CN109858665A - Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO - Google Patents
Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO Download PDFInfo
- Publication number
- CN109858665A CN109858665A CN201811488151.0A CN201811488151A CN109858665A CN 109858665 A CN109858665 A CN 109858665A CN 201811488151 A CN201811488151 A CN 201811488151A CN 109858665 A CN109858665 A CN 109858665A
- Authority
- CN
- China
- Prior art keywords
- weather
- photovoltaic
- data
- anfis
- pso
- 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
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses the photovoltaic short term power prediction techniques based on Feature Selection and ANFIS-PSO, comprising: the somewhere photovoltaic generation power for any time period that selection has occurred and the historical data of weather；Feature Selection is carried out to period this area's photovoltaic generation power and weather historical data, selecting influences the historical data of the strong weather of the degree of correlation as weather characteristics data to photovoltaic generation power；To the training network that the weather characteristics data input adaptive neuro-fuzzy inference algorithm of screening is built, training network carries out data training to the weather characteristics data of screening and constructs fuzzy inference system；When being trained to weather characteristics data, training network is optimized using particle swarm algorithm；The weather characteristics data for inputting the prediction of this area's future certain time period, obtain the photovoltaic power generation power prediction result of this area's future certain time period.Input data greatly simplifies, and saves the data training time, is not easy to fall into local minimum in training pattern optimization process.
Description
Technical field
The present invention relates to photovoltaic power generation power prediction studying technological domains, are specially based on Feature Selection and ANFIS-PSO
Photovoltaic short term power prediction technique.
Background technique
In recent years, with the continuous appearance of national policy, solar energy power generating has become the emerging of state key development
Industry.Currently, solar photovoltaic generation system gradually switchs to distribution by centralization, distributed photovoltaic generating system is new within 2018
Increasing scale is more than centralized photovoltaic for the first time.At the same time, National Development and Reform Commission, the Ministry of Finance combine to have printed and distributed with Bureau of Energy and " close
In the notice of photovoltaic power generation relevant issues in 2018 ", it explicitly points out in notice and distributed photovoltaic orderly development is supported to become next
The most important thing of renewable energy work is walked, and in 10,000,000 kilowatts of Program Construction or so scale distribution formula photovoltaic projects in this year,
Large-scale distributed photovoltaic access will have an immense impact on to optimal dispatch and safety and stability.In this context, how
Prediction accurately and timely is carried out to distributed photovoltaic power output, becomes urgent problem to be solved.
Domestic and foreign scholars have carried out photovoltaic research to photovoltaic power generation power prediction method, and existing method mainly includes the time
Sequence algorithm, support vector machines, neural network and hybrid algorithm etc..But these algorithms are asked mainly for the prediction of photovoltaic plant
Topic, prediction algorithm still less for the forecasting research of distributed photovoltaic and previous generally require the elder generation according to professional person
It tests knowledge and chooses information useful in input feature vector, remove the information of redundancy.Under normal circumstances, photovoltaic generating system output is influenced
The factor of power is complex, includes pressure information, intensity of solar radiation, photovoltaic panel surface temperature, aerial cloud cluster etc., analysis
The input data of diversification makes modeling pattern relative complex.In addition, the period of traditional model training mode is longer, and optimize
Process is easily trapped into local minimum, prevent model learning is from being optimal effect.
It is as above in order to further solve the problems, such as, it is pre- to invent a kind of distributed photovoltaic power generation power based on RF-ANFIS-PSO
It surveys model and carrys out Accurate Prediction distributed photovoltaic power generation power.
Summary of the invention
In view of the above problems, the present invention provides the photovoltaic short term power prediction sides based on Feature Selection and ANFIS-PSO
Method, input data greatly simplify, and save the data training time, are not easy to fall into local minimum in training pattern optimization process
Value, can allow model learning to be optimal as a result, the problems in background technique can be solved effectively.
To achieve the above object, the invention provides the following technical scheme: the photovoltaic based on Feature Selection and ANFIS-PSO is short
Phase power forecasting method, comprising:
S1 selects the somewhere photovoltaic generation power of any time period occurred and the historical data of weather；
S2 carries out Feature Selection to period this area's photovoltaic generation power and weather historical data, selects
The historical data of the strong weather of the degree of correlation is influenced as weather characteristics data on photovoltaic generation power；
S3 is input in trained network the weather characteristics data screened in S2, and training network uses adaptive neural network
Fuzzy Logic Reasoning Algorithm is built, and training network carries out data training to the weather characteristics data of screening and constructs fuzzy inference system；
S4 when being trained to weather characteristics data, optimizes training network using particle swarm algorithm；
The photovoltaic power generation power prediction model of this area's future certain time period is obtained according to the weather characteristics data of screening；
S5, the weather characteristics data of input this area's future certain time period prediction, obtains this area's future sometime
The photovoltaic power generation power prediction result of section.
As a preferred technical solution of the present invention, the Feature Selection method in the S2 uses the feature of random forest
Contribution Analysis method, each bifurcated gain situation of random forest (Random Forest, RF) carry out Feature Selection, effectively sieve
Choosing influences the validity feature of photovoltaic power generation power prediction, and the historical data of weather is about the important of photovoltaic generation power historical data
Degree is ranked up analysis, calculates the historical data for influencing the big weather of photovoltaic generation power as weather characteristics data.
RF is by there is the repeated sampling put back to (Bootstrap Sampling) mode, from original training set B at random
It extracts several samples and generates new sample set, the forest set of k decision tree composition is then generated according to each subsample.
As a preferred technical solution of the present invention, Adaptive Neural-fuzzy Inference algorithm (Adaptive in S3
Network-based Fuzzy Inference System, ANFIS) fuzzy logic unit and neural network are organically combined,
To form a kind of new fuzzy inference system, the premise parameter and consequent parameter in model are by back-propagation algorithm and minimum two
Multiplication study obtains.
As a preferred technical solution of the present invention, particle swarm algorithm (Particle swarm in S4
Optimization, PSO) optimization algorithm simulates the predation of birds, and in N-dimensional search space, m is in primary group
Number of particles, position statement of i-th of particle in search space are as follows:
To the motion profile expression formula of particle are as follows:
W=wdamp*w (4)
Single particle optimal trajectory is denoted asI.e. individual extreme value, population optimal trajectory are denoted asThat is global extremum, particle movement speed are denoted asWherein, w indicates inertia power
Weight, wdampFor Inertia Weight damping ratio, c1c2For aceleration pulse, meaning is to guarantee have the particle of smaller value attached in target value
It closely hovers, there is the larger value particle to rush at or beyond object boundary, c1Represent self influencing for particle, c2Embody social shadow
It rings, r1 r2It is the random number changed in [0,1] range, for guaranteeing population diversity.
As a preferred technical solution of the present invention, the duration of any time period occurred in the S1 is not less than S5
The duration of middle future certain time period.
As a preferred technical solution of the present invention, the historical data of weather includes humidity, air pressure, air in the S2
Density, surface temperature, intensity of solar radiation, average dew point, low clouds coverage rate, medium cloud coverage rate and high cloud coverage rate.
Compared with prior art, the beneficial effects of the present invention are:
1, input data greatly simplifies, and saves the data training time；
2, it is not easy to fall into local minimum in training pattern optimization process, model learning can be allowed to be optimal result
3, optimal solution can comparatively fast be converged to using the ANFIS algorithm of PSO training, effectively shortens optimal time, divided
In cloth photovoltaic power generation power prediction scene, prediction essence can preferably be weighed using the ANFIS-PSO model of RF Feature Selection
Degree and prediction two aspect factor of resource consumption.
Detailed description of the invention
Fig. 1 is Adaptive Neuro-fuzzy Inference model support composition of the present invention；
Fig. 2 is that the present invention is based on the photovoltaic power generation power prediction illustratons of model of RF-ANFIS-PSO；
Fig. 3 is the feature scoring figure of each input feature vector of the present invention；
Fig. 4 temperature and irradiation level are to photovoltaic power output impact analysis figure.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Embodiment:
The present invention provides a kind of technical solution referring to FIG. 1 to FIG. 4:
Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO, comprising:
S1 selects the somewhere photovoltaic generation power of any time period occurred and the historical data of weather；
S2 carries out Feature Selection to period this area's photovoltaic generation power and weather historical data, selects
The historical data of the strong weather of the degree of correlation is influenced as weather characteristics data on photovoltaic generation power；
S3 is input in trained network the weather characteristics data screened in S2, and training network uses adaptive neural network
Fuzzy Logic Reasoning Algorithm is built, and training network carries out data training to the weather characteristics data of screening and constructs fuzzy inference system；
S4 when being trained to weather characteristics data, optimizes training network using particle swarm algorithm；
The photovoltaic power generation power prediction model of this area's future certain time period is obtained according to the weather characteristics data of screening；
S5, the weather characteristics data of input this area's future certain time period prediction, obtains this area's future sometime
The photovoltaic power generation power prediction result of section.
As a preferred technical solution of the present invention, the Feature Selection method in the S2 uses the feature of random forest
Contribution Analysis method, each bifurcated gain situation of random forest (Random Forest, RF) carry out Feature Selection, effectively sieve
Choosing influences the validity feature of photovoltaic power generation power prediction, and the historical data of weather is about the important of photovoltaic generation power historical data
Degree is ranked up analysis, calculates the historical data for influencing the big weather of photovoltaic generation power as weather characteristics data.
RF is by there is the repeated sampling put back to (Bootstrap Sampling) mode, from original training set B at random
It extracts several samples and generates new sample set, the forest set of k decision tree composition is then generated according to each subsample.
As a preferred technical solution of the present invention, particle swarm algorithm (Particle swarm in S4
Optimization, PSO) optimization algorithm simulates the predation of birds, and in N-dimensional search space, m is in primary group
Number of particles, position statement of i-th of particle in search space are as follows:
To the motion profile expression formula of particle are as follows:
W=wdamp*w (4)
Single particle optimal trajectory is denoted asI.e. individual extreme value, population optimal trajectory are denoted asThat is global extremum, particle movement speed are denoted asWherein, w indicates inertia power
Weight, wdampFor Inertia Weight damping ratio, c1c2For aceleration pulse, meaning is to guarantee have the particle of smaller value attached in target value
It closely hovers, there is the larger value particle to rush at or beyond object boundary, c1Represent self influencing for particle, c2Embody social shadow
It rings, r1 r2It is the random number changed in [0,1] range, for guaranteeing population diversity.
As a preferred technical solution of the present invention, Adaptive Neural-fuzzy Inference algorithm (Adaptive in S3
Network-based Fuzzy Inference System, ANFIS) fuzzy logic unit and neural network are organically combined,
To form a kind of new fuzzy inference system, the premise parameter and consequent parameter in model are by back-propagation algorithm and minimum two
Multiplication study obtains.
As shown in Figure 1, being Adaptive Neuro-fuzzy Inference model framework.
Generally formed by five layers.First layer is blurring layer (Fuzzification Layer), each node in first layer
Corresponding input fuzzy membership expression formula are as follows:
O1,i=μ Ai(x), (5) i=1,2
O1,i=μ Bi-2(y), (6) i=3,4
In formula: x and y respectively indicates the input of i-th of node.O1,iFor fuzzy membership, μ AiWith μ Bi-2It is usually maximum
Value is equal to 1, and minimum value is equal to 0 bell shaped function.Wherein, μ AiCorresponding expression formula are as follows:
In formula: indicating the correspondence parameter of membership function, the parameter in membership function will be determined by training.
The second layer is rule-based reasoning layer (Rule Layer), this layer needs to calculate the excitation density of each rule, is motivated
Intensity expression formula are as follows:
O2,i=wi=μ Ai(x)μBi(y), (8) i=1,2
In formula: Wi indicates the weight of corresponding fuzzy rule.
Third layer is known as normalizing layer (Normalization layer), this layer of main function is swashing each rule
Encourage intensity normalization, the expression formula of normalization layer output result are as follows:
In formula:For the excitation density after the normalization of the i-th rule, contribution of i-th rule to final result is indicated.
4th layer is referred to as blurring layer (Defuzzification layer), this layer calculates the output of every rule, pass through
The weighted results value for calculating each rule embodies contribution of each rule to totally exporting, and exports expression formula are as follows:
In formula: { ai,bi, c } and it is consequent parameter collection.
The last layer is output layer (Summation Layer), and output layer need to calculate all node summations, and model exports table
Up to formula are as follows:
During model learning, ANFIS network can be adaptively calculated and adjust relevant parameter, so that input data
Mapping relations between output label can be expressed more accurately.
As a preferred technical solution of the present invention, the duration of any time period occurred in the S1 is not less than S5
The duration of middle future certain time period.
As a preferred technical solution of the present invention, the historical data of weather includes humidity, air pressure, air in the S2
Density, surface temperature, intensity of solar radiation, average dew point, low clouds coverage rate, medium cloud coverage rate and high cloud coverage rate.
The working principle of the invention: the Feature Selection function of RF is used first, is selected from historical data and weather data
Maximum factor is influenced on photovoltaic power generation power prediction.Then the data after feature selecting are inputted and trains ANFIS mould
Type is avoided falling into Local Minimum, be optimized in training process using PSO in order to enable model comparatively fast obtains optimal solution, training
RF-ANFIS-PSO photovoltaic power generation power prediction algorithm modeling afterwards is completed.It is illustrated in figure 2 the photovoltaic based on RF-ANFIS-PSO
Generated power forecasting model.
Wherein, the parameter selection of this paper PSO is as follows: Population Size PopSize is 50, and maximum number of iterations MaxIt is
1000 generations, autognosis constant c1It is 1, social recognition constant c2It is 2, inertia constant w is 1, and Inertia Weight damping ratio is 0.99.
Daxing District somewhere distributed photovoltaic actual operating data is selected to be verified, distributed photovoltaic total installation of generating capacity
For 700kW, Weather information is from numerical weather forecast (Numerical weather prediction, NWP).Predict target
For next hour photovoltaic output power.Wherein, annual data in 2016 are training data, and data in January, 2017 are test
Data.Test machine uses 10 system of Win and Core i5 processor, and relative program is completed under Matlab 12a environment and is compiled
It translates.
1, feature scoring and selection
It is ranked up analysis using different degree of the RF algorithm to photovoltaic power generation power prediction feature, the input of initial option is special
Sign includes weather data and historical data, and wherein weather data is 9 candidate variables obtained from NWP, including humidity, air pressure,
Atmospheric density, surface temperature, intensity of solar radiation, average dew point, low clouds coverage rate, medium cloud coverage rate and high cloud coverage rate.It goes through
History data are that the history photovoltaic generating system at 24 moment before predicting object time is really contributed.Feature weight is calculated by RF algorithm
After spending, each feature scoring is as shown in Figure 3.
From the figure 3, it may be seen that selected alternative features all have good signature contributions degree.Wherein, intensity of solar radiation, table
Face temperature, low clouds coverage rate, medium cloud coverage rate and high cloud coverage rate be the weather that is affected to photovoltaic power generation power prediction because
Element.Influence of the history photovoltaic data at nearlyr moment adjacent with prediction time to prediction target is bigger.In order to further prove this
The validity that text is selected using RF algorithm characteristics, experiment will carry out the relevance of the maximum variable of predicted impact and photovoltaic power output
Analysis.
From fig. 4, it can be seen that the output power of photovoltaic is bigger when irradiation intensity is bigger, the linear relevant relationship of the two；?
0 DEG C to 25 DEG C section, photovoltaic power output is more steady, and when too high or too low for temperature, the output power of photovoltaic is on a declining curve.By
This is as it can be seen that irradiation level and temperature have larger impact to photovoltaic power generation power prediction accuracy, and also side demonstrates this paper feature
The validity of selection.
In view of selecting less characteristic quantity that can effectively reduce model for distributed photovoltaic power generation power prediction problem
Therefore cycle of training selects intensity of solar radiation, surface temperature, low clouds coverage rate, medium cloud coverage rate and high cloud coverage rate herein
It really contributes with the photovoltaic generation power at preceding 12 moment as the input feature vector after screening.
2, prediction model performance evaluation
For distributed photovoltaic power generation power prediction, prediction model has in terms of computation complexity and memory space
Larger limitation.Therefore, the precision problem of photovoltaic power generation power prediction should be paid close attention to.Simultaneously also to pay close attention to prediction model training when
Between consumption problem with computing resource.For the estimated performance of integration test this paper algorithm, this section contrived experiment is excellent by (1) tradition
The ANFIS model of change method training；(2) without the ANFIS-PSO model of Feature Selection；(3) after RF selects feature
ANFIS-PSO model.Three kinds of models are compared analysis, compare off-line training time corresponding to three kinds of scenes, on-line prediction
Time and precision of prediction are as shown in table 1.
A reuse algorithm performance evaluation of table more than 1
As shown in Table 1, for these three scenes, the model training time is all significantly higher than the on-line prediction time.Therefore, from
Line training link is completed in high-performance calculation unit, and on-line prediction link is completed in embedded unit, both of which difference
The mode of deployment is more suitable for distributed photovoltaic power generation prediction model.Scene (1) and (2) is compared as it can be seen that conventional method was trained
ANFIS model and the precision of prediction difference of PSO training ANFIS model are smaller, can be very fast using the ANFIS algorithm of PSO training
Optimal solution is converged to, optimal time is effectively shortened.Scene (2) and (3) are compared as it can be seen that the photovoltaic after being extracted feature using RF is sent out
Electrical power prediction model, precision of prediction is slightly below the photovoltaic power generation power prediction model of full dose feature training, this is because RF is mentioned
Information after taking has given up the smaller feature of partial contribution degree, but due to reducing the smaller feature of most of contribution degree, model instruction
The white silk time is greatly shortened.It can be seen that in distributed photovoltaic power generation power prediction scene, using RF Feature Selection
ANFIS-PSO model can preferably weigh precision of prediction and prediction two aspect factor of resource consumption.
3, the photovoltaic power generation power prediction situation analysis under different weather
In order to further verify the prediction effect and generalization ability of RF-ANFIS-PSO algorithm, respectively by this paper prediction algorithm
With neural network algorithm (Artificial neural network, ANN), autoregressive moving-average model (Auto-
Regressive and Moving Average Model, ARMA) precision of prediction in test data set compares point
Analysis.Training data and test data are all made of data described above, since every kind of method applicable elements are different, used by calculating
Data set is slightly changed.In addition, also selecting persistency method as a comparison, selecting the true value of previous moment as target
The predicted value at moment.In Skill error assessment index, using persistency method as evaluation criteria algorithm.
Table 2 considers the photovoltaic power generation power prediction error assessment of multiple models
As shown in Table 2, in multiple photovoltaic generation power models, highest precision of prediction is ANFIS-PSO method.Its
In, ARMA algorithm can only usage history sequence data be trained, the missings of Weather information data so that prediction error generally compared with
It is high；ANN algorithm is limited for complex data learning ability, although prediction effect ratio ARMA increases, still with this paper algorithm
There is gap, ANFIS-PSO algorithm can effectively avoid the limitation of traditional models fitting complicated function, effectively improve prediction essence
Degree.For RMSE error criterion, this paper algorithm is reduced compared to the error of ANN algorithm, persistency method and ARMA algorithm
1.54%, 1.3% and 3.68%.For nMAE error criterion, average nMAE of this paper algorithm on seven days photovoltaic test sets is
5.07.Within the corresponding period, deviation power is the 78.97% of ANN, persistence model and ARMA, 58.74% and
68.68%.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (6)
1. the photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO characterized by comprising
S1 selects the somewhere photovoltaic generation power of any time period occurred and the historical data of weather；
S2 carries out Feature Selection to period this area's photovoltaic generation power and weather historical data, selects to light
Lying prostrate generated output influences the historical data of the strong weather of the degree of correlation as weather characteristics data；
S3 is input in trained network the weather characteristics data screened in S2, and training network is fuzzy using adaptive neural network
Reasoning algorithm is built, and training network carries out data training to the weather characteristics data of screening and constructs fuzzy inference system；
S4 when being trained to weather characteristics data, optimizes training network using particle swarm algorithm；
The photovoltaic power generation power prediction model of this area's future certain time period is obtained according to the weather characteristics data of screening；
S5, the weather characteristics data of input this area's future certain time period prediction, obtains this area's future certain time period
Photovoltaic power generation power prediction result.
2. the photovoltaic short term power prediction technique according to claim 1 based on Feature Selection and ANFIS-PSO, feature
Be: the Feature Selection method in the S2 uses the signature contributions degree analysis method of random forest, and the historical data of weather is closed
It is ranked up analysis in the different degree of photovoltaic generation power historical data, calculates the history for influencing the big weather of photovoltaic generation power
Data are as weather characteristics data.
3. the photovoltaic short term power prediction technique according to claim 1 based on Feature Selection and ANFIS-PSO, feature
Be: Adaptive Neural-fuzzy Inference algorithm organically combines fuzzy logic unit and neural network in S3, to form one kind
New fuzzy inference system, premise parameter and consequent parameter in model are obtained by back-propagation algorithm and least square calligraphy learning
?.
4. the photovoltaic short term power prediction technique according to claim 1 based on Feature Selection and ANFIS-PSO, feature
Be: particle swarm optimization algorithm simulates the predation of birds in S4, and in N-dimensional search space, m is grain in primary group
Subnumber amount, position statement of i-th of particle in search space are as follows:
To the motion profile expression formula of particle are as follows:
W=wdamp*w (4)
Single particle optimal trajectory is denoted asI.e. individual extreme value, population optimal trajectory are denoted asThat is global extremum, particle movement speed are denoted asWherein, w indicates inertia power
Weight, wdampFor Inertia Weight damping ratio, c1c2For aceleration pulse, meaning is to guarantee have the particle of smaller value attached in target value
It closely hovers, there is the larger value particle to rush at or beyond object boundary, c1Represent self influencing for particle, c2Embody social shadow
It rings, r1r2It is the random number changed in [0,1] range, for guaranteeing population diversity.
5. the photovoltaic short term power prediction side according to any one of claims 1-4 based on Feature Selection and ANFIS-PSO
Method, it is characterised in that: duration of the duration of any time period occurred in the S1 not less than certain time period following in S5.
6. the photovoltaic short term power prediction side according to any one of claims 1-4 based on Feature Selection and ANFIS-PSO
Method, it is characterised in that: the historical data of weather includes humidity, air pressure, atmospheric density, surface temperature, solar radiation in the S2
Intensity, average dew point, low clouds coverage rate, medium cloud coverage rate and high cloud coverage rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811488151.0A CN109858665A (en) | 2018-12-06 | 2018-12-06 | Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811488151.0A CN109858665A (en) | 2018-12-06 | 2018-12-06 | Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109858665A true CN109858665A (en) | 2019-06-07 |
Family
ID=66890752
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811488151.0A Pending CN109858665A (en) | 2018-12-06 | 2018-12-06 | Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109858665A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110717630A (en) * | 2019-10-09 | 2020-01-21 | 国核电力规划设计研究院有限公司 | Reliability prediction method and device for turbine overspeed protection system |
| CN111815027A (en) * | 2020-06-09 | 2020-10-23 | 山东大学 | Photovoltaic station generated power prediction method and system |
| CN113627546A (en) * | 2021-08-16 | 2021-11-09 | 阳光新能源开发有限公司 | Method for determining reflectivity data, method for determining electric quantity and related device |
| CN116706907A (en) * | 2023-08-09 | 2023-09-05 | 深圳航天科创泛在电气有限公司 | Photovoltaic power generation prediction method based on fuzzy reasoning and related equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268638A (en) * | 2014-09-11 | 2015-01-07 | 广州市香港科大霍英东研究院 | Photovoltaic power generation system power predicting method of elman-based neural network |
| CN105139264A (en) * | 2015-06-06 | 2015-12-09 | 安徽工程大学 | Photovoltaic generation capacity prediction method based on particle swarm algorithm wavelet neural network |
| EP3343496A1 (en) * | 2016-12-28 | 2018-07-04 | Robotina d.o.o. | Method and system for energy management in a facility |
| CN108832663A (en) * | 2018-07-18 | 2018-11-16 | 北京天诚同创电气有限公司 | The prediction technique and equipment of the generated output of micro-capacitance sensor photovoltaic generating system |
-
2018
- 2018-12-06 CN CN201811488151.0A patent/CN109858665A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104268638A (en) * | 2014-09-11 | 2015-01-07 | 广州市香港科大霍英东研究院 | Photovoltaic power generation system power predicting method of elman-based neural network |
| CN105139264A (en) * | 2015-06-06 | 2015-12-09 | 安徽工程大学 | Photovoltaic generation capacity prediction method based on particle swarm algorithm wavelet neural network |
| EP3343496A1 (en) * | 2016-12-28 | 2018-07-04 | Robotina d.o.o. | Method and system for energy management in a facility |
| CN108832663A (en) * | 2018-07-18 | 2018-11-16 | 北京天诚同创电气有限公司 | The prediction technique and equipment of the generated output of micro-capacitance sensor photovoltaic generating system |
Non-Patent Citations (2)
| Title |
|---|
| 刘鑫: "自适应神经网络在异常天气下光伏出力预测中的应用研" * |
| 王介生: "基于粒子群算法的ANFIS模型参数优化" * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110717630A (en) * | 2019-10-09 | 2020-01-21 | 国核电力规划设计研究院有限公司 | Reliability prediction method and device for turbine overspeed protection system |
| CN111815027A (en) * | 2020-06-09 | 2020-10-23 | 山东大学 | Photovoltaic station generated power prediction method and system |
| CN113627546A (en) * | 2021-08-16 | 2021-11-09 | 阳光新能源开发有限公司 | Method for determining reflectivity data, method for determining electric quantity and related device |
| CN116706907A (en) * | 2023-08-09 | 2023-09-05 | 深圳航天科创泛在电气有限公司 | Photovoltaic power generation prediction method based on fuzzy reasoning and related equipment |
| CN116706907B (en) * | 2023-08-09 | 2024-01-23 | 深圳航天科创泛在电气有限公司 | Photovoltaic power generation prediction method based on fuzzy reasoning and related equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Mahmoud et al. | An advanced approach for optimal wind power generation prediction intervals by using self-adaptive evolutionary extreme learning machine | |
| Wang et al. | A review of artificial intelligence based building energy prediction with a focus on ensemble prediction models | |
| CN109858665A (en) | Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO | |
| Huang et al. | One‐day‐ahead hourly forecasting for photovoltaic power generation using an intelligent method with weather‐based forecasting models | |
| CN105046374A (en) | Power interval predication method based on nucleus limit learning machine model | |
| Ramkumar et al. | A short-term solar photovoltaic power optimized prediction interval model based on FOS-ELM algorithm | |
| Feng et al. | Parallel cooperation search algorithm and artificial intelligence method for streamflow time series forecasting | |
| Xing et al. | Research of a novel short-term wind forecasting system based on multi-objective Aquila optimizer for point and interval forecast | |
| CN114362175B (en) | Wind power prediction method and system based on depth certainty strategy gradient algorithm | |
| Zou et al. | Wind turbine power curve modeling using an asymmetric error characteristic-based loss function and a hybrid intelligent optimizer | |
| Rizwan et al. | Artificial intelligence based approach for short term load forecasting for selected feeders at madina saudi arabia | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN114492922A (en) | Medium-and-long-term power generation capacity prediction method | |
| CN106529732A (en) | Carbon emission efficiency prediction method based on neural network and random frontier analysis | |
| CN111008790A (en) | Hydropower station group power generation electric scheduling rule extraction method | |
| CN114781723A (en) | Short-term photovoltaic output prediction method based on multi-model fusion | |
| Zhu | Research on adaptive combined wind speed prediction for each season based on improved gray relational analysis | |
| Yakut et al. | Modeling of energy consumption forecast with economic indicators using particle swarm optimization and genetic algorithm: an application in Turkey between 1979 and 2050 | |
| Yang et al. | Day‐ahead wind power combination forecasting based on corrected numerical weather prediction and entropy method | |
| CN110059871A (en) | Photovoltaic power generation power prediction method | |
| CN109615142A (en) | A kind of wind farm wind velocity combination forecasting method based on wavelet analysis | |
| Ding et al. | A statistical upscaling approach of region wind power forecasting based on combination model | |
| Hua | Power quality prediction of active distribution network based on CNN-LSTM deep learning model | |
| Yousefi et al. | Strategic planning for minimizing CO2 emissions using LP model based on forecasted energy demand by PSO Algorithm and ANN | |
| Taik et al. | Hybrid particle swarm and neural network approach for streamflow forecasting |
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 |