CN110197298A - The method and device of multi-model sun normal direction radiation prediction based on clustering algorithm - Google Patents
The method and device of multi-model sun normal direction radiation prediction based on clustering algorithm Download PDFInfo
- Publication number
- CN110197298A CN110197298A CN201910388405.XA CN201910388405A CN110197298A CN 110197298 A CN110197298 A CN 110197298A CN 201910388405 A CN201910388405 A CN 201910388405A CN 110197298 A CN110197298 A CN 110197298A
- Authority
- CN
- China
- Prior art keywords
- data
- meteorological
- meteorological data
- normal direction
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000005855 radiation Effects 0.000 title claims abstract description 47
- 238000001514 detection method Methods 0.000 claims description 41
- 238000010801 machine learning Methods 0.000 claims description 32
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 11
- 241001269238 Data Species 0.000 claims description 10
- 238000007689 inspection Methods 0.000 claims description 8
- 238000007477 logistic regression Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000012706 support-vector machine Methods 0.000 claims description 7
- 238000013480 data collection Methods 0.000 claims description 4
- 238000003064 k means clustering Methods 0.000 claims description 3
- 238000011161 development Methods 0.000 abstract description 5
- 238000010276 construction Methods 0.000 abstract description 2
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 230000000875 corresponding effect Effects 0.000 description 33
- 238000010248 power generation Methods 0.000 description 11
- 238000010606 normalization Methods 0.000 description 10
- 230000002596 correlated effect Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 210000005036 nerve Anatomy 0.000 description 4
- 238000011524 similarity measure Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000007418 data mining Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 239000000428 dust Substances 0.000 description 2
- 239000004576 sand Substances 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Marketing (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Development Economics (AREA)
- Primary Health Care (AREA)
- Entrepreneurship & Innovation (AREA)
- General Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention is based on the method and devices of the multi-model sun normal direction of clustering algorithm radiation prediction, belong to solar energy thermal-power-generating technical field, a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm of the present invention, include: step 110: collecting the history meteorological data of estimation range, the history meteorological data is the multi-class meteorological element for influencing solar energy normal direction and directly radiating;Step 120: the history meteorological data is standardized, the history meteorological data of standardization is clustered using clustering algorithm, and the result of cluster is grouped, each group of the grouping is each moment meteorology sample set comprising each meteorological element data of standardization;The prediction submodel for selecting real time meteorological data close, so that prediction has specific aim;The resources model more refined is the basis of solar energy thermal-power-generating development and project construction, is also beneficial to solar energy thermal-power-generating station and is preferably exerted advantages of oneself according to resource situation.
Description
Technical field
The invention belongs to solar energy thermal-power-generating technical fields, the multi-model sun normal direction radiation specially based on clustering algorithm
The method and device of prediction.
Background technique
Prediction for solar energy resources is the major issue of field of solar thermal power generation and time series field.Currently,
Realized that commercialized solar power generation form is mainly photovoltaic power generation and photo-thermal power generation.
Comparatively, earlier, therefore, the current resources for solar power generation are main for the development starting of photovoltaic power generation
The characteristics of for photovoltaic power generation and be unfolded, most of related works are directed to the prediction of total solar radiation.Photo-thermal power generation
Mainly directly radiated to normal direction related, the numerical value directly radiated compared with global radiation is easier to occur fluctuating and change, and not
Under same meteorological condition, such as: cloud layer is thicker, and cloud layer is relatively thin, dust and sand weather, fine day, and changing rule differs greatly prediction not
Accurately.
Summary of the invention
The present invention provides the method and devices of the multi-model sun normal direction radiation prediction based on clustering algorithm, it is therefore intended that
It solves the above problems, solves photo-thermal power generation and mainly directly radiated to normal direction related, the numerical value directly radiated compared with global radiation is more
It is easy to appear fluctuation and variation, and under different meteorological conditions, such as: cloud layer is thicker, and cloud layer is relatively thin, dust and sand weather, fine day,
Its changing rule differs greatly the true problem of forecasting inaccuracy.
To achieve the above object, The technical solution adopted by the invention is as follows:
A method of the multi-model sun normal direction based on clustering algorithm radiates prediction, comprising:
Step 110: collecting the history meteorological data of estimation range, the history meteorological data is that influence solar energy normal direction is straight
Connect the multi-class meteorological element of radiation;
Step 120: the history meteorological data being standardized, by the history meteorology number of standardization
According to using clustering algorithm to be clustered, and data are grouped according to cluster result, each group of the grouping include through
The cluster set of the meteorological element data of standardization, the meteorological element data and the cluster result phase of the cluster set
Together;
Step 130: carrying out models fitting using machine learning algorithm for each group of the data respectively, obtain multiple
Predict submodel;
Step 140: input real time meteorological data is standardized real time meteorological data, after standardization
Real time meteorological data and the multiple prediction submodel progress similitude matching, choose corresponding prediction submodel, the phase
Cluster centre similitude or sample like property matching to calculate the real time meteorological data after multiple prediction submodels and standardization
This distance similarity;
Step 150: being based on corresponding submodel, export prediction result.
In the step 120, the history meteorological data of standardization is clustered using clustering algorithm specifically,
The history meteorological data of standardization is subjected to sample clustering, the complex network society using complex network community detection
Group is detected
For the data point in existing database, calculate its similitude two-by-two, and record two datas reference numeral and
Similarity figure;
Cyberrelationship is established according to the similarity figure, corporations are carried out to the cyberrelationship using corporations' detection algorithm
Detection.
It is described that the cyberrelationship is carried out corporations to detect being specially to use Girvan-Newman using corporations' detection algorithm
Algorithm or label propagation algorithm carry out corporations' detection to the cyberrelationship;
In the step 110, the multi-class meteorological element is specifically included, wind direction, wind speed, global radiation, the direct spoke of normal direction
It penetrates, scatters, air pressure, temperature, cloud amount, visibility, cloud atlas.
In the step 120, clustering algorithm is specially hierarchical clustering, modular clustering or k-means clustering method.
It is in step 130, described to carry out models fitting using machine learning algorithm specifically:
Data in each subset are divided into training set and verifying collection, packet mode is used and randomly selected;
Models fitting is carried out using the machine learning method of non-modelling by mechanism.
The machine learning method using non-modelling by mechanism carries out models fitting specifically, using neural network, supporting
It is quasi- that the machine learning method of vector machine or logistic regression carries out model.
After the step 150, further includes:
It is continuously added real time data, the real time data is standardized, by the history of standardization
Meteorological data is clustered using clustering algorithm, obtains new grouping, and using machine learning algorithm to the new grouping into
Row models fitting, obtains new submodel, it is described it is each it is new be grouped into standardization comprising each meteorological element data
Each moment meteorology sample set.
Further include:
Step 160, it if submodel prediction deviation is larger or the meteorological condition of estimation range is unusual for the multiple prediction, weighs
The step of multiple step 110 to step 150.
After step 110, further includes:
Step 111: real time meteorological data is added in the history meteorological data, historical data is updated.
A kind of device of the multi-model sun normal direction radiation prediction based on clustering algorithm, comprising:
Meteorological data collection module, for collecting the history meteorological data of estimation range, the history meteorological data is shadow
Ring the multi-class meteorological element that solar energy normal direction directly radiates;
Cluster module, for being standardized to the history meteorological data, by the history of standardization
Meteorological data is clustered using clustering algorithm, and is grouped data according to cluster result, each group of packet of the grouping
The cluster set of meteorological element data containing normalized processing, the meteorological element data and the cluster of the cluster set
As a result identical;
Submodel establishes module, quasi- using machine learning algorithm progress model for being directed to each group of the data respectively
It closes, obtains multiple prediction submodels;
Meteorological data input module is standardized real time meteorological data for inputting real time meteorological data, will
Real time meteorological data and the multiple prediction submodel after standardization carry out similitude matching, choose corresponding prediction
Model, the similitude matching are the multiple cluster centres for predicting submodels with the real time meteorological data after standardization of calculating
Similitude or sample distance similarity;
As a result output module exports prediction result for being based on corresponding submodel.
The cluster module is complex network community detection module;
Complex network community detection module is standardized the history meteorological data, by standardization
The history meteorological data carries out sample clustering using complex network community detection, and is grouped to the result of cluster, described
Each group of grouping is each moment meteorology sample set comprising each meteorological element data of standardization;The complex network
Corporations' detection includes: to calculate its similitude two-by-two for the data point in existing database, and record the corresponding of two datas and compile
Number and similarity figure, cyberrelationship is established according to the similarity figure, using corporations' detection algorithm to the cyberrelationship
Carry out corporations' detection.
The invention has the advantages that meteorological element multi-class in history meteorological data is gathered using clustering algorithm
After class grouping, different prediction models is respectively created, model can be chosen when inputting real time meteorological data, selection is real-time
The close prediction submodel of meteorological data, so that prediction has specific aim, and then the adaptability of method for improving, it can be more
Higher precision of prediction is obtained in changeable actual environment;The resources model more refined is solar energy thermal-power-generating development
With the basis of project construction, it can effectively instruct the operation in power station to operate, to improve system generated energy, and then be conducive to too
Positive energy thermal power station preferably exerts advantages of oneself according to resource situation.
Detailed description of the invention
Fig. 1 is a kind of method flow of the multi-model sun normal direction radiation prediction based on clustering algorithm of one embodiment of the invention
Figure;
Fig. 2 is a kind of method stream of the multi-model sun normal direction radiation prediction based on clustering algorithm of another embodiment of the present invention
Cheng Tu;
Fig. 3 is a kind of knot of the device of the multi-model sun normal direction radiation prediction based on clustering algorithm of one embodiment of the invention
Structure block diagram;
Fig. 4 is a kind of device of the multi-model sun normal direction radiation prediction based on clustering algorithm of another embodiment of the present invention
Structural block diagram;
Fig. 5 is in a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm of another embodiment of the present invention
The cluster result of one group of history meteorological data;
Fig. 6 is in a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm of another embodiment of the present invention
The measured data prediction result schematic diagram that solar energy directly radiates.
Specific embodiment
Firstly the need of explanation, in each embodiment of the present invention, related term are as follows:
Standardization, standardization, that is, data normalization (normalization) processing are a basic works of data mining
Make, different evaluation index often has different a dimension and dimensional unit, such situation influence whether data analysis as a result,
In order to eliminate the dimension impact between index, need to carry out data normalization processing, to solve the comparativity between data target.
For initial data after data normalization is handled, each index is in the same order of magnitude, is appropriate for Comprehensive Correlation evaluation.
Cluster, cluster are claimed the process that the set of physics or abstract object is divided into the multiple classes being made of similar object
For cluster, cluster is in the prior art mature technology.
Models fitting refers to the real data obtained based on sampling, the pass between discovery and summary input quantity and output quantity
System carries out the fitting of rule in such a way that mathematical model is built, to realize summary and refinement for real model and rule.
In the present invention, machine learning method need to be used to carry out models fitting for every class data, the object of models fitting is sun normal direction
The mapping relations between other relevant weather parameters in the development trend and time series of radiation data.
Submodel first passes through headed by submodel after clustering algorithm clustered history meteorological data, uses each cluster
The model that the data of group are established.
Real time meteorological data needs for all relevant weather data in specific time period according to the difference of observation place
P value or mutual information or Spearman related coefficient or Pearson correlation coefficient, relevance detection are obtained using t inspection institute first
After index progress related resource data are filtered, obtain exact real time data and constitute, usually, original real time data
Length of time series should not shorter than 1 day, and the index of correlation on historical time sequence node should include: wind direction as far as possible, wind speed, always
Radiation, normal direction directly radiate, and scatter, air pressure, temperature, cloud amount, visibility, cloud atlas.
In the following, will be carried out by scheme of several specific embodiments to data backup provided in an embodiment of the present invention detailed
Introduce explanation.
Referring to FIG. 1, it illustrates a kind of multi-model sun methods based on clustering algorithm that one embodiment of the invention provides
To the method flow diagram of radiation prediction, it is somebody's turn to do the multi-model solar energy normal direction based on clustering algorithm and directly radiates prediction technique, comprising:
110, the history meteorological data of estimation range is collected, the history meteorological data is to influence the direct spoke of solar energy normal direction
The multi-class meteorological element penetrated;
120, the history meteorological data is standardized, the history meteorological data of standardization is made
It is clustered with clustering algorithm, and is grouped data according to cluster result, each group of the grouping includes through standard
Change the cluster set of the meteorological element data of processing, the meteorological element data of the cluster set are identical as the cluster result;
130, models fitting is carried out using machine learning algorithm for each group of the data respectively, obtains multiple predictions
Submodel;
140, input real time meteorological data, 141, real time meteorological data is standardized, after standardization
Real time meteorological data and the multiple prediction submodel progress similitude matching, choose corresponding prediction submodel, the phase
Cluster centre similitude or sample like property matching to calculate the real time meteorological data after multiple prediction submodels and standardization
This distance similarity;
150, it is based on corresponding submodel, exports prediction result.
In above-described embodiment, processing step is as follows:
1, data prediction;
Firstly, it is necessary to collect the history meteorological data of desired estimation range.According to the difference of historical data, each moment can
At the time of to select different dimension n and m association.Such as t moment, t-1 is collected, when t-2 ..., t-m are m total
It carves, corresponding solar elevation, solar azimuth, normal direction directly radiate, global radiation, wind speed, wind direction, cloud amount, temperature and influence
The n correlated characteristic that solar energy method mutually directly radiates, it should be noted that features described above can be according to specific historical data situation
Difference is adjusted, and the process and result for having no effect on the invention obtain.
2, it clusters;
(1) due to the data of separate sources, such as wind speed and temperature, source gap is larger, and also gap is larger for the regularity of distribution, because
This, needs to be counted first against each data source, and carries out corresponding data normalization (normalization work).
(2) after the completion of normalizing, the correlation between computation model is needed, Euclidean distance or mutual information can be selected, or
T inspection institute obtains p value or mutual information or Spearman related coefficient or Pearson correlation coefficient, relevance Testing index method
To measure the similitude (similarity) between two features.
A) each history resource data will be calculated as a variable, correspond to a sample point;
B) for the data point in existing database, its similitude is calculated two-by-two;
C) after the completion of Similarity measures, the reference numeral of two datas and similarity figure should be recorded;
D) similarity figure between all data is recorded, corresponding cyberrelationship is established, carries out society using clustering method
The detection of group, correlated results are as shown in Figure 5.
(3) according to above-mentioned cluster result, existing data set is divided into several subsets, it is according to above-mentioned analysis as a result, every
Data in a subset have stronger relevance, carry out models fitting that can be stronger.
3, models fitting;
In each subset, models fitting is carried out using machine learning algorithm;
A) data in each subset are divided into training set and verifying collects, packet mode is using randomly selecting or in order
It is cross-checked;
B) using the machine learning method that can be used for non-modelling by mechanism, such as neural network (BP neural network, radial base nerve
Network, deep neural network), support vector machines, logistic regression method, carry out models fitting.
4, Model Matching;
(1) when needing to carry out real data prediction, used input data arranges prediction when should be according to model foundation
The input of data.
(2) according to the input that sorts out, calculate between present input data and center of all categories or average value it is European away from
From or similarity relationships, carry out the matching of model, corresponding submodel is answered in determination.
5, prediction result is exported.
The corresponding direct radiation data of normal direction is exported according to submodel, it, can root if necessary to the prediction data of longer sequence
Process is slided accordingly, continues t+1, the prediction at t+2 or even t+k moment.
The major advantage of the present embodiment is:
1, the difference based on solar energy thermal-power-generating and photovoltaic power generation, directly radiates sun normal direction and predicts, and be directed to
Normal direction radiates opposite global radiation and is easier the characteristics of fluctuating, and carries out building for multi-model forecasting system;Propose it is a set of it is new too
Positive normal direction directly radiates prediction technique.
2, the correlation according to the direct radiation history data of normal direction in time series, is carried out by data mining algorithm
The cluster of historical data;Realize the automatic classification of resource data.
3, according to the direct radiation history data of normal direction between the correlation and all kinds of meteorologic factors in time series
Relevance carries out the cluster of historical data based on various features;It has been achieved under DIFFERENT METEOROLOGICAL CONDITIONS, automatic of each submodel
Match.
4, according to cluster result, models fitting is carried out for every class data using machine learning method;Utilize historical data
Establish the evolutionary model that the normal direction more refined directly radiates.
5, in real-time prediction, the similitude first between calculating present input data and center of all categories or average value is closed
System, and then types of models be applicable at that time is judged, it is predicted further according to trained model;Correspondence can be exported
The prediction result at moment.
In Fig. 1, multiple submodels 131 are the model obtained, and multiple submodels are used for prediction result, and real time meteorological data exists
After standardization, matches corresponding submodel and predicted.
Further, referring to FIG. 1, in the step 120, the history meteorological data of standardization is used cluster
Algorithm is clustered specifically, the history meteorological data of standardization is carried out sample using complex network community detection
Cluster, the complex network community detection include:
For the data point in existing database, calculate its similitude two-by-two, and record two datas reference numeral and
Similarity figure;
Cyberrelationship is established according to the similarity figure, corporations are carried out to the cyberrelationship using corporations' detection algorithm
Detection.
In above-described embodiment, processing step is as follows:
1, data prediction;
Firstly, it is necessary to collect the history meteorological data of desired estimation range.According to the difference of historical data, each moment can
At the time of to select different dimension n and m association.Such as t moment, t-1 is collected, when t-2 ..., t-m are m total
It carves, corresponding solar elevation, solar azimuth, normal direction directly radiate, global radiation, wind speed, wind direction, cloud amount, temperature and influence
The n correlated characteristic that solar energy normal direction directly radiates, it should be noted that features described above can be according to specific historical data situation
Difference is adjusted, and the process and result for having no effect on the invention obtain.
2, it clusters;
(1) due to the data of separate sources, such as wind speed and temperature, source gap is larger, and also gap is larger for the regularity of distribution, because
This, needs to be counted first against each data source, and carries out corresponding data normalization (normalization work).
(2) after the completion of normalizing, (k-mean, hierarchical clustering, k neighbour) can be directly selected, routine clustering algorithm will be original
Data are divided into multiple subsets.When due to project characteristic, when routine clustering algorithm effect is undesirable, selection complex network community is detected
Method carry out sample clustering:
A) point that each sample will be counted as in complex network, the relevant history resource data of each point is then it
Corresponding feature;
B) for the data point in existing database, calculating its similitude two-by-two (can be selected Euclidean distance or mutual information side
Method or t inspection institute obtain p value or mutual information or Spearman related coefficient or Pearson correlation coefficient relevance Testing index
To measure the similitude between two features);
C) after the completion of Similarity measures, the reference numeral of two datas and similarity figure should be recorded;
D) according to the similarity figure between all data points, corresponding cyberrelationship is established, then carries out the inspection of corporations
It surveys, can be used Girvan-Newman algorithm (GN algorithm), label propagation algorithm (Label Propagation Algorithm,
LPA) or other corporations' detection algorithms, the correlated results based on GN algorithm are as shown in Figure 5.Each corporation in figure, so that it may right
It should be a class in clustering algorithm.
(3) according to the category division of above-mentioned clustering algorithm output as a result, corresponding can be divided into existing data set several
Subset, according to above-mentioned analysis as a result, the data in each subset have stronger relevance, the carry out model being more advantageous to
Fitting and rule discovery.
3, models fitting;
In each subset, models fitting is carried out using machine learning algorithm;
A) data in each subset are divided into training set and verifying collects, packet mode is using randomly selecting or in order
It is cross-checked;
B) using the machine learning method that can be used for non-modelling by mechanism, such as neural network (BP neural network, radial base nerve
Network, deep neural network), support vector machines, logistic regression method, carry out models fitting.
4, Model Matching;
(1) when needing to carry out real data prediction, used input data arranges prediction when should be according to model foundation
The input of data.
(2) according to the input that sorts out, calculate between present input data and center of all categories or average value it is European away from
From or similarity relationships, carry out the matching of model, corresponding submodel is answered in determination.
5, prediction result is exported.
The corresponding direct radiation data of normal direction is exported according to submodel, it, can root if necessary to the prediction data of longer sequence
Process is slided accordingly, continues t+1, the prediction at t+2 or even t+k moment.
It in the present embodiment, is detected using complex network community, can effectively promote forecasting accuracy, promote precision of prediction;
Optionally, another group of new submodel can establish using complex network community detection, it is real when choosing submodel
When meteorological data in Matching Model, have more more options, it is more accurate to choose corresponding submodel.
Further, described that corporations' detection specially use is carried out to the cyberrelationship using corporations' detection algorithm
Girvan-Newman algorithm or label propagation algorithm carry out corporations' detection to the cyberrelationship.
In above-described embodiment, Girvan-Newman algorithm or label propagation algorithm can effectively promote forecasting accuracy,
Promote precision of prediction;
Further, in the step 110, the multi-class meteorological element is specifically included, wind direction, wind speed, global radiation, method
To directly radiating, scatter, air pressure, temperature, cloud amount, visibility, cloud atlas.
In above-described embodiment, wind direction, wind speed, global radiation, normal direction directly radiated, and is scattered, air pressure, temperature cloud amount, visibility,
Cloud atlas directly affects solar energy normal direction and directly radiates.
Optionally, it can establish another group of new submodel using Girvan-Newman algorithm or label propagation algorithm,
When choosing submodel, real time meteorological data has more more options in Matching Model, more accurate to choose corresponding submodel.
Further, in the step 120, clustering algorithm is specially hierarchical clustering, modular clustering or k-means cluster
Method.
In above-described embodiment, cluster knot can be effectively promoted using hierarchical clustering, modular clustering or k-means clustering method
The accuracy of fruit.
It is further, in step 130, described to carry out models fitting using machine learning algorithm specifically:
Data in each subset are divided into training set and verifying collection, packet mode is used and randomly selected;
Models fitting is carried out using the machine learning method of non-modelling by mechanism.
It is above-mentioned to implement, it can avoid causing due to the regularity of itself sequence of data by the way of randomly selecting
Training and verifying there is relatively large deviation, randomly select it is possible to prevente effectively from this is happened.
Further, the machine learning method using non-modelling by mechanism carries out models fitting specifically, using nerve
It is quasi- that the machine learning method of network, support vector machines or logistic regression carries out model.
In above-described embodiment, neural network, support vector machine or the machine learning method of logistic regression, neural network are used
The advantages of, it is that its development history is more remote, neural network is more mature.The advantages of support vector machines is SVM in Small Amount sample
The non-linear relation being easy to get when scale between data and feature, can to avoid use neural network structure select drawn game
Portion's minimum problem, interpretation is strong, can solve higher-dimension problem.The advantages of machine learning method of logistic regression, 1) prediction
The result is that probability of the boundary between 0 and 1;2) it can be adapted for continuity and classification independent variable;3) it is easy to use and explain.
Further, after the step 150, further includes:
It is continuously added real time data, the real time data is standardized, by the history of standardization
Meteorological data is clustered using clustering algorithm, obtains new grouping, and using machine learning algorithm to the new grouping into
Row models fitting, obtains new submodel, it is described it is each it is new be grouped into standardization comprising each meteorological element data
Each moment meteorology sample set.
In above-described embodiment, since prediction model can be during continuous iteration be predicted, prediction result is at any time
Increasing can be gradually inaccurate, in order to promote the accuracy of prediction, needs constantly to fill into the number for the real-time weather having occurred that
According to this real time meteorological data is used not only for the prediction of the following meteorology, also new submodel is provided with for constantly updating, so that in advance
The continuous self-renewing of model is surveyed, the accuracy of prediction is continuously increased.
With the accumulation of time series data, this model further training data and verify data can be modified and
It updates.
Further, further includes:
Step 160, it if submodel prediction deviation is larger or the meteorological condition of estimation range is unusual for the multiple prediction, weighs
The step of multiple step 110 to step 150.
In above-described embodiment, when there is relatively large deviation in prediction result or unusual meteorological condition occurs in somewhere, it should to mould
Type is reset, and process according to the invention carries out models fitting and prediction again.To guarantee the accuracy of prediction.
Further, referring to FIG. 2, after step 110, further includes:
Step 111: real time meteorological data is added in the history meteorological data, historical data is updated.
In above-described embodiment, for history meteorological data, it is also desirable to which constantly alternate real time meteorological data, guarantees again
When establishing submodel, more acurrate submodel is obtained, and then the model prediction made is more accurate.
Referring to FIG. 3, it illustrates a kind of multi-model sun methods based on clustering algorithm that one embodiment of the invention provides
To the structural block diagram of the device of radiation prediction, it is somebody's turn to do the multi-model solar energy normal direction based on clustering algorithm and directly radiates prediction meanss,
Include:
Meteorological data collection module, for collecting the history meteorological data of estimation range, the history meteorological data is shadow
Ring the multi-class meteorological element that solar energy normal direction directly radiates;
Cluster module, for being standardized to the history meteorological data, by the history of standardization
Meteorological data is clustered using clustering algorithm, and is grouped data according to cluster result, each group of packet of the grouping
The cluster set of meteorological element data containing normalized processing, the meteorological element data and the cluster of the cluster set
As a result identical;
Submodel establishes module, quasi- using machine learning algorithm progress model for being directed to each group of the data respectively
It closes, obtains multiple prediction submodels;
Meteorological data input module is standardized real time meteorological data for inputting real time meteorological data, will
Real time meteorological data and the multiple prediction submodel after standardization carry out similitude matching, choose corresponding prediction
Model, the similitude matching are the multiple cluster centres for predicting submodels with the real time meteorological data after standardization of calculating
Similitude or sample distance similarity;
As a result output module exports prediction result for being based on corresponding submodel.
In above-described embodiment, in above-described embodiment, processing step is as follows:
1, data prediction;
Firstly, it is necessary to collect the history meteorological data of desired estimation range.Meteorological data collection module 210 is according to history number
According to difference, at the time of each moment can select different dimension n and m association.Such as t moment, t-1 is collected,
At t-2 ..., the t-m total m moment, corresponding solar elevation, solar azimuth, normal direction directly radiate, global radiation, wind speed, wind
To, cloud amount, temperature and influence n correlated characteristic directly radiating of solar energy normal direction, it should be noted that features described above can basis
The difference of specific historical data situation, is adjusted, and the process and result for having no effect on the invention obtain.
2, cluster module 220 is clustered;
(1) due to the data of separate sources, such as wind speed and temperature, source gap is larger, and also gap is larger for the regularity of distribution, because
This, needs to be counted first against each data source, and carries out corresponding data normalization (normalization work).
(2) after the completion of normalizing, the correlation between computation model is needed, Euclidean distance or mutual information can be selected, or
T inspection institute obtains p value or mutual information or Spearman related coefficient or Pearson correlation coefficient relevance Testing index method
To measure the similitude (similarity) between two features.
A) each history resource data will be calculated as a variable, correspond to a sample point;
B) for the data point in existing database, its similitude is calculated two-by-two;
C) after the completion of Similarity measures, the reference numeral of two datas and similarity figure should be recorded;
D) similarity figure between all data is recorded, corresponding cyberrelationship is established, carries out society using clustering method
The detection of group, correlated results are as shown in Figure 5.
(3) according to above-mentioned cluster result, existing data set is divided into several subsets, it is according to above-mentioned analysis as a result, every
Data in a subset have stronger relevance, carry out models fitting that can be stronger.
3, submodel establishes module 230 and establishes each submodel with models fitting;
In each subset, models fitting is carried out using machine learning algorithm;
A) data in each subset are divided into training set and verifying collects, packet mode is using randomly selecting or in order
It is cross-checked;
B) using the machine learning method that can be used for non-modelling by mechanism, such as neural network (BP neural network, radial base nerve
Network, deep neural network), support vector machines, logistic regression method, carry out models fitting.
4,240 receiving real-time data of meteorological data input module and corresponding model is matched;
(1) when needing to carry out real data prediction, used input data arranges prediction when should be according to model foundation
The input of data.
(2) according to the input that sorts out, calculate between present input data and center of all categories or average value it is European away from
From or similarity relationships, carry out the matching of model, corresponding submodel is answered in determination.
5, result output module 250 exports prediction result.
The corresponding direct radiation data of normal direction is exported according to submodel, it, can root if necessary to the prediction data of longer sequence
Process is slided accordingly, continues t+1, the prediction at t+2 or even t+k moment.
The major advantage of the present embodiment is:
1, the difference based on solar energy thermal-power-generating and photovoltaic power generation, directly radiates sun normal direction and predicts, and be directed to
Normal direction radiates opposite global radiation and is easier the characteristics of fluctuating, and carries out building for multi-model forecasting system;Propose it is a set of it is new too
Positive normal direction directly radiates prediction technique.
2, the correlation according to the direct radiation history data of normal direction in time series, is carried out by data mining algorithm
The cluster of historical data;Realize the automatic classification of resource data.
3, according to the direct radiation history data of normal direction between the correlation and all kinds of meteorologic factors in time series
Relevance carries out the cluster of historical data based on various features;It has been achieved under DIFFERENT METEOROLOGICAL CONDITIONS, automatic of each submodel
Match.
4, according to cluster result, models fitting is carried out for every class data using machine learning method;Utilize historical data
Establish the evolutionary model that the normal direction more refined directly radiates.
5, in real-time prediction, the similitude first between calculating present input data and center of all categories or average value is closed
System, and then types of models be applicable at that time is judged, it is predicted further according to trained model;Correspondence can be exported
The prediction result at moment.
In Fig. 1, multiple submodels 131 are the model obtained, and multiple submodels are used for prediction result, and real time meteorological data exists
After standardization, matches corresponding submodel and predicted.
Further, referring to FIG. 4, the cluster module is complex network community detection module;
Complex network community detection module 221 is standardized the history meteorological data, by standardization
The history meteorological data carry out sample clustering using complex network community detection, and the result of cluster is grouped, institute
Each group of grouping is stated as each moment meteorology sample set comprising each meteorological element data of standardization;The complex web
The detection of network corporations includes: to calculate its similitude two-by-two, and record the correspondence of two datas for the data point in existing database
Number and similarity figure, are established cyberrelationship according to the similarity figure, are closed using corporations' detection algorithm to the network
System carries out corporations' detection.
In above-described embodiment, complex network community detection module 221 carries out sample clustering, specifically:
A) point that each sample will be counted as in complex network, the relevant history resource data of each point is then it
Corresponding feature;
B) for the data point in existing database, calculating its similitude two-by-two (can be selected Euclidean distance or mutual information side
Method measures the similitude between two features);
C) after the completion of Similarity measures, the reference numeral of two datas and similarity figure should be recorded;
D) according to the similarity figure between all data points, corresponding cyberrelationship is established, then carries out the inspection of corporations
It surveys, can be used Girvan-Newman algorithm (GN algorithm), label propagation algorithm (Label Propagation Algorithm,
LPA) or other corporations' detection algorithms, the correlated results based on GN algorithm are as shown in Figure 5.Each corporation in figure, so that it may right
It should be a class in clustering algorithm.
To sum up, by the way that method of the invention to be compared with other correlation techniques, it was demonstrated that effectiveness of the invention is adopted
It uses the time series of 1 clock very much as raw data set, is emulated, as a result as shown in Figure 6.
As shown in the table, BP neural network, radial base neural net (RBF), the single mode of extreme learning machine (ELM) will be used
Type system be to be compared with the Multiple Models Algorithm of the method for the present invention, can be with using Minimum Mean Square Error (MSE) as evaluation index
Find out that the method for the present invention has great advantages.
Method | Models fitting mode | Minimum Mean Square Error (MSE) |
BP | Single model | 328.82 |
RBF | Single model | 276.43 |
ELM | Single model | 157.76 |
The present invention | Multi-model | 82.54 |
It is to be appreciated that the directional instruction (such as up, down, left, right, before and after ...) of institute is only used for solving in the present embodiment
It releases in relative positional relationship, the motion conditions etc. under a certain particular pose (as shown in the picture) between each component, if this is specific
When posture changes, then directionality instruction also correspondingly changes correspondingly.
In addition, the description for being related to " first ", " second " etc. is used for description purposes only, it is not understood to indicate or imply it
Relative importance or the quantity for implicitly indicating indicated technical characteristic.Define " first " as a result, the feature of " second " can
To explicitly or implicitly include at least one this feature.
Technical solution between each embodiment can be combined with each other, but must be with those of ordinary skill in the art's energy
It is enough realize based on, will be understood that the knot of this technical solution when conflicting or cannot achieve when occurs in the combination of technical solution
Conjunction is not present, also not the present invention claims protection scope within.The component and structure and step that the present embodiment does not describe in detail
The rapid well-known components for belonging to the industry and common structure or conventional means, do not describe one by one here.
Claims (10)
1. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm characterized by comprising
Step 110: collecting the history meteorological data of estimation range, the history meteorological data is to influence the direct spoke of solar energy normal direction
The multi-class meteorological element penetrated;
Step 120: the history meteorological data being standardized, the history meteorological data of standardization is made
It is clustered with clustering algorithm, and is grouped data according to cluster result, each group of the grouping includes through standard
Change the cluster set of the meteorological element data of processing, the meteorological element data of the cluster set are identical as the cluster result;
Step 130: carrying out models fitting using machine learning algorithm for each group of the data respectively, obtain multiple predictions
Submodel;
Step 140: input real time meteorological data is standardized real time meteorological data, by the reality after standardization
When meteorological data and the multiple prediction submodel progress similitude matching, choose corresponding prediction submodel, the similitude
Matching for calculate it is multiple prediction submodels and standardization after real time meteorological data cluster centre similitudes or sample away from
From similitude;
Step 150: being based on corresponding submodel, export prediction result.
2. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as described in claim 1
In in the step 120, the history meteorological data of standardization being clustered using clustering algorithm specifically, will mark
The history meteorological data of standardization processing carries out sample clustering, the complex network community inspection using complex network community detection
Survey includes:
For the data point in existing database, its similitude is calculated two-by-two, and record the reference numeral of two datas and similar
Property numerical value;
Cyberrelationship is established according to the similarity figure, corporations' inspection is carried out to the cyberrelationship using corporations' detection algorithm
It surveys.
3. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as claimed in claim 2
In described to carry out corporations to detect being specially to use Girvan-Newman algorithm to the cyberrelationship using corporations' detection algorithm
Or label propagation algorithm carries out corporations' detection to the cyberrelationship;
In the step 110, the multi-class meteorological element is specifically included: wind direction, wind speed, global radiation, normal direction are directly radiated, are dissipated
It penetrates, air pressure, temperature, cloud amount, visibility and cloud atlas.
In the step 120, clustering algorithm is specially hierarchical clustering, modular clustering or k-means clustering method.
4. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as described in claim 1
In in step 130, described to carry out models fitting using machine learning algorithm specifically:
Data in each subset are divided into training set and verifying collection, packet mode is used and randomly selected;
Models fitting is carried out using the machine learning method of non-modelling by mechanism.
5. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as claimed in claim 4
In the machine learning method using non-modelling by mechanism carries out models fitting specifically, using neural network, support vector machine
Or the machine learning method of logistic regression carries out model and intends.
6. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as described in claim 1
In after the step 150, further includes:
It is continuously added real time data, the real time data is standardized, the history of standardization is meteorological
Data are clustered using clustering algorithm, obtain new grouping, and carry out mould to the new grouping using machine learning algorithm
Type fitting, obtains new submodel, when described each new each comprising each meteorological element data for being grouped into standardization
Carve meteorological sample set.
7. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as described in claim 1
In, further includes:
Step 160, it if submodel prediction deviation is larger or the meteorological condition of estimation range is unusual for the multiple prediction, repeats to walk
Rapid 110 to step 150 the step of.
8. a kind of method of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as described in claim 1
In after step 110, further includes:
Step 111: real time meteorological data is added in the history meteorological data, historical data is updated.
9. a kind of device of the multi-model sun normal direction radiation prediction based on clustering algorithm characterized by comprising
Meteorological data collection module, for collecting the history meteorological data of estimation range, the history meteorological data is to influence too
The multi-class meteorological element that positive energy normal direction directly radiates;
Cluster module, it is for being standardized to the history meteorological data, the history of standardization is meteorological
Data are clustered using clustering algorithm, and are grouped data according to cluster result, and each group of the grouping includes
The cluster set of the meteorological element data of normalized processing, the meteorological element data of the cluster set and the cluster result
It is identical;
Submodel establishes module, carries out models fitting using machine learning algorithm for being directed to each group of the data respectively,
Obtain multiple prediction submodels;
Meteorological data input module is standardized real time meteorological data for inputting real time meteorological data, by standard
Change treated real time meteorological data and the progress similitude matching of the multiple prediction submodel, chooses corresponding prediction submodule
Type, the similitude matching are the multiple cluster centre phases for predicting submodels and the real time meteorological data after standardization of calculating
Like property or sample distance similarity;
As a result output module exports prediction result for being based on corresponding submodel.
10. a kind of device of the multi-model sun normal direction radiation prediction based on clustering algorithm, feature exist as claimed in claim 9
In the cluster module is complex network community detection module;
Complex network community detection module is standardized the history meteorological data, will be described in standardization
History meteorological data carries out sample clustering using complex network community detection, and is grouped to the result of cluster, the grouping
Each group be standardization each moment meteorology sample set comprising each meteorological element data;The complex network community
Detection include: its similitude calculated two-by-two for the data point in existing database, and record two datas reference numeral and
Similarity figure establishes cyberrelationship according to the similarity figure, is carried out using corporations' detection algorithm to the cyberrelationship
Corporations' detection.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910388405.XA CN110197298A (en) | 2019-05-10 | 2019-05-10 | The method and device of multi-model sun normal direction radiation prediction based on clustering algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910388405.XA CN110197298A (en) | 2019-05-10 | 2019-05-10 | The method and device of multi-model sun normal direction radiation prediction based on clustering algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110197298A true CN110197298A (en) | 2019-09-03 |
Family
ID=67752535
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910388405.XA Pending CN110197298A (en) | 2019-05-10 | 2019-05-10 | The method and device of multi-model sun normal direction radiation prediction based on clustering algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110197298A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111814398A (en) * | 2020-07-08 | 2020-10-23 | 国网河北省电力有限公司 | Map-based earth surface solar radiance prediction method integrating space-time attention |
CN112801357A (en) * | 2021-01-21 | 2021-05-14 | 长江慧控科技(武汉)有限公司 | Sunshine radiation amount prediction method, device, equipment and storage medium |
CN112926625A (en) * | 2020-11-25 | 2021-06-08 | 北方工业大学 | Method for analyzing deviation influence factors of satellite radiation data |
CN113159041A (en) * | 2021-03-15 | 2021-07-23 | 成都信息工程大学 | Torrent flow line extraction and identification method based on deep learning |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101383748A (en) * | 2008-10-24 | 2009-03-11 | 北京航空航天大学 | Community division method in complex network |
CN102269124A (en) * | 2011-06-30 | 2011-12-07 | 内蒙古电力勘测设计院 | Ultra-short term wind power station generated power forecasting system |
CN103955749A (en) * | 2014-03-19 | 2014-07-30 | 内蒙古电力勘测设计院 | Sun direct radiation value prediction method and system |
CN106650784A (en) * | 2016-11-04 | 2017-05-10 | 许继集团有限公司 | Feature clustering comparison-based power prediction method and device for photovoltaic power station |
CN108053061A (en) * | 2017-12-08 | 2018-05-18 | 天津大学 | A kind of solar energy irradiation level Forecasting Methodology based on improvement convolutional neural networks |
-
2019
- 2019-05-10 CN CN201910388405.XA patent/CN110197298A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101383748A (en) * | 2008-10-24 | 2009-03-11 | 北京航空航天大学 | Community division method in complex network |
CN102269124A (en) * | 2011-06-30 | 2011-12-07 | 内蒙古电力勘测设计院 | Ultra-short term wind power station generated power forecasting system |
CN103955749A (en) * | 2014-03-19 | 2014-07-30 | 内蒙古电力勘测设计院 | Sun direct radiation value prediction method and system |
CN106650784A (en) * | 2016-11-04 | 2017-05-10 | 许继集团有限公司 | Feature clustering comparison-based power prediction method and device for photovoltaic power station |
CN108053061A (en) * | 2017-12-08 | 2018-05-18 | 天津大学 | A kind of solar energy irradiation level Forecasting Methodology based on improvement convolutional neural networks |
Non-Patent Citations (1)
Title |
---|
KUK YEOL BAE等: "Hourly Solar Irradiance Prediction Based on Support Vector Machine and Its Error Analysis", 《IEEE TRANSACTIONS ON POWER SYSTEMS》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111814398A (en) * | 2020-07-08 | 2020-10-23 | 国网河北省电力有限公司 | Map-based earth surface solar radiance prediction method integrating space-time attention |
CN111814398B (en) * | 2020-07-08 | 2023-09-29 | 国网河北省电力有限公司 | Map-based earth surface solar radiance prediction method integrating space-time attention |
CN112926625A (en) * | 2020-11-25 | 2021-06-08 | 北方工业大学 | Method for analyzing deviation influence factors of satellite radiation data |
CN112926625B (en) * | 2020-11-25 | 2023-12-22 | 北方工业大学 | Deviation influence factor analysis method for satellite radiation data |
CN112801357A (en) * | 2021-01-21 | 2021-05-14 | 长江慧控科技(武汉)有限公司 | Sunshine radiation amount prediction method, device, equipment and storage medium |
CN112801357B (en) * | 2021-01-21 | 2024-04-02 | 长江慧控科技(武汉)有限公司 | Solar radiation quantity prediction method, device, equipment and storage medium |
CN113159041A (en) * | 2021-03-15 | 2021-07-23 | 成都信息工程大学 | Torrent flow line extraction and identification method based on deep learning |
CN113159041B (en) * | 2021-03-15 | 2022-10-04 | 成都信息工程大学 | Rapid flow line extraction and identification method based on deep learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110197298A (en) | The method and device of multi-model sun normal direction radiation prediction based on clustering algorithm | |
CN111369070B (en) | Multimode fusion photovoltaic power prediction method based on envelope clustering | |
US7228235B2 (en) | System and method for enhanced measure-correlate-predict for a wind farm location | |
CN107944622B (en) | Wind power prediction method based on continuous time interval clustering | |
CN110990461A (en) | Big data analysis model algorithm model selection method and device, electronic equipment and medium | |
CN102411729B (en) | Wind power prediction method based on adaptive linear logic network | |
CN103489046A (en) | Method for predicting wind power plant short-term power | |
CN103853106A (en) | Energy consumption prediction parameter optimization method of building energy supply device | |
CN103473621A (en) | Wind power station short-term power prediction method | |
CN112818604A (en) | Wind turbine generator risk degree assessment method based on wind power prediction | |
CN111628494B (en) | Low-voltage distribution network topology identification method and system based on logistic regression method | |
US20240054267A1 (en) | Method for planning a layout of a renewable energy site | |
CN116911806B (en) | Internet + based power enterprise energy information management system | |
CN115036922B (en) | Distributed photovoltaic power generation electric quantity prediction method and system | |
Molina-Garcia et al. | Vertical wind profile characterization and identification of patterns based on a shape clustering algorithm | |
CN115013258A (en) | Intelligent soft measurement method and device for wind speed in front of wind turbine generator and server | |
CN117421926A (en) | Intelligent wind power equipment management and control method and system based on digital twin technology | |
CN113435101B (en) | Particle swarm optimization-based power failure prediction method for support vector machine | |
CN109508820A (en) | Campus electricity demand forecasting modeling method based on differentiation modeling | |
CN108932554A (en) | The method for optimizing configuration and device of a kind of wind power plant flow field measuring point | |
CN112508278A (en) | Multi-connected system load prediction method based on evidence regression multi-model | |
CN115907228A (en) | Short-term power load prediction analysis method based on PSO-LSSVM | |
CN110163437A (en) | Day-ahead photovoltaic power generation power prediction method based on DPK-means | |
Qin | Software reliability prediction model based on PSO and SVM | |
Cho et al. | Application of Parallel ANN-PSO to Hourly Solar PV Estimation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190903 |
|
RJ01 | Rejection of invention patent application after publication |