CN104992248A - Microgrid photovoltaic power station generating capacity combined forecasting method - Google Patents
Microgrid photovoltaic power station generating capacity combined forecasting method Download PDFInfo
- Publication number
- CN104992248A CN104992248A CN201510406307.6A CN201510406307A CN104992248A CN 104992248 A CN104992248 A CN 104992248A CN 201510406307 A CN201510406307 A CN 201510406307A CN 104992248 A CN104992248 A CN 104992248A
- Authority
- CN
- China
- Prior art keywords
- prediction
- msub
- photovoltaic power
- mrow
- result
- 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
- 238000013277 forecasting method Methods 0.000 title abstract 4
- 238000012360 testing method Methods 0.000 claims abstract description 70
- 238000010248 power generation Methods 0.000 claims abstract description 47
- 238000013528 artificial neural network Methods 0.000 claims abstract description 46
- 238000000034 method Methods 0.000 claims abstract description 46
- 238000012706 support-vector machine Methods 0.000 claims abstract description 34
- 239000002245 particle Substances 0.000 claims abstract description 31
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims description 29
- 238000005457 optimization Methods 0.000 claims description 28
- 230000008569 process Effects 0.000 claims description 12
- 230000009466 transformation Effects 0.000 claims description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 description 9
- 238000012545 processing Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 3
- 230000014509 gene expression Effects 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 230000005855 radiation Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000012897 Levenberg–Marquardt algorithm Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
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
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A microgrid photovoltaic power station generating capacity combined forecasting method belongs to the new energy power generation output forecasting field. The method comprises the following steps: 1) taking generating capacity data of a photovoltaic power station in a microgrid as input/output data of a forecasting model; 2) carrying out generating capacity forecasting through wavelet transform and a BP neural network; 3) carrying out generating capacity forecasting by utilizing a particle swarm algorithm to optimize a support vector machine; 4) obtaining model test weights by utilizing a variance-covariance combination method; and 5) calculating combined forecasting result according to the weights and model output corresponding to the new input. Compared with an existing forecasting method, the forecasting method starts from a new angle, avoids considering complex external factors and adopts a combination method, and has the advantages of being good in real-time performance, high in accuracy and high in adaptability and the like.
Description
Technical Field
The invention relates to the technical field of new energy power generation output prediction, in particular to a micro-grid photovoltaic power station power generation amount combined prediction method adopting a neural network and a support vector machine.
Background
With the increasing exhaustion of fossil energy and the increasing awareness of human environmental protection, the energy and power problem becomes more and more important, and micro-grids have become hot spots for research in recent years as new energy sources, especially as a combination and supply mode of power. The micro-grid combines the distributed power supply, the energy storage device, the load device and the control device together to form a unified and controllable small power generation and distribution system, is an effective means for connecting the distributed power supply to the power grid, promotes the development of the distributed power generation technology, and has wide prospects.
Solar photovoltaic power generation as a micro power supply in a micro power grid is a clean, flexible and mature novel power generation mode, and plays an increasingly important role in future energy supply. In recent years, the photovoltaic industry in China is rapidly developed, and the installed capacity of photovoltaic power generation in China is estimated to reach 20GW in 2015, which is located in the front of the world, and 47GW in 2020.
The photovoltaic power generation system is influenced by solar radiation intensity, photovoltaic module temperature, environment and certain random factors, the change of generated energy has randomness and uncontrollable property, and the new problem is brought to the reliable, safe and economic operation of the microgrid when the photovoltaic power generation system is connected to the microgrid. At the moment, the prediction of the power generation amount of the photovoltaic power generation system becomes more important, and the accurate prediction of the power generation amount of the photovoltaic power generation system can provide help for the optimized operation and the planning and scheduling of the micro-grid, so that the influence on the micro-grid and the power system is reduced.
At present, a method for predicting photovoltaic power generation mainly comprises a principle prediction method for predicting power generation by establishing an empirical formula aiming at energy loss in a photoelectric conversion link and an inversion link in the process of solar power generation; establishing a regression prediction method of a regression equation according to historical data of photovoltaic power generation; an artificial intelligence and a novel theoretical method for predicting the generated energy by utilizing an artificial neural network, a support vector machine, a grey theory and the like. The principle prediction method and the regression prediction method are difficult to reflect the dynamic and nonlinear relation between the generated energy and weather and other factors, and the prediction effect on the complex and changeable photovoltaic power generation system is poor; artificial intelligence and a novel theoretical method mostly adopt a single prediction model, and complex external interference factors such as radiation intensity, radiation angle, temperature, cloud cover and the like are considered, so that the prediction precision and flexibility are limited, original data are complicated, and the data analysis and acquisition difficulty is high.
Disclosure of Invention
The invention aims to overcome the defects in the prior art, perform power generation amount prediction of the photovoltaic power station in the microgrid from a new angle, avoid considering complex external factors, only need to obtain power generation amount data of the related photovoltaic power station in the microgrid, and optimize a combined prediction method of a support vector machine by adopting wavelet transformation, a BP neural network and a particle swarm algorithm, thereby realizing the power generation amount prediction of the photovoltaic power station in the microgrid with better real-time property, higher accuracy and stronger adaptability.
In order to achieve the purpose, the invention adopts the technical scheme that:
the utility model provides a little grid photovoltaic power plant generated energy combination prediction method which is characterized in that, includes the following steps:
1) in the microgrid, generating capacity data in a known period of a target photovoltaic power station is used as sample output data, generating capacity data in the same known period of a plurality of photovoltaic power stations connected with the target photovoltaic power station is used as sample input data, generating capacity data in a period to be predicted of the target photovoltaic power station is used as prediction output data, and generating capacity data in the same prediction period of the plurality of photovoltaic power stations connected with the target photovoltaic power station is used as prediction input data;
2) the method comprises the steps of adopting wavelet transformation and a BP neural network to predict the power generation amount of the photovoltaic power station of the microgrid to obtain a first test result and a first predicted result
The specific process is as follows:
201) performing db2 wavelet 1 layer decomposition and single branch reconstruction on the sample input data, the sample output data and the prediction input data;
202) selecting a training sample and a test sample from the input and output data of the sample after wavelet transformation;
203) respectively establishing a high-frequency component BP neural network prediction model and a low-frequency component BP neural network prediction model by utilizing the training samples and the test samples;
204) training the two prediction models by using training samples;
205) inputting a test sample to the two prediction models, and respectively outputting a high-frequency component BP neural network test prediction result and a low-frequency component BP neural network test prediction result which are subjected to algebraic summation to serve as a test result I;
206) calculating the relative prediction error of the first test result;
207) inputting the prediction input data after wavelet transformation to the two prediction models, and respectively outputting a high-frequency component BP neural network prediction result and a low-frequency component BP neural network prediction result which are algebraically summed to serve as a prediction result I;
3) the power generation amount of the photovoltaic power station of the microgrid is predicted by optimizing a support vector machine through a particle swarm algorithm, and a second test result and a second prediction result are obtained
The specific process is as follows:
301) selecting the same training sample and test sample before the corresponding wavelet transform in step 202);
302) establishing a support vector machine prediction model by utilizing a training sample and a test sample, wherein the type is epsilon-SVR, a radial basis kernel function is adopted, and an optimization problem solving algorithm is a sequence minimum optimization algorithm;
303) optimizing a punishment parameter c, a kernel function parameter g and an epsilon loss function parameter p of the support vector machine by adopting a particle swarm optimization;
304) training the prediction model by using a training sample;
305) inputting a test sample to the prediction model, and outputting a particle swarm optimization support vector machine test prediction result as a test result II;
306) calculating the relative prediction error of the test result II;
307) inputting prediction input data to the prediction model, and outputting a prediction result of the particle swarm optimization support vector machine as a prediction result II;
4) weighting of two prediction models using variance-covariance combination
The specific process is as follows:
401) calculating the variance of the results of the step 206) and the step 306);
402) the weight is obtained from the variance, and the following formula (1) is specifically used:
wherein, ω is1And11the variance, ω, of the weights and relative prediction errors of the wavelet transform and BP neural network prediction models, respectively2And22respectively optimizing the weight of a prediction model of the support vector machine and the variance of a relative prediction error for the particle swarm algorithm;
5) multiplying the weights of the two prediction models obtained in the step 4) by corresponding prediction results to obtain combined power generation amount prediction data, specifically using the following formula (2):
f＝ω1f1+ω2f2 (2)
wherein f is combined power generation amount prediction data, ω1And f1Respectively as the weight and prediction result of wavelet transform and BP neural network prediction model I, omega2And f2And respectively optimizing the weight of the support vector machine prediction model and the prediction result II for the particle swarm algorithm.
Compared with the prior art, the technical scheme adopted by the invention can achieve the following technical effects:
(1) the method comprises the steps that the power generation amount of the photovoltaic power station in the microgrid is predicted from a new angle, complex external factors are avoided being considered, and only power generation amount data of the related photovoltaic power station in the microgrid are needed to be obtained;
(2) aiming at the characteristics of strong randomness and volatility of photovoltaic power generation, the wavelet decomposition and single-branch reconstruction are carried out on photovoltaic power generation data, and high-frequency components and low-frequency components are respectively predicted, so that the prediction precision and the adaptability are improved;
(3) the particle swarm optimization algorithm has the advantages of small possibility of entering local minimum, simple algorithm and the like, and the particle swarm optimization algorithm is adopted to optimize the support vector machine, so that the parameter optimization is accelerated, and the prediction precision and the prediction efficiency are improved;
(4) the variance-covariance combined prediction method is adopted, and for each specific prediction, the proportion of the method with high accuracy in the prediction result is dynamically adjusted through the weight, so that the method has better real-time performance, higher accuracy and stronger adaptability.
Drawings
FIG. 1 is a general framework diagram of a method for predicting the power generation combination of a microgrid photovoltaic power station, according to the present invention;
FIG. 2 is a flow chart of wavelet transform and BP neural network prediction;
FIG. 3 is a comparison graph of wavelet transform and BP neural network test predicted values and actual values;
FIG. 4 is a graph of wavelet transform and BP neural network test prediction results versus prediction error;
FIG. 5 is a flow chart of an algorithm for optimizing support vector machine parameters by particle swarm optimization;
FIG. 6 is a comparison graph of the test predicted value and the actual value of the particle swarm optimization support vector machine;
FIG. 7 is a graph of the relative prediction error of the test prediction result of the particle swarm optimization support vector machine;
FIG. 8 is a comparison graph of predicted values and actual values for the combined method;
FIG. 9 is a graph of prediction results versus prediction error for the combined method.
Detailed Description
The present invention will be described in detail with reference to the accompanying drawings and examples, which are provided for illustration only and are not intended to limit the scope of the invention.
Fig. 1 is a schematic general framework diagram of a method for predicting the combination of the generated energy of the photovoltaic power stations in the microgrid, and as shown in fig. 1, the method for predicting the combination of the generated energy of the photovoltaic power stations in the microgrid comprises the following steps:
1) in the microgrid, generating capacity data in a known period of a target photovoltaic power station is used as sample output data, generating capacity data in the same known period of a plurality of photovoltaic power stations connected with the target photovoltaic power station is used as sample input data, generating capacity data in a period to be predicted of the target photovoltaic power station is used as prediction output data, and generating capacity data in the same prediction period of the plurality of photovoltaic power stations connected with the target photovoltaic power station is used as prediction input data.
In the embodiment, the target photovoltaic power station in a certain microgrid demonstration project and the generating capacity data of 6 photovoltaic power stations connected with the target photovoltaic power station in 2013 year each day are selected for establishing, training and testing a prediction model;
after the prediction model is well established, trained and tested, and the weight corresponding to the prediction model is obtained, the generated energy data of the 6 photovoltaic power stations 2014 each year is used as input data, and the generated energy data of the target photovoltaic power station 2014 each year is output when the prediction model is input;
therefore, in this embodiment, the sample output data is the power generation amount data of the target photovoltaic power station 2013, the sample input data is the power generation amount data of the 6 photovoltaic power stations 2013, the predicted output data is the power generation amount data of the target photovoltaic power station 2014, and the predicted input data is the power generation amount data of the 6 photovoltaic power stations 2014.
2) The method comprises the steps of adopting wavelet transformation and a BP neural network to predict the power generation amount of the photovoltaic power station of the microgrid to obtain a first test result and a first prediction result, wherein a flow chart of the wavelet transformation and the BP neural network prediction is shown in figure 2.
The specific process is as follows:
201) performing db2 wavelet 1 layer decomposition and single branch reconstruction on the sample input data, the sample output data and the prediction input data;
in the present embodiment, a db2 wavelet (db 2 wavelet function in Daubechies system) is adopted to decompose and reconstruct the original data signal into a high-frequency component d by a single branch in a 1-layer mode1And a low frequency component a11-layer decomposition and single-branch reconstruction original signal s ═ a1+d1；
202) Selecting a training sample and a test sample from the input and output data of the sample after wavelet transformation;
in the embodiment, the selected training sample is the same as the selected test sample and corresponds to the generated energy data of the whole year every day in 2013 before wavelet transformation;
203) respectively establishing a high-frequency component BP neural network prediction model and a low-frequency component BP neural network prediction model by utilizing the training samples and the test samples;
the two prediction models BP neural networks adopt 3-layer network structures of an input layer, a hidden layer and an output layer;
the number of neurons of the input layer of the BP neural network of the two prediction models is 6, the number of neurons of the output layer is 1, the number of neurons of the hidden layer is obtained by repeated trial and error, the high-frequency component BP neural network prediction model is 12, and the low-frequency component BP neural network prediction model is 18;
the two prediction models BP neural network hidden layer transfer functions both adopt tansig functions, and the function expressions are (1):
the transfer functions of the two prediction models BP neural network output layers both adopt purelin functions, and the function expressions are (2):
f(x)＝x (2)
the two prediction model BP neural network training algorithms both adopt a Levenberg-Marquardt algorithm;
the training samples and the test samples need to be normalized, and the following formula (3) is used specifically:
wherein,is the normalized sample power generation data, xiIs sample power generation amount data before normalization processing, xmaxAnd xminRespectively is the maximum value and the minimum value in the sample power generation amount data before normalization processing;
204) training the two prediction models by using training samples;
205) inputting a test sample to the two prediction models, and respectively outputting a high-frequency component BP neural network test prediction result and a low-frequency component BP neural network test prediction result which are subjected to algebraic summation to serve as a test result I;
inputting a test sample into the two prediction models, firstly testing and predicting to obtain a normalized high-frequency component and low-frequency component prediction result, and performing inverse normalization on the result, wherein the following formula (4) is specifically used:
wherein x isiIs the predicted value of the generated energy after the reverse normalization processing,for normalized prediction results, xmaxAnd xminRespectively is the maximum value and the minimum value in the sample power generation amount data before normalization processing;
after inverse normalization is carried out, a high-frequency component BP neural network test prediction result and a low-frequency component BP neural network test prediction result are obtained respectively, algebraic sum is carried out, the algebraic sum is used as a test result I, and the test result I and an actual value pair are shown in the figure 3;
206) the relative prediction error of the first test result is obtained, and the obtained result is shown in fig. 4, and specifically the following formula (5) is used:
wherein e isiIs the relative prediction error corresponding to the ith test sample, X (i) is the actual value corresponding to the ith test sample,is the predicted value of the ith test sample.
207) Inputting the prediction input data after wavelet transformation to the two prediction models, and respectively outputting a high-frequency component BP neural network prediction result and a low-frequency component BP neural network prediction result which are taken as a first prediction result after algebraic sum, wherein the specific process is similar to the step 205).
3) And adopting a particle swarm optimization support vector machine to predict the power generation amount of the photovoltaic power station of the microgrid, and obtaining a second test result and a second prediction result.
The specific process is as follows:
301) selecting the same training sample and test sample before the corresponding wavelet transform in step 202);
302) establishing a support vector machine prediction model by using a training sample and a test sample, wherein the training sample and the test sample need to be subjected to normalization treatment, and a formula (3) is specifically used;
the prediction model is epsilon-SVR (epsilon-Support Vector Regression), and a function expression of a radial basis kernel function is (6):
K(Xi，Xj)＝exp(-γ||Xi-Xj||)2 (6)
wherein K (-) is a kernel function, Xi、XjInputting a vector for the hyperplane, and taking gamma as a nuclear parameter;
the Optimization problem solving algorithm is a sequence minimum Optimization algorithm (SMO);
303) optimizing a punishment parameter c, a kernel function parameter g and an epsilon loss function parameter p of the support vector machine by adopting a particle swarm optimization;
the fitness function of the particle swarm optimization is the average Mean Square Error (MSE) under Cross Validation (K-fold Cross Validation, K-CV);
as shown in fig. 5, an algorithm flowchart of a particle swarm optimization support vector machine parameter is shown, where a local search parameter c1 of the particle swarm optimization is initialized to 1.5, a global search parameter c2 is initialized to 1.7, the maximum evolutionary number is 200, the maximum population number is 20, a speed and position relation parameter k is initialized to 0.6, a speed update elastic coefficient is initialized to 1, a population update elastic coefficient is initialized to 1, a cross validation parameter v is initialized to 5, a variation range of a penalty parameter c is set to [0.1, 100], a variation range of a kernel function parameter g is set to [0.01, 1000], and a variation range of an epsilon loss function parameter p is set to [0.01, 100 ];
304) training the prediction model by using a training sample;
305) inputting a test sample to the prediction model, and outputting a particle swarm optimization support vector machine test prediction result as a test result II;
in this embodiment, the optimal penalty parameter c of the support vector machine is 2.53158, the kernel function parameter g is 0.01, and the epsilon loss function parameter p is 0.01;
firstly, testing and predicting to obtain a normalized prediction result, and performing inverse normalization on the result by specifically using a formula (4);
after the inverse normalization is carried out, a test prediction result of the particle swarm optimization support vector machine is obtained and is used as a test result II, and a comparison graph of the test result II and an actual value is shown in FIG. 6;
306) calculating the relative prediction error of the second test result, wherein the calculation result is shown in fig. 7, and specifically using a formula (5);
307) inputting prediction input data into the prediction model, and outputting a prediction result of the particle swarm optimization support vector machine as a prediction result II, wherein the specific process is similar to the step 305).
4) The two prediction models are weighted using a variance-covariance combination method.
The specific process is as follows:
401) calculating the variance of the results of step 206) and step 306), specifically calculating the variance using the following formula (7):
wherein, the variance of the relative prediction error, n is the number of the test samples, e1、e2…enFor each of the relative prediction errors corresponding to the test samples,for phases of n test specimensAverage value of prediction error;
402) the weight is obtained from the variance, and the following formula (8) is specifically used:
wherein, ω is1And11the variance, ω, of the weights and relative prediction errors of the wavelet transform and BP neural network prediction models, respectively2And22respectively optimizing the weight of a prediction model of the support vector machine and the variance of a relative prediction error for the particle swarm algorithm;
in this embodiment, the prediction weight ω of the wavelet transform and the BP neural network prediction model is obtained1To 0.524539, the particle swarm optimization support vector machine prediction model prediction weight ω2Is 0.475461.
5) Multiplying the weights of the two prediction models obtained in the step 4) by the corresponding prediction results to obtain combined power generation amount prediction data, specifically using the following formula (9):
f＝ω1f1+ω2f2 (9)
wherein f is combined power generation amount prediction data, ω1And f1Respectively as the weight and prediction result of wavelet transform and BP neural network prediction model I, omega2And f2And respectively optimizing the weight of the support vector machine prediction model and the prediction result II for the particle swarm algorithm.
The ratio of the predicted value to the actual value of the combination method is shown in fig. 8, and the relative prediction error of the prediction result of the combination method is shown in fig. 9.
Table 1 shows the average values of the relative prediction errors of the wavelet transform and BP neural network prediction model, the particle swarm optimization support vector machine prediction model, and the combination method in this embodiment, for monthly and annual prediction errors. The analysis table can be obtained, the combination method can combine the advantages of the two prediction models, the prediction accuracy is biased to be smaller in the two prediction methods, the overall prediction error is reduced, and the advantages of the combination prediction method, such as higher accuracy and stronger adaptability, are further explained.
TABLE 1
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various changes or modifications may be made by those skilled in the art. Any conceivable modifications, improvements or replacements within the idea and principle of the present invention should be included in the protection scope of the present invention.
Claims (1)
1. A method for predicting the generated energy combination of a photovoltaic power station of a micro-grid is characterized by comprising the following steps:
1) in the microgrid, generating capacity data in a known period of a target photovoltaic power station is used as sample output data, generating capacity data in the same known period of a plurality of photovoltaic power stations connected with the target photovoltaic power station is used as sample input data, generating capacity data in a period to be predicted of the target photovoltaic power station is used as prediction output data, and generating capacity data in the same prediction period of the plurality of photovoltaic power stations connected with the target photovoltaic power station is used as prediction input data;
2) the method comprises the steps of adopting wavelet transformation and a BP neural network to predict the power generation amount of the photovoltaic power station of the microgrid to obtain a first test result and a first predicted result
The specific process is as follows:
201) performing db2 wavelet 1 layer decomposition and single branch reconstruction on the sample input data, the sample output data and the prediction input data;
202) selecting a training sample and a test sample from the input and output data of the sample after wavelet transformation;
203) respectively establishing a high-frequency component BP neural network prediction model and a low-frequency component BP neural network prediction model by utilizing the training samples and the test samples;
204) training the two prediction models by using training samples;
205) inputting a test sample to the two prediction models, and respectively outputting a high-frequency component BP neural network test prediction result and a low-frequency component BP neural network test prediction result which are subjected to algebraic summation to serve as a test result I;
206) calculating the relative prediction error of the first test result;
207) inputting the prediction input data after wavelet transformation to the two prediction models, and respectively outputting a high-frequency component BP neural network prediction result and a low-frequency component BP neural network prediction result which are algebraically summed to serve as a prediction result I;
3) the power generation amount of the photovoltaic power station of the microgrid is predicted by optimizing a support vector machine through a particle swarm algorithm, and a second test result and a second prediction result are obtained
The specific process is as follows:
301) selecting the same training sample and test sample before the corresponding wavelet transform in step 202);
302) establishing a support vector machine prediction model by utilizing a training sample and a test sample, wherein the type is epsilon-SVR, a radial basis kernel function is adopted, and an optimization problem solving algorithm is a sequence minimum optimization algorithm;
303) optimizing a punishment parameter c, a kernel function parameter g and an epsilon loss function parameter p of the support vector machine by adopting a particle swarm optimization;
304) training the prediction model by using a training sample;
305) inputting a test sample to the prediction model, and outputting a particle swarm optimization support vector machine test prediction result as a test result II;
306) calculating the relative prediction error of the test result II;
307) inputting prediction input data to the prediction model, and outputting a prediction result of the particle swarm optimization support vector machine as a prediction result II;
4) weighting of two prediction models using variance-covariance combination
The specific process is as follows:
401) calculating the variance of the results of the step 206) and the step 306);
402) the weight is obtained from the variance, and the following formula (1) is specifically used:
wherein, ω is1And11the variance, ω, of the weights and relative prediction errors of the wavelet transform and BP neural network prediction models, respectively2And22respectively optimizing the weight of a prediction model of the support vector machine and the variance of a relative prediction error for the particle swarm algorithm;
5) multiplying the weights of the two prediction models obtained in the step 4) by corresponding prediction results to obtain combined power generation amount prediction data, specifically using the following formula (2):
f＝ω1f1+ω2f2 (2)
wherein f is combined power generation amount prediction data, ω1And f1Respectively as the weight and prediction result of wavelet transform and BP neural network prediction model I, omega2And f2And respectively optimizing the weight of the support vector machine prediction model and the prediction result II for the particle swarm algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406307.6A CN104992248A (en) | 2015-07-07 | 2015-07-07 | Microgrid photovoltaic power station generating capacity combined forecasting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510406307.6A CN104992248A (en) | 2015-07-07 | 2015-07-07 | Microgrid photovoltaic power station generating capacity combined forecasting method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104992248A true CN104992248A (en) | 2015-10-21 |
Family
ID=54304060
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510406307.6A Pending CN104992248A (en) | 2015-07-07 | 2015-07-07 | Microgrid photovoltaic power station generating capacity combined forecasting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104992248A (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105528517A (en) * | 2015-12-01 | 2016-04-27 | 北京国电通网络技术有限公司 | Photovoltaic power station power prediction method and system on basis of neural network and wavelet decomposition |
| CN105913151A (en) * | 2016-04-12 | 2016-08-31 | 河海大学常州校区 | Photovoltaic power station power generation amount predication method based on adaptive mutation particle swarm and BP network |
| CN106202700A (en) * | 2016-07-07 | 2016-12-07 | 西安美林数据技术股份有限公司 | A kind of photovoltaic generation exert oneself prediction data analysing method |
| CN107358059A (en) * | 2017-09-01 | 2017-11-17 | 北京天诚同创电气有限公司 | Short-term photovoltaic energy Forecasting Methodology and device |
| CN107562992A (en) * | 2017-07-25 | 2018-01-09 | 华南理工大学 | A kind of Photovoltaic array maximum power tracking method based on SVM and particle cluster algorithm |
| CN108259099A (en) * | 2018-01-25 | 2018-07-06 | 湘潭大学 | A kind of TD-SCDMA base station electromagnetic radiation Forecasting Methodology |
| CN108255180A (en) * | 2018-01-23 | 2018-07-06 | 中南大学 | A kind of intelligence manufacture environment robot and the vehicle computational intelligence driving means of delivery and system |
| CN108280518A (en) * | 2018-01-23 | 2018-07-13 | 中南大学 | A kind of distributed environment robot and the vehicle mobile interchange means of delivery and system |
| CN108287548A (en) * | 2018-01-23 | 2018-07-17 | 中南大学 | A kind of automation guide rail toter and the robot collaboration means of delivery and system |
| CN109066707A (en) * | 2018-09-11 | 2018-12-21 | 东南大学 | One kind being based on NARMA-L2 model energy management method for micro-grid |
| CN109446230A (en) * | 2018-07-27 | 2019-03-08 | 中国计量大学 | A kind of big data analysis system and method for photovoltaic power generation influence factor |
| CN109636054A (en) * | 2018-12-21 | 2019-04-16 | 电子科技大学 | Solar energy power generating amount prediction technique based on classification and error combination prediction |
| CN111367349A (en) * | 2018-12-26 | 2020-07-03 | 株洲中车时代电气股份有限公司 | Photovoltaic MPPT control method and system based on prediction model |
| WO2020228568A1 (en) * | 2019-05-14 | 2020-11-19 | 京东方科技集团股份有限公司 | Method for training power generation amount prediction model of photovoltaic power station, power generation amount prediction method and device of photovoltaic power station, training system, prediction system and storage medium |
| CN112418476A (en) * | 2019-08-23 | 2021-02-26 | 武汉剑心科技有限公司 | Ultra-short-term power load prediction method |
| WO2021051332A1 (en) * | 2019-09-19 | 2021-03-25 | 深圳市桥博设计研究院有限公司 | Bridge seismic damage monitoring method based on wavelet neural network and support vector machine |
| CN116362418A (en) * | 2023-05-29 | 2023-06-30 | 天能电池集团股份有限公司 | Online prediction method for application-level manufacturing capacity of intelligent factory of high-end battery |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663513A (en) * | 2012-03-13 | 2012-09-12 | 华北电力大学 | Combination forecast modeling method of wind farm power by using gray correlation analysis |
| CN102938093A (en) * | 2012-10-18 | 2013-02-20 | 安徽工程大学 | Wind power forecasting method |
| CN103023065A (en) * | 2012-11-20 | 2013-04-03 | 广东工业大学 | Wind power short-term power prediction method based on relative error entropy evaluation method |
| CN103218674A (en) * | 2013-04-07 | 2013-07-24 | 国家电网公司 | Method for predicating output power of photovoltaic power generation system based on BP (Back Propagation) neural network model |
-
2015
- 2015-07-07 CN CN201510406307.6A patent/CN104992248A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663513A (en) * | 2012-03-13 | 2012-09-12 | 华北电力大学 | Combination forecast modeling method of wind farm power by using gray correlation analysis |
| CN102938093A (en) * | 2012-10-18 | 2013-02-20 | 安徽工程大学 | Wind power forecasting method |
| CN103023065A (en) * | 2012-11-20 | 2013-04-03 | 广东工业大学 | Wind power short-term power prediction method based on relative error entropy evaluation method |
| CN103218674A (en) * | 2013-04-07 | 2013-07-24 | 国家电网公司 | Method for predicating output power of photovoltaic power generation system based on BP (Back Propagation) neural network model |
Non-Patent Citations (2)
| Title |
|---|
| 彭怀午 等: "基于组合预测方法的风电场短期风速预测", 《太阳能学报》 * |
| 杨锡运 等: "基于熵权法的光伏输出功率组合预测模型", 《太阳能学报》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105528517B (en) * | 2015-12-01 | 2019-02-26 | 北京国电通网络技术有限公司 | Based on neural network, wavelet decomposition predicting power of photovoltaic plant method, system |
| CN105528517A (en) * | 2015-12-01 | 2016-04-27 | 北京国电通网络技术有限公司 | Photovoltaic power station power prediction method and system on basis of neural network and wavelet decomposition |
| CN105913151A (en) * | 2016-04-12 | 2016-08-31 | 河海大学常州校区 | Photovoltaic power station power generation amount predication method based on adaptive mutation particle swarm and BP network |
| CN106202700A (en) * | 2016-07-07 | 2016-12-07 | 西安美林数据技术股份有限公司 | A kind of photovoltaic generation exert oneself prediction data analysing method |
| CN107562992A (en) * | 2017-07-25 | 2018-01-09 | 华南理工大学 | A kind of Photovoltaic array maximum power tracking method based on SVM and particle cluster algorithm |
| CN107562992B (en) * | 2017-07-25 | 2020-04-28 | 华南理工大学 | Photovoltaic array maximum power tracking method based on SVM and particle swarm algorithm |
| CN107358059A (en) * | 2017-09-01 | 2017-11-17 | 北京天诚同创电气有限公司 | Short-term photovoltaic energy Forecasting Methodology and device |
| CN108280518B (en) * | 2018-01-23 | 2020-06-26 | 中南大学 | Distributed environment robot and vehicle mobile interconnection carrying method and system |
| CN108255180B (en) * | 2018-01-23 | 2019-05-03 | 中南大学 | A kind of intelligence manufacture environment robot and the vehicle computational intelligence driving means of delivery and system |
| CN108280518A (en) * | 2018-01-23 | 2018-07-13 | 中南大学 | A kind of distributed environment robot and the vehicle mobile interchange means of delivery and system |
| CN108287548A (en) * | 2018-01-23 | 2018-07-17 | 中南大学 | A kind of automation guide rail toter and the robot collaboration means of delivery and system |
| CN108255180A (en) * | 2018-01-23 | 2018-07-06 | 中南大学 | A kind of intelligence manufacture environment robot and the vehicle computational intelligence driving means of delivery and system |
| CN108259099A (en) * | 2018-01-25 | 2018-07-06 | 湘潭大学 | A kind of TD-SCDMA base station electromagnetic radiation Forecasting Methodology |
| CN109446230A (en) * | 2018-07-27 | 2019-03-08 | 中国计量大学 | A kind of big data analysis system and method for photovoltaic power generation influence factor |
| CN109066707A (en) * | 2018-09-11 | 2018-12-21 | 东南大学 | One kind being based on NARMA-L2 model energy management method for micro-grid |
| CN109066707B (en) * | 2018-09-11 | 2021-03-02 | 东南大学 | Micro-grid energy management method based on NARMA-L2 model |
| CN109636054A (en) * | 2018-12-21 | 2019-04-16 | 电子科技大学 | Solar energy power generating amount prediction technique based on classification and error combination prediction |
| CN111367349A (en) * | 2018-12-26 | 2020-07-03 | 株洲中车时代电气股份有限公司 | Photovoltaic MPPT control method and system based on prediction model |
| WO2020228568A1 (en) * | 2019-05-14 | 2020-11-19 | 京东方科技集团股份有限公司 | Method for training power generation amount prediction model of photovoltaic power station, power generation amount prediction method and device of photovoltaic power station, training system, prediction system and storage medium |
| CN112418476A (en) * | 2019-08-23 | 2021-02-26 | 武汉剑心科技有限公司 | Ultra-short-term power load prediction method |
| WO2021051332A1 (en) * | 2019-09-19 | 2021-03-25 | 深圳市桥博设计研究院有限公司 | Bridge seismic damage monitoring method based on wavelet neural network and support vector machine |
| CN116362418A (en) * | 2023-05-29 | 2023-06-30 | 天能电池集团股份有限公司 | Online prediction method for application-level manufacturing capacity of intelligent factory of high-end battery |
| CN116362418B (en) * | 2023-05-29 | 2023-08-22 | 天能电池集团股份有限公司 | Online prediction method for application-level manufacturing capacity of intelligent factory of high-end battery |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104992248A (en) | Microgrid photovoltaic power station generating capacity combined forecasting method | |
| CN110070226B (en) | Photovoltaic power prediction method and system based on convolutional neural network and meta-learning | |
| CN103559561B (en) | A kind of ultra-short term prediction method of photovoltaic plant irradiance | |
| CN108053061B (en) | Solar irradiance prediction method based on improved convolutional neural network | |
| Graditi et al. | Energy yield estimation of thin-film photovoltaic plants by using physical approach and artificial neural networks | |
| Zheng et al. | Stochastic optimization of cost-risk for integrated energy system considering wind and solar power correlated | |
| CN105426956A (en) | Ultra-short-period photovoltaic prediction method | |
| CN107220723A (en) | A kind of predicting power of photovoltaic plant method | |
| Ghanbarzadeh et al. | Solar radiation forecasting based on meteorological data using artificial neural networks | |
| CN104050517A (en) | Photovoltaic power generation forecasting method based on GRNN | |
| CN102479339A (en) | Method and system for forecasting short-term wind speed of wind farm based on hybrid neural network | |
| CN112052913B (en) | Distributed photovoltaic power station power data virtual acquisition method | |
| CN105184679A (en) | Wind power plant power multi-time scale joint prediction method | |
| Mensour et al. | Modeling of solar energy potential in Souss-Massa area-Morocco, using intelligence Artificial Neural Networks (ANNs) | |
| CN108537359A (en) | Photovoltaic power generation power prediction based on wavelet analysis and BP neural network | |
| Doorga et al. | Forecasting mesoscale distribution of surface solar irradiation using a proposed hybrid approach combining satellite remote sensing and time series models | |
| Alblawi et al. | PV solar power forecasting based on hybrid MFFNN-ALO | |
| De Giorgi et al. | Short-term power forecasting by statistical methods for photovoltaic plants in south Italy | |
| CN104361399A (en) | Solar irradiation intensity minute-scale predication method | |
| CN112116171A (en) | Novel photovoltaic power generation power prediction method based on neural network | |
| Alharbi | Daily global solar radiation forecasting using ANN and extreme learning machine: A case study in Saudi Arabia | |
| CN103927598A (en) | Photovoltaic generation power prediction method based on self-learning radial basis function | |
| Kamthania et al. | Determination of efficiency of hybrid photovoltaic thermal air collectors using artificial neural network approach for different PV technology | |
| Premalatha et al. | Estimation of global solar radiation in India using artificial neural network | |
| Feng et al. | Research on the prediction of short-term wind power based on wavelet neural networks |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20151021 |
|
| RJ01 | Rejection of invention patent application after publication |