' q The migration probability coefficient of the q-th scene in the typical scene set is forecasted for wind and light;
wherein x 'is' p The wind and light forecast scene value of the p-th hour in the optimal scheduling period is represented; x is x p,q Representing the value of the p hour in the q-th scene in the typical scene values of the wind and light forecast.
7. The wind-light-water combined short-term optimization scheduling method based on the wind-light uncertainty prediction scene according to claim 1, wherein in the step S5, a wind-light-water combined short-term optimization scheduling model is constructed by taking the optimal source-load matching degree as a target, and wind-light-water combined scheduling is performed based on a continuously generated decision scene of short-term optimization scheduling, and the calculation method is as follows:
Nh p ＝N w +N s +N h
wherein N is v The matching rate is the source load; ny p Total system output for the p-th hour; nh (Nh) p System load for the p-th hour; n (N) w 、N s 、N h : the output values of wind power, photoelectricity and hydropower are respectively;
the power generation flow rate is the j-th power generation flow rate of the hydropower station; h j Is the power generation water head at the j-th time of the hydropower station.
8. A wind-light-water combined short-term optimization scheduling system based on a wind-light uncertainty prediction scene is characterized by comprising the following steps:
the data processing module is used for carrying out hierarchical division on the input historical wind-solar power output data, carrying out characteristic value calculation on the historical data of each hierarchical level and generating scenes;
the scene generation module is used for establishing a wind-solar scene prediction model and a scene reduction model, generating a scene set by adopting characteristic values of each level sample based on day-ahead prediction data, reducing the scene set by using iterative clustering, and outputting a wind-solar prediction typical scene set;
the scene migration probability calculation module is used for calculating migration probability among actual measurement values of the existing time correlation and outputting migration probability coefficients of all scenes in the wind-light forecast typical scene set;
the optimization scheduling module is used for making a wind-light-water combined optimization scheduling plan, and performing wind-light-water combined short-term optimization scheduling calculation according to the principle of best source load matching degree based on a wind-light prediction typical scene set and each scene migration probability coefficient.
9. A computer device, characterized in that the computer device comprises a memory, a processor and a computer program stored on the memory and operable on the processor, the processor implementing a wind-light-water joint short-term optimization scheduling method based on a wind-light uncertainty prediction scenario according to any one of claims 1-7 when executing the computer program.
10. A computer readable storage medium, characterized in that it stores a computer program for executing a wind-light-water joint short-term optimization scheduling method based on a wind-light uncertainty prediction scenario according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211612301.0A CN116128211A (en) | 2022-12-15 | 2022-12-15 | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211612301.0A CN116128211A (en) | 2022-12-15 | 2022-12-15 | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116128211A true CN116128211A (en) | 2023-05-16 |
Family
ID=86309127
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211612301.0A Pending CN116128211A (en) | 2022-12-15 | 2022-12-15 | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116128211A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117154725A (en) * | 2023-10-31 | 2023-12-01 | 长江三峡集团实业发展(北京)有限公司 | Water-wind-solar multi-energy complementary scheduling method, device, computer equipment and medium |
-
2022
- 2022-12-15 CN CN202211612301.0A patent/CN116128211A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117154725A (en) * | 2023-10-31 | 2023-12-01 | 长江三峡集团实业发展(北京)有限公司 | Water-wind-solar multi-energy complementary scheduling method, device, computer equipment and medium |
| CN117154725B (en) * | 2023-10-31 | 2024-01-26 | 长江三峡集团实业发展(北京)有限公司 | Water-wind-solar multi-energy complementary scheduling method, device, computer equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102478584B (en) | Wind power station wind speed prediction method based on wavelet analysis and system thereof | |
| CN110929953A (en) | Photovoltaic power station ultra-short term output prediction method based on cluster analysis | |
| Ye et al. | Combined approach for short-term wind power forecasting based on wave division and Seq2Seq model using deep learning | |
| CN105512766A (en) | Wind power plant power predication method | |
| CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
| Chen et al. | Research on wind power prediction method based on convolutional neural network and genetic algorithm | |
| CN114021848A (en) | Generating capacity demand prediction method based on LSTM deep learning | |
| CN115759467A (en) | Time-division integrated learning photovoltaic prediction method for error correction | |
| CN116014722A (en) | Sub-solar photovoltaic power generation prediction method and system based on seasonal decomposition and convolution network | |
| CN116629416A (en) | Photovoltaic power station power prediction method and device | |
| CN116128211A (en) | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene | |
| Kader et al. | Active power control strategy for wind farms based on power prediction errors distribution considering regional data | |
| CN105205560A (en) | Photovoltaic power supply power prediction method based on positive and negative error variable weights | |
| Xu et al. | NWP feature selection and GCN-based ultra-short-term wind farm cluster power forecasting method | |
| Shen et al. | PV power interval prediction based on EEMD-LSTM method | |
| Huang et al. | Power Prediction Method of Distributed Photovoltaic Digital Twin System Based on GA-BP | |
| Fan et al. | Ultra-Short-Term Wind Power Forecasting Based on Fluctuation Pattern Clustering and Prediction | |
| Chen et al. | Short-Term Photovoltaic Power Prediction Algorithm Based on Improved BP Neural Network | |
| Wu et al. | Research on New Energy Probability Prediction Technology Based on Ensemble Weather Forecast | |
| Zhang et al. | PV Prediction Using Hierarchical BiLSTM-RFR Model Considering Meteorological Factors | |
| CN110874611B (en) | Wind power cluster power ultra-short term prediction error grade grading compensation method | |
| Sai et al. | Bus net Load Forecast Considering Photovoltaic Output Correlation | |
| BALBAL et al. | Time Series Forecasting on Solar Energy Production Data Using LSTM | |
| Yan et al. | Check for updates Multi-source-Load Prediction Based on Multi-task Learning | |
| Gao et al. | Hybrid Solar and Wind Power Prediction Based on Optimal Decomposition and WOA Optimization BiLSTM |
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 |