CN114997475A - Short-term prediction method for photovoltaic power generation of fusion model based on Kmeans - Google Patents
Short-term prediction method for photovoltaic power generation of fusion model based on Kmeans Download PDFInfo
- Publication number
- CN114997475A CN114997475A CN202210540910.3A CN202210540910A CN114997475A CN 114997475 A CN114997475 A CN 114997475A CN 202210540910 A CN202210540910 A CN 202210540910A CN 114997475 A CN114997475 A CN 114997475A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- prediction
- fusion model
- fusion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000004927 fusion Effects 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000010248 power generation Methods 0.000 title claims abstract description 24
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000013135 deep learning Methods 0.000 claims abstract description 4
- 238000003066 decision tree Methods 0.000 claims description 33
- 238000012549 training Methods 0.000 claims description 19
- 230000006870 function Effects 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000004913 activation Effects 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000010801 machine learning Methods 0.000 abstract description 2
- 238000012545 processing Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 5
- 230000002354 daily effect Effects 0.000 description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000006386 neutralization reaction Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—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
-
- 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
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a short-term prediction method for photovoltaic power generation of a fusion model based on Kmeans, and belongs to the field of data processing. The method specifically comprises the following steps: preprocessing photovoltaic data; analyzing characteristics; clustering and dividing historical data; respectively constructing XGboost, LightGBM and Multilayer Perceptron models according to the clustered photovoltaic data and the characteristic set thereof, and obtaining respective prediction results; and fusing the XGboost model, the LightGBM model and the Multilayer Perceptron model to obtain a prediction model, and outputting a prediction result. According to the invention, through sufficient data preprocessing, a GBDT model based on an integrated thought is combined, and three machine learning algorithms are fused in a deep learning manner, so that the prediction precision and effectiveness are improved under different climatic conditions compared with a single model.
Description
Technical Field
The invention belongs to the technical field of photovoltaic prediction, and relates to a short-term prediction method of photovoltaic power generation based on a Kmeans fusion model.
Background
Solar energy is considered to be one of the most competitive new energy sources in the future. With the carbon neutralization goal proposed in 2060 years in China, the photovoltaic industry will have a greater development space in the future. The rapid increase of the photovoltaic power generation proportion in China puts higher requirements on power dispatching departments.
Because the generated power of the wind power generation device has great uncertainty, an ideal predicted value cannot be obtained by using the traditional regression method for prediction. There are numerous non-linear correlations of the meteorological data with photovoltaic power, which is precisely a strong term for machine learning to classify and regress. At present, the theory of photovoltaic power generation power prediction is in a widely researched stage, and comprises point prediction, interval prediction, probability prediction and other prediction modes, the main prediction methods comprise neural networks, grey prediction, Markov chains, support vector machines and the like, reliable reference is provided for the scheduling problem of a power grid, the prediction precision of the main problem occurring in extreme or non-sunny days is low, the relation between photovoltaic prediction and meteorological factors and geographic factors is considered, and the comprehensive consideration of an applicable training model and data to improve the prediction precision and shorten the prediction time becomes an important direction of future research.
Disclosure of Invention
The invention aims to solve the problem that the prediction precision difference is large under the weather characteristics of sunny days and non-sunny days, and provides a short-term prediction method of photovoltaic power generation based on a Kmeans fusion model, so that the prediction precision of photovoltaic prediction under different weather conditions is improved.
The technical scheme adopted by the invention is that a short-term prediction method for photovoltaic power generation of a fusion model based on Kmeans specifically comprises the following steps:
step 1: acquiring original data and power generation data corresponding to the data, and preprocessing the data;
step 1.1: setting an irradiance threshold, wherein the irradiance threshold is used for distinguishing the day and the night, and selecting data corresponding to the irradiance threshold;
step 1.2: deleting data with irradiance reaching rated generating power and without power output and data without weather identification;
step 1.3: dividing the data obtained in the step 1.2 into sunny data and rainy data; taking the daily average value, the maximum and minimum values and the extreme difference of the temperature, the humidity, the pressure, the wind speed, the wind direction and the irradiance attributes and the wave crest and trough distances of the generating power in the data obtained in the step (2) as input characteristics; the characteristics obtained by linearly combining time, generated power, temperature, humidity, pressure, wind speed, wind direction and irradiance are also used as input characteristics; the corresponding generated power is an output characteristic;
step 2: carrying out data standardization on the input features obtained in the step 1, and carrying out cluster division on the standardized features by using a Kmeans algorithm;
the method of normalization is shown below:
wherein x represents data before normalization, x * Expressing the normalized data, mu expressing the data mean value under each characteristic attribute, and sigma expressing the data standard deviation under each characteristic attribute;
and step 3: respectively training a sunny model and a rainy model by adopting the following method for prediction;
step 3.1: respectively training an XGboost model, a LighGBM model and a Multilayer Perceptron model by adopting the data obtained in the step (2);
step 3.2: constructing a fusion model based on the idea of deep learning, taking the results obtained by the three models as the input of the fusion model, setting the fusion model with three fully-connected layers, wherein the number of neurons in each layer is 40, 80 and 20 in sequence, and the fusion model adopts a Relu activation function and an adam optimizer;
step 3.3: taking the output prediction results of the three models as the input of the fusion model, and training the fusion model;
and 4, step 4: during actual prediction, firstly, selecting the sunny model or the rainy model obtained in the step 3 for prediction according to whether the prediction data are sunny data or rainy data; and then inputting the preprocessed data into trained XGboost, LighGBM and Multilayer Perceptron models respectively, and inputting output structures of the XGboost, LighGBM and Multilayer Perceptron models into a fusion model, wherein an output result of the fusion model is a final prediction result.
Further, in step 3, the XGBoost model, the ligahgbm model, and the Multilayer Perceptron model are all constructed based on decision trees, where each model includes a plurality of decision trees, and each decision tree is constructed according to the following formula:
wherein Gini (t) represents the probability that a randomly selected sample in the sample set is misclassified, a i Representing categories, wherein k is the number of decision trees, and p (ai | t) is the probability that the sample belongs to the ith category under the condition of t;
in this model, the global penalty function for minimizing the decision tree is set as:
wherein, C a (T) is a function value of the overall loss of the decision tree, and represents the generalization capability of the decision tree, T is a node of the decision tree, | T | is the total number of the nodes, and H t (T) is the information entropy of the sample under the T-th branch, N t Is the number of training samples of the branch, and alpha is a penalty coefficient;
the prediction settings for each model based on the decision tree results are as follows:
wherein,
for the final prediction result of the corresponding model, K represents the number of trees contained in the prediction model, f k And adding the prediction results of the decision trees for the prediction results obtained by each decision tree to obtain a final prediction result.
The invention has the following characteristics and effects:
the invention relates to a short-term prediction method of photovoltaic power generation based on a Kmeans fusion model, which is used for preprocessing the existing historical moment data; considering the influence of weather characteristics, trying to mine the influence of other characteristics on photovoltaic power generation; clustering division is carried out on the photovoltaic power by using a Kmeans algorithm, and a data set is divided into characteristic subsets with obvious characteristic differences; respectively training the two types of data by utilizing XGboost and LightGBM models based on a GBDT decision tree algorithm model and a traditional multilayer perceptron model; the three algorithms are fused and retrained through a multilayer perceptron based on the deep learning idea to obtain a final prediction result, and the prediction accuracy is higher than that of the traditional single-model prediction algorithm; the prediction effect of the fusion model is stably represented in data sets of different weather types, and reference can be provided for the power dispatching system.
Drawings
FIG. 1 is a flow chart of a short-term prediction method of photovoltaic power generation based on a Kmeans fusion model;
FIG. 2 is a partial prediction result of an XGboost single model in the short-term prediction method for photovoltaic power generation based on a Kmeans fusion model;
FIG. 3 is a partial prediction result of a LightGBM single model in the short-term prediction method of the photovoltaic power generation based on the Kmeans fusion model;
FIG. 4 is a partial prediction result of a Multilayer Perceptron single model in the Kmeans-based fusion model photovoltaic power generation short-term prediction method;
FIG. 5 is a partial prediction result of a fusion model in the short-term photovoltaic power generation prediction method based on the Kmeans fusion model.
Detailed Description
The techniques of the present invention are described in further detail below in conjunction with specific embodiments.
As shown in fig. 1, the prediction process of the short-term prediction method of photovoltaic power generation based on the Kmeans fusion model in this embodiment is shown. Performing data preprocessing on the obtained historical meteorological power data and the like to obtain more standard input data, and then expanding a characteristic set according to characteristic climatic factors; and then, carrying out standardization and clustering processing on the data to obtain data types with different weather characteristics, dividing a data set, respectively training by three models, taking the obtained result as the input of a fusion multilayer perceptron, and training the multilayer perceptron model with a double hidden layer structure to obtain a final prediction result.
The embodiment specifically comprises the following steps:
(1) and collecting related data. In the embodiment, historical power generation data and meteorological data of 2017, 1 month to 2018, 12 months in a certain place are taken as research objects, the recording time interval of an initial data set is 15 minutes, partial data are missing, and partial abnormal data are included.
(2) Firstly, preprocessing data to ensure that input data of model training is effective, wherein the definition of abnormal data mainly comes from irradiance and power data which are shown as negative values; approximate non-generated data below an irradiance threshold; deleting data with negative irradiance, correcting data with negative power to be 0, setting an irradiance threshold value to be 6.5 units, and deleting data lower than the irradiance threshold value;
(3) linear combinations among different characteristics are added as new characteristics to reflect the common influence of different weather characteristics on power; and adding the average value, the peak-valley distance, the maximum and minimum value, the corresponding time, the variance and the like of the daily data sequence as new characteristics to reflect the weather type of the day.
(4) Data were normalized using Z-Score as shown below:
wherein
As standard data,. mu. i Is the mean, σ, of the ith feature i Obtained as standard deviation of ith feature
The mean value is 0 and the standard deviation is 1 within the range of-1.5 to 1.5, and the most value change is not needed to be considered when data is newly added.
(5) Further standardizing the data, assuming that the power grid does not carry out photovoltaic power generation at night, setting an observation interval of 08: 00-19: and 00, taking the interval of 15 minutes, considering time factors such as seasons and the like, decomposing the time characteristics of the data, adding characteristics of a certain month and a certain day, facilitating daily operation of the data, dividing the data according to the characteristics of the day, and supplementing missing data through linear interpolation.
(6) Performing Kmeans clustering operation on the obtained standard data set, taking 44 photovoltaic power values every day as a sample vector, setting the clustering number as 2, randomly selecting a clustering center, calculating Euclidean distances of all samples to the clustering center by traversal, defining a loss function as the following formula, updating the clustering center until the clustering center does not change any more, outputting the clustering center and the conditions of the clustered samples,
where J (c, μ) represents the sum of the distances of all data points to the cluster center, x i Represents the ith sample sequence, c i Is x i The cluster to which the cluster belongs to is,
represents the center point corresponding to the cluster, and M is the total number of samples.
(7) Carrying out training set, verification set and test set division on the obtained two types of data;
(8) XGboost and LightGBM are respectively constructed by the clustered photovoltaic power data training sets
And a Multilayer Perceptron model;
(9) the construction of the GBDT algorithm in each model is based on a decision tree, and the decision tree is constructed by using the following formula:
where Gini (t) represents the probability that a randomly selected sample in the sample set is misclassified. ai represents a category, k is the number of decision trees, and p (ai | t) is the probability that a sample belongs to the ith category under the condition of t;
(10) the global penalty function for minimizing the decision tree is set as:
wherein C is a (T) is a function value of the overall loss of the decision tree, the generalization capability of the decision tree is represented, T represents a node of the decision tree, | T | indicates the total number of the nodes, Ht is the information entropy of a sample under the T-th branch, Nt is the number of training samples of the branch, and alpha is a penalty term coefficient.
(11) The prediction for the decision tree result is set as:
wherein,
for the final prediction result, K denotes the number of trees contained in the prediction model, f k And adding the prediction results of the decision trees for the prediction results obtained by the decision trees to obtain a final prediction result.
(12) Constructing a fusion model, setting the number of hidden layers to be three layers, setting the number of neurons to be 40, 80 and 20 respectively, and adopting a Relu activation function and an adam optimizer.
(13) Obtaining a prediction result of the fusion model, wherein the weights distributed by the three models are 0.6, 0.2 and 0.2 respectively for the data cluster with the characteristics in sunny days; for a data cluster characterized by rainy days, the three models are assigned weights of 0.0001, 0.8 and 0.1999 respectively.
(14) Selecting a root Mean Square Error (MSE) and an average absolute percentage error (MAPE) to evaluate a prediction result, wherein the partial missing data Wi is 0, and the calculated MAPE is infinite, so that a daily maximum power value with a screening threshold value of 3% is set, and the calculation is shown as the following formula:
n is the number of samples and is the number of samples,
and W i The predicted value and the true value of the ith sample are respectively.
The invention has the beneficial effects that:
according to the mode, the short-term prediction method of the photovoltaic power generation based on the Kmeans fusion model comprises the steps of preprocessing initial data, adding a new feature set through feature analysis, and clustering and dividing the data through a Kmeans algorithm to obtain two types of data subsets with different weather features; respectively constructing XGboost, LightGBM and Multilayer Perceptron prediction models for different data sets for prediction; then a multi-hidden-layer fusion model is constructed, and the prediction results of the three independent models on the training set are input into the fusion model to obtain a final prediction result, so that the influence of weather data is more emphasized compared with the traditional single prediction model; through the training mode, the nonlinear embodiment capability of the prediction system is increased, the prediction system is stable under two different weather characteristic data sets, and the prediction accuracy is increased.
Examples
Table 1 shows the comparison of the training results of the single model with the prediction error of the prediction results of the proposed fusion model.
Table 1 shows the comparison between the prediction analysis of the single model and the prediction analysis of the fusion model, and it can be seen that the proposed prediction result of the fusion model is optimal, the MSE of the feature data set in sunny days is 0.827, the MSE of the feature data set in rainy days is 0.884, and for MAPE, since a 3% screening threshold is set, the prediction effect on a small power value is not shown, and the prediction effect of the obtained fusion model is smaller in two weather types. The MSE of the prediction result obtained by the fusion model under two kinds of weather is close, and the difference of the prediction model on different weather types is greatly reduced.
Fig. 2-5 are graphs of the prediction results of the present embodiment using different uni-prediction models and fusion models, and it can be seen from the graphs that: the predictive effect of the LightGBM model at the peak and the valley is better than that of a multi-layer perceptron under the same feature set. Therefore, the stability of the prediction precision can be predicted based on the fusion prediction model of neural network relearning under different climatic characteristics.
Claims (2)
1. A short-term prediction method for photovoltaic power generation of a fusion model based on Kmeans specifically comprises the following steps:
step 1: acquiring original data and power generation data corresponding to the data, and preprocessing the data;
step 1.1: setting an irradiance threshold, wherein the irradiance threshold is used for distinguishing the day and the night, and selecting data corresponding to the irradiance threshold;
step 1.2: deleting data with irradiance reaching rated generating power and without power output and data without weather identification;
step 1.3: dividing the data obtained in the step 1.2 into sunny data and rainy data; taking the daily average value, the maximum and minimum values and the extreme difference of the temperature, the humidity, the pressure, the wind speed, the wind direction and the irradiance attributes and the wave crest and trough distances of the generating power in the data obtained in the step (2) as input characteristics; the characteristics obtained by linearly combining time, generated power, temperature, humidity, pressure, wind speed, wind direction and irradiance are also used as input characteristics; the corresponding generated power is an output characteristic;
step 2: carrying out data standardization on the input features obtained in the step 1, and carrying out cluster division on the standardized features by using a Kmeans algorithm;
the method of normalization is shown below:
wherein x represents data before normalization, x * Expressing the normalized data, mu expresses the data mean value under each characteristic attribute, and sigma expresses the data standard deviation under each characteristic attribute;
and step 3: respectively training a sunny model and a rainy model by adopting the following method for prediction;
step 3.1: respectively training an XGboost model, a LighhGBM model and a Multilayer Perceptron model by adopting the data obtained in the step 2;
step 3.2: based on the thought of deep learning, constructing a fusion model, taking results obtained by the three models as input of the fusion model, setting the fusion model with three fully-connected layers, wherein the number of neurons in each layer is 40, 80 and 20 in sequence, and the fusion model adopts a Relu activation function and an adam optimizer;
step 3.3: taking the output prediction results of the three models as the input of the fusion model, and training the fusion model;
and 4, step 4: during actual prediction, firstly, selecting the sunny model or the rainy model obtained in the step 3 for prediction according to whether the prediction data are sunny data or rainy data; and then inputting the preprocessed data into trained XGboost, LighGBM and Multilayer Perceptron models respectively, and inputting output structures of the XGboost, LighGBM and Multilayer Perceptron models into a fusion model, wherein an output result of the fusion model is a final prediction result.
2. The method for short-term photovoltaic power generation prediction based on a Kmeans fusion model in claim 1, wherein the XGboost model, the LighGBM model and the Multilayer Perceptron model in the step 3 are all constructed based on decision trees, wherein each model comprises a plurality of decision trees, and each decision tree is constructed according to the following formula:
wherein Gini (t) represents the probability that a randomly selected sample in the sample set is misclassified, a i Representing categories, wherein k is the number of decision trees, and p (ai | t) is the probability that the sample belongs to the ith category under the condition of t;
in this model, the global penalty function for minimizing the decision tree is set as:
wherein, C a (T) is the overall loss function value of the decision tree and represents the generalization capability of the decision tree, T is the node of the decision tree, | T | is the total number of the nodes, H t (T) is the information entropy of the sample under the T-th branch, N t The number of training samples of the branch is shown, and alpha is a penalty term coefficient;
the prediction settings for each model based on the decision tree results are as follows:
wherein,
for the final prediction result of the corresponding model, K represents the number of trees contained in the prediction model, f k And adding the prediction results of the decision trees for the prediction results obtained by each decision tree to obtain a final prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210540910.3A CN114997475B (en) | 2022-05-17 | 2022-05-17 | Kmeans-based fusion model photovoltaic power generation short-term prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210540910.3A CN114997475B (en) | 2022-05-17 | 2022-05-17 | Kmeans-based fusion model photovoltaic power generation short-term prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114997475A true CN114997475A (en) | 2022-09-02 |
| CN114997475B CN114997475B (en) | 2024-04-09 |
Family
ID=83027662
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210540910.3A Active CN114997475B (en) | 2022-05-17 | 2022-05-17 | Kmeans-based fusion model photovoltaic power generation short-term prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114997475B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118174297A (en) * | 2024-05-13 | 2024-06-11 | 山东浪潮智慧能源科技有限公司 | Photovoltaic power prediction method, system, terminal and medium based on multi-model fusion |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106899300A (en) * | 2017-02-15 | 2017-06-27 | 电子科技大学 | A kind of redundant cyclic averaging method for gradually-appoximant analog-digital converter |
| CN111369070A (en) * | 2020-03-13 | 2020-07-03 | 西安理工大学 | Envelope clustering-based multimode fusion photovoltaic power prediction method |
| CN111414717A (en) * | 2020-03-02 | 2020-07-14 | 浙江大学 | XGboost-L ightGBM-based unit power prediction method |
| US11070056B1 (en) * | 2020-03-13 | 2021-07-20 | Dalian University Of Technology | Short-term interval prediction method for photovoltaic power output |
| CN113868935A (en) * | 2020-06-30 | 2021-12-31 | 北京金风科创风电设备有限公司 | Turbulence intensity calculation method, model training method, device, equipment and medium |
| KR102361423B1 (en) * | 2020-11-10 | 2022-02-14 | (주)위세아이텍 | Artificial intelligence system and method for predicting maintenance demand |
-
2022
- 2022-05-17 CN CN202210540910.3A patent/CN114997475B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106899300A (en) * | 2017-02-15 | 2017-06-27 | 电子科技大学 | A kind of redundant cyclic averaging method for gradually-appoximant analog-digital converter |
| CN111414717A (en) * | 2020-03-02 | 2020-07-14 | 浙江大学 | XGboost-L ightGBM-based unit power prediction method |
| CN111369070A (en) * | 2020-03-13 | 2020-07-03 | 西安理工大学 | Envelope clustering-based multimode fusion photovoltaic power prediction method |
| US11070056B1 (en) * | 2020-03-13 | 2021-07-20 | Dalian University Of Technology | Short-term interval prediction method for photovoltaic power output |
| CN113868935A (en) * | 2020-06-30 | 2021-12-31 | 北京金风科创风电设备有限公司 | Turbulence intensity calculation method, model training method, device, equipment and medium |
| KR102361423B1 (en) * | 2020-11-10 | 2022-02-14 | (주)위세아이텍 | Artificial intelligence system and method for predicting maintenance demand |
Non-Patent Citations (1)
| Title |
|---|
| 杨荣新: "基于Stacking模型融合的光伏发电功率预测", 计算机系统应用, vol. 29, no. 05, 15 May 2020 (2020-05-15), pages 36 - 45 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118174297A (en) * | 2024-05-13 | 2024-06-11 | 山东浪潮智慧能源科技有限公司 | Photovoltaic power prediction method, system, terminal and medium based on multi-model fusion |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114997475B (en) | 2024-04-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lipu et al. | Artificial intelligence based hybrid forecasting approaches for wind power generation: Progress, challenges and prospects | |
| Li et al. | An integrated missing-data tolerant model for probabilistic PV power generation forecasting | |
| CN110717610B (en) | Wind power prediction method based on data mining | |
| CN114792156B (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| Xiong et al. | A short-term wind power forecast method via xgboost hyper-parameters optimization | |
| CN112465256A (en) | Building power consumption prediction method and system based on Stacking model fusion | |
| CN111967675A (en) | Photovoltaic power generation amount prediction method and prediction device | |
| Chen et al. | Research on wind power prediction method based on convolutional neural network and genetic algorithm | |
| CN114399081A (en) | Photovoltaic power generation power prediction method based on weather classification | |
| CN115238956A (en) | Photovoltaic power generation short-term prediction method based on feature engineering | |
| CN114997475B (en) | Kmeans-based fusion model photovoltaic power generation short-term prediction method | |
| Subramanian et al. | Solar power prediction using Machine learning | |
| CN116307111A (en) | Reactive load prediction method based on K-means clustering and random forest algorithm | |
| Yuan et al. | Short-term wind speed forecasting using STLSSVM hybrid model | |
| Wu et al. | Overview of Day-ahead Solar Power Forecasts Based on Weather Classifications and a Case Study in Taiwan | |
| CN116911418A (en) | Photovoltaic power generation power ultra-short-term prediction method based on wavelet transformation and BP neural network optimization | |
| CN115392387A (en) | Low-voltage distributed photovoltaic power generation output prediction method | |
| CN115271242A (en) | Training method, prediction method and device of photovoltaic power generation power prediction model | |
| CN109034453A (en) | A kind of Short-Term Load Forecasting Method based on multiple labeling neural network | |
| CN114492945A (en) | Short-term photovoltaic power prediction method, medium and equipment in electric power market background | |
| Yuan et al. | A novel hybrid short-term wind power prediction framework based on singular spectrum analysis and deep belief network utilized improved adaptive genetic algorithm | |
| Wu et al. | Overview of day-ahead solar power forecasts based on weather classifications | |
| CN113723670A (en) | Photovoltaic power generation power short-term prediction method with variable time window | |
| Kaushik et al. | Performance Analysis of Regression Models in Solar PV Forecasting | |
| Jankauskas et al. | Short-term wind energy forecasting with advanced recurrent neural network models: a comparative study |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |