CN115271191A - Method for constructing wind-light power combined prediction model System and prediction method - Google Patents
Method for constructing wind-light power combined prediction model System and prediction method Download PDFInfo
- Publication number
- CN115271191A CN115271191A CN202210851575.9A CN202210851575A CN115271191A CN 115271191 A CN115271191 A CN 115271191A CN 202210851575 A CN202210851575 A CN 202210851575A CN 115271191 A CN115271191 A CN 115271191A
- Authority
- CN
- China
- Prior art keywords
- wind
- photovoltaic power
- loss value
- power
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000003062 neural network model Methods 0.000 claims abstract description 85
- 238000005457 optimization Methods 0.000 claims abstract description 59
- 238000010276 construction Methods 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims description 61
- 238000012545 processing Methods 0.000 claims description 41
- 238000004422 calculation algorithm Methods 0.000 claims description 31
- 238000012937 correction Methods 0.000 claims description 25
- 239000013598 vector Substances 0.000 claims description 21
- 238000007781 pre-processing Methods 0.000 claims description 18
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 16
- 230000009467 reduction Effects 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 10
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000012847 principal component analysis method Methods 0.000 claims description 3
- 230000008878 coupling Effects 0.000 abstract description 4
- 238000010168 coupling process Methods 0.000 abstract description 4
- 238000005859 coupling reaction Methods 0.000 abstract description 4
- 230000002452 interceptive effect Effects 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 210000002569 neuron Anatomy 0.000 description 5
- 230000004913 activation Effects 0.000 description 4
- 238000010248 power generation Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000000053 physical method Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
- 238000012546 transfer 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- 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
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Educational Administration (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a construction method, a system and a prediction method of a wind-solar power joint prediction model, which comprise the following steps: performing iterative optimization on the model parameter values of the fully-connected neural network model by using the sample data set, until the optimization times reach preset times or a joint prediction loss value corresponding to the current fully-connected neural network model meets a preset range, taking the trained model as a wind-solar power joint prediction model; the combined prediction loss value is obtained by combining the wind power prediction loss value and the photovoltaic power prediction loss value. According to the method, the model parameters are iteratively optimized by combining the real-time prediction loss values of the wind power and the photovoltaic power, so that the model can learn the interactive coupling relation between the wind power and the photovoltaic power, and the prediction precision of the model is further improved.
Description
Technical Field
The invention relates to the technical field of wind and light power prediction, in particular to a construction method, a system and a prediction method of a wind and light power combined prediction model.
Background
The wind and light power generation power prediction is a necessary premise for ensuring the safe and stable operation of a high-proportion new energy power system. From the self-perspective of the wind and light power station, the accurate power prediction result can improve the operation and maintenance level of the wind power station and the photovoltaic power station and reduce the energy abandon rate on one hand; on the other hand, the method is a precondition of participating in electric power market trading and has important significance for improving market competitiveness of wind power and photovoltaic.
Traditional wind and solar power generation power prediction can be mainly divided into a physical method and a statistical method. When the wind power is predicted by adopting a physical method, firstly, the influence of terrain, altitude and surface roughness change on a flow field of a wind power plant is considered, a computational fluid mechanics model of the wind power plant is established, the wind condition at the hub height position of each wind power unit is calculated by taking an initial numerical weather forecast result as input, then, the wind speed is converted into power through a unit wind speed-power curve, and the power of each unit is added to obtain the power of the whole plant. When the photovoltaic power generation power is predicted by adopting a physical method, the initial numerical weather forecast result is taken as input, the geographic information and the component parameters of the photovoltaic power station are combined, the output of each photovoltaic array is calculated based on physical equations such as a solar irradiation transfer equation and a photovoltaic component operation equation, and the output is added to obtain the power of the whole field. The statistical method adopts a statistical learning algorithm to establish a mapping relation between the operation data of the wind and light power station or between the numerical weather forecast data and the operation data, and carries out power prediction based on the established mapping model. Common methods include autoregressive moving average models, kalman filtering, support vector machines, correlation vector machines, least squares, random forests, artificial neural networks, deep learning, combinatorial methods, and the like. However, the traditional wind and photovoltaic power generation power prediction is only to construct a wind power or photovoltaic prediction model respectively, or the wind power prediction model and the photovoltaic prediction model have independence, and the prediction of wind power and photovoltaic power cannot be realized simultaneously.
Disclosure of Invention
The invention provides a construction method, a system and a prediction method of a wind-solar power combined prediction model, which optimize model parameters based on a wind power prediction loss value and a photovoltaic power prediction loss value, so that the model can learn the interactive coupling relation between the wind power and the photovoltaic power, and the prediction precision of the model is improved.
In order to solve the technical problem, an embodiment of the present invention provides a method for constructing a wind-solar power joint prediction model, including:
respectively observing a plurality of wind power plants and a plurality of photovoltaic power stations in a preset area in real time to obtain a sample data set; the sample data set comprises a plurality of wind power plant data corresponding to each wind power plant, a plurality of photovoltaic power plant data corresponding to each photovoltaic power plant, first total wind power of the preset area at different moments and first total photovoltaic power of the preset area at different moments;
according to the Adam algorithm, carrying out iterative optimization on model parameter values of a pre-built fully-connected neural network model by using the sample data set until the optimization times reach preset times or a combined prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and taking the current fully-connected neural network model as a wind-solar power combined prediction model; and the combined prediction loss value is calculated according to a preset algorithm and by combining the wind power prediction loss value and the photovoltaic power prediction loss value.
By implementing the embodiment of the invention, in the process of carrying out iterative optimization on the model parameter values of the pre-built fully-connected neural network model, the wind power prediction loss value and the photovoltaic power prediction loss value are combined, and the joint prediction loss value corresponding to the current fully-connected neural network model is obtained through calculation and is used as the approximate target value of the iterative optimization, so that the model can learn the interactive coupling relation between the wind power and the photovoltaic power, and the prediction precision of the model can be optimized for multiple times.
As a preferred scheme, the obtaining of the joint prediction loss value specifically includes:
preprocessing all the wind power plant data and all the photovoltaic power station data in the sample data set to obtain corresponding data to be input;
inputting the data to be input into the current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area;
according to a preset algorithm, calculating to obtain a wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain a photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power;
and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain the combined prediction loss value.
According to the preferred scheme of the embodiment of the invention, through the current fully-connected neural network model, the predicted values corresponding to all wind power plant data and all photovoltaic power plant data in the sample data set, namely a plurality of second total wind power and a plurality of second total photovoltaic power, are output, and in combination with all first total wind power and first total photovoltaic power in the sample data set, the corresponding wind power predicted loss value and photovoltaic power predicted loss value are calculated to reflect the difference between the measured value and the predicted value and serve as the representation of the prediction accuracy of the wind and photovoltaic power combined prediction model. In addition, considering that the wind power field data and the photovoltaic power station data have different influence degrees on wind and light power combined prediction and the detection precision of the photovoltaic power field data and the photovoltaic power station data are different, the wind power prediction loss value and the photovoltaic power prediction loss value are weighted respectively, different weight coefficients are given to different prediction loss values, and the prediction precision of the wind and light power combined prediction model is further improved.
As a preferred scheme, the preprocessing is performed on all the wind farm data and all the photovoltaic power station data in the sample data set to obtain corresponding data to be input, and specifically:
forming a corresponding first variable matrix by using all the wind power plant data and all the photovoltaic power station data in the sample data set;
performing decentralized processing on the first variable matrix to obtain a corresponding second variable matrix, and calculating to obtain a corresponding covariance matrix according to the second variable matrix;
calculating a plurality of eigenvalues corresponding to the covariance matrix and eigenvectors corresponding to the eigenvalues by adopting an eigenvalue decomposition method;
and arranging all the eigenvalues in a descending order, selecting the first k eigenvalues according to an arrangement result, and performing projection transformation on the first variable matrix by using the eigenvectors corresponding to the first k eigenvalues to obtain the corresponding data to be input, thereby realizing the dimension reduction optimization processing of the first variable matrix.
According to the implementation of the preferred scheme of the embodiment of the invention, all wind power plant data and all photovoltaic power plant data are subjected to dimensionality reduction optimization processing, the multidimensional variable matrix comprising a large amount of initial data is converted into to-be-input data comprising main components capable of reflecting key information of the initial data, the data of an input model is prevented from comprising a large amount of redundant information, and therefore the prediction rate and the prediction precision of the model are optimized.
As a preferred scheme, according to the Adam algorithm, performing iterative optimization on model parameter values of a pre-built fully-connected neural network model by using the sample data set until the number of times of optimization reaches a preset number of times or a joint prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and then using the current fully-connected neural network model as a wind-light power joint prediction model specifically:
according to an Adam algorithm, calculating to obtain a parameter correction value corresponding to the fully-connected neural network model by combining all the wind power plant data, all the photovoltaic power station data, all the first total wind power and all the first total photovoltaic power in the sample data set;
performing iterative optimization processing on the model parameter value of the fully-connected neural network model by using the parameter correction value, updating the model parameter value of the fully-connected neural network model according to the current parameter correction value during each iterative optimization processing, and acquiring the joint prediction loss value corresponding to the current fully-connected neural network model until the optimization times reach preset times or the current joint prediction loss value meets a preset range, and taking the current fully-connected neural network model as the wind-light power joint prediction model;
wherein the fully connected neural network model comprises an input layer, an output layer and a plurality of hidden layers, the model parameter values comprise network weights and bias vectors corresponding to the hidden layers.
According to the optimal scheme of the embodiment of the invention, parameter correction values corresponding to the fully-connected neural network model are calculated by using an Adam algorithm and a sample data set, and the model parameter values of the fully-connected neural network model are updated according to the current parameter correction values during each iterative optimization process, so that the model parameter values of the fully-connected neural network model can be optimized, and meanwhile, based on the preset optimization times or the preset combined prediction loss value range, the model parameter values can be trained and optimized for multiple times to gradually improve the prediction performance of the model.
In order to solve the same technical problem, an embodiment of the present invention further provides a system for constructing a wind-solar power joint prediction model, including:
the data acquisition module is used for respectively carrying out real-time observation on a plurality of wind power plants and a plurality of photovoltaic power plants in a preset area, to obtain a sample data set; the sample data set comprises a plurality of wind power plant data corresponding to each wind power plant, a plurality of photovoltaic power plant data corresponding to each photovoltaic power plant, first total wind power of the preset area at different moments and first total photovoltaic power of the preset area at different moments;
the model construction module is used for carrying out iterative optimization on model parameter values of a pre-built fully-connected neural network model by utilizing the sample data set according to an Adam algorithm until the optimization times reach preset times or a combined prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and then taking the current fully-connected neural network model as a wind-light power combined prediction model; and the combined prediction loss value is calculated according to a preset algorithm and by combining the wind power prediction loss value and the photovoltaic power prediction loss value.
As a preferred scheme, the system for constructing the wind-solar power joint prediction model further includes:
the loss analysis module is used for preprocessing all the wind power plant data and all the photovoltaic power plant data in the sample data set to obtain corresponding data to be input; inputting the data to be input into the current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area; according to a preset algorithm, calculating to obtain the wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain the photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power; and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain the combined prediction loss value.
As a preferred scheme, the loss analysis module specifically includes:
the preprocessing unit is used for forming a corresponding first variable matrix by utilizing all the wind power plant data and all the photovoltaic power plant data in the sample data set; performing decentralized processing on the first variable matrix to obtain a corresponding second variable matrix, calculating to obtain a corresponding covariance matrix according to the second variable matrix; calculating a plurality of eigenvalues corresponding to the covariance matrix and eigenvectors corresponding to the eigenvalues by adopting an eigenvalue decomposition method; arranging all the eigenvalues in a descending order, selecting the first k eigenvalues according to an arrangement result, and performing projection transformation on the first variable matrix by using the eigenvectors corresponding to the first k eigenvalues to obtain the corresponding data to be input, thereby realizing the dimension reduction optimization processing of the first variable matrix;
the loss analysis unit is used for inputting the data to be input into the current fully-connected neural network model so as to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area; according to a preset algorithm, calculating to obtain the wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain the photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power; and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain the combined prediction loss value.
As a preferred scheme, the model building module specifically includes:
the data processing unit is used for calculating to obtain a parameter correction value corresponding to the fully-connected neural network model by combining all the wind power plant data, all the photovoltaic power station data, all the first total wind power and all the first total photovoltaic power in the sample data set according to an Adam algorithm;
the model construction optimization unit is used for performing iterative optimization processing on model parameter values of the fully-connected neural network model by using the parameter correction values, updating the model parameter values of the fully-connected neural network model according to the current parameter correction values during each iterative optimization processing, acquiring the joint prediction loss value corresponding to the current fully-connected neural network model, and taking the current fully-connected neural network model as the wind and light power joint prediction model until the optimization times reach preset times or the current joint prediction loss value meets a preset range; wherein the fully connected neural network model comprises an input layer, an output layer and a plurality of hidden layers, the model parameter values comprise network weights and bias vectors corresponding to the hidden layers.
In order to solve the same technical problem, an embodiment of the present invention further provides a wind-solar power joint prediction method, including:
acquiring a plurality of wind power plant data corresponding to a plurality of wind power plants and a plurality of photovoltaic power station data corresponding to a plurality of photovoltaic power stations in an area to be predicted, and preprocessing all the wind power plant data and all the photovoltaic power station data to obtain corresponding matrixes to be input;
inputting the matrix to be input into a wind-solar power joint prediction model, outputting a power prediction vector corresponding to the region to be predicted, and acquiring the total wind power and the total photovoltaic power of the region to be predicted at different moments according to the power prediction vector to realize wind-solar power joint prediction of the region to be predicted;
the wind power plant data are wind speed, irradiance and temperature of each wind power plant at different moments, the photovoltaic power plant data are wind speed, irradiance and temperature of each photovoltaic power plant at different moments, and the wind-light power joint prediction model is constructed by using the construction method of the wind-light power joint prediction model.
As a preferred scheme, the preprocessing is performed on all the wind farm data and all the photovoltaic power station data to obtain corresponding matrices to be input, and specifically includes:
forming a corresponding variable matrix by using all the wind power plant data and all the photovoltaic power station data in the sample data set;
and performing dimension reduction optimization processing on the variable matrix by adopting a principal component analysis method to obtain the corresponding matrix to be input.
Drawings
FIG. 1: the invention provides a flow schematic diagram of a construction method of a wind-solar power joint prediction model in an embodiment;
FIG. 2: the invention provides a structural schematic diagram of a construction system of a wind-solar power joint prediction model;
FIG. 3: the invention provides a flow diagram of a wind-solar power joint prediction method.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, it is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The first embodiment is as follows:
referring to fig. 1, a method for constructing a wind-solar power joint prediction model according to an embodiment of the present invention includes steps S1 to S2, where the steps are as follows:
the method comprises the following steps that S1, a plurality of wind power plants and a plurality of photovoltaic power plants in a preset area are observed in real time respectively to obtain a sample data set; the sample data set comprises a plurality of wind power plant data corresponding to each wind power plant, a plurality of photovoltaic power station data corresponding to each photovoltaic power station, first total wind power of a preset area at different moments and first total photovoltaic power of the preset area at different moments.
In this embodiment, in one day, every 15 minutes, i wind farms and j photovoltaic power stations in a preset area are respectively observed in real time, so as to obtain 96 wind farm data corresponding to each wind farm and 96 photovoltaic power station data corresponding to each photovoltaic power station.
It should be noted that the wind farm data includes, but is not limited to, wind speed, irradiance, and temperature corresponding to the wind farm, and the photovoltaic plant data includes, but is not limited to, wind speed, irradiance, and temperature of the photovoltaic plant.
S2, performing iterative optimization on model parameter values of the pre-built fully-connected neural network model by using a sample data set according to an Adam algorithm until the optimization times reach preset times or a combined prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and taking the current fully-connected neural network model as a wind-solar power combined prediction model; the combined prediction loss value is obtained by calculation according to a preset algorithm and by combining the wind power prediction loss value and the photovoltaic power prediction loss value.
Preferably, step S2 includes steps S21 to S22, and each step is as follows:
step S21, according to an Adam algorithm, calculating to obtain parameter correction values corresponding to the fully-connected neural network model by combining all wind power plant data, all photovoltaic power plant data, all first total wind power and all first total photovoltaic power in the sample data set; the fully-connected neural network model comprises an input layer, an output layer and a plurality of hidden layers, and the model parameter values comprise network weights and bias vectors corresponding to the hidden layers.
And S22, carrying out iterative optimization processing on the model parameter values of the fully-connected neural network model by using the parameter correction values, updating the model parameter values of the fully-connected neural network model according to the current parameter correction values during each iterative optimization processing, and acquiring a joint prediction loss value corresponding to the current fully-connected neural network model until the optimization times reach preset times or the current joint prediction loss value meets a preset range, and taking the current fully-connected neural network model as the wind-solar power joint prediction model.
Preferably, the process of acquiring the joint prediction loss value in step S22 includes steps S221 to S223, and each step specifically includes the following steps:
step S221, preprocessing all the wind power plant data and all the photovoltaic power station data in the sample data set to obtain corresponding data to be input.
In this embodiment, the pretreatment process includes steps (1) to (6), which are specifically as follows:
step (1), all wind power plant data and all photovoltaic power plant data in the sample data set are utilized to form a corresponding first variable matrix X = { X = (the number of the data in the wind power plant is one) of1,x2,…,xnH, n =288, where xnAre column vectors, each comprising (i + j) elements.
Step (2), subtracting the mean value corresponding to each column vector from each column vector in the first variable matrix X to realize decentralized processing so as to obtain the corresponding second variable matrix
Step (3) of calculating a covariance matrix
Step (4), solving the covariance matrix by adopting a characteristic value decomposition method
A plurality of corresponding eigenvalues and eigenvectors corresponding to the eigenvalues.
And (5) arranging all the eigenvalues in descending order, selecting the first k eigenvalues according to the arrangement result, and taking the eigenvectors corresponding to the first k eigenvalues as row vectors to form an eigenvector matrix P.
And (6) converting the projection of the input variable matrix into a new space constructed by k eigenvectors to obtain corresponding data X 'to be input, namely X' = PX, and realizing the dimension reduction optimization processing of the first variable matrix X.
Step S222, inputting data to be input into the current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area.
Step S223, please refer to equation (1), calculating to obtain a wind power predicted loss value L according to all the first total wind power and all the second total wind powerfAnd referring to the formula (2), calculating to obtain a photovoltaic power prediction loss value L according to all the first total photovoltaic power and all the second total photovoltaic powerg。
Wherein, T is the total time point number, here is 96; y istThe method comprises the steps of setting the actual total wind power of a region at the time t, namely the first total wind power at the time t in a preset region; f. oftThe predicted value of the total wind power of the region at the moment t of the model is a second total wind power at the moment t in the preset region; y isgThe actual total photovoltaic power of the area at the time t, namely the first total photovoltaic power at the time t in the preset area; f. ofgAnd (4) predicting the total photovoltaic power of the area at the moment t of the model, namely presetting the second total photovoltaic power at the moment t in the area.
Step S224, please refer to equation (3), predicting the loss value L of the wind power respectivelyfAnd photovoltaic power prediction loss value LgCarrying out weighting processing, and weighting results corresponding to the wind power prediction loss value and the photovoltaic powerAdding the weighted results corresponding to the rate prediction loss values to obtain a joint prediction loss value Ltotal。
Ltotal＝WfLf+WgLg
(3)
Wherein, Wf、WgThe weight coefficient of the wind power prediction task1 and the weight coefficient of the photovoltaic power prediction task2 are respectively.
In this embodiment, the fully-connected neural network model is essentially a multi-layer perceptron including a plurality of hidden layers, nodes in adjacent layers are fully connected, and there is no connection between nodes in the same layer, and the fully-connected neural network model mainly includes an input layer 01, a first hidden layer 02, a second hidden layer 03, a third hidden layer 04, and an output layer 05. Wherein, the input layer mainly has k × n neurons, which respectively correspond to the one-dimensional input vector expanded by the data to be input obtained after the dimension reduction optimization in the step S221; the first hidden layer 02, the second hidden layer 03 and the third hidden layer 04 are respectively provided with 800 neurons, 600 neurons and 300 neurons; the output layer has 96 x 2 neurons, the wind power total wind power prediction task1 and the photovoltaic power prediction task2 respectively correspond to 96 total wind power at different moments of the fully-connected neural network model and 96 total photovoltaic power at different moments of the fully-connected neural network model.
Please refer to equations (4) and (5), which show the relationship between the network weights and the bias vectors corresponding to the hidden layers of the fully-connected neural network model:
αl＝Wlhl-1+bl
(4)
hl＝f(αl)
(5)
wherein l is the layer number, αlIntermediate calculation variables, W, for the l-th hidden layerlNetwork weight, h, corresponding to the l-th hidden layerl-1Node activation values for the (l-1) th hidden layer, blIs hidden by the first layerOffset vector, h, corresponding to the reservoirlAnd f (-) is an activation parameter for the node activation value corresponding to the l-th hidden layer. Additionally, each hidden layer mainly employs lreol (leakage corrected Linear Unit) activation function, i.e. f (x) = max (0.01x, x), and for the present embodiment, x = α = maxl。
In this embodiment, an iteration end condition and associated initial parameters are set in advance. For example, the iteration end condition is that the optimization number reaches a preset number, such as 10000, or that the joint prediction loss value satisfies a preset range, such as less than 1 × 10-5. Additionally, the initial parameters are: step length eta =0.0001, moment estimation exponential decay rate rho1＝0.9，ρ2=0.998, denominator perturbation e =1 × 10-9The initial weight θ (including the weight and the bias vector corresponding to the hidden layer) is initialized randomly, the initial value r of the first moment variable is =0, s =0, and the initial value t of the iteration number is =0. Then, referring to equations (6) (7) (8) (9) (10) (11) (12) (13), adam optimization algorithm is performed until a preset iteration end condition is satisfied (i.e. t ≧ 10000 or the joint prediction loss value is less than 1 × 10)-5) And exiting iteration and taking the current fully-connected neural network model as the wind-solar power joint prediction model.
r＝ρ1r+(1-ρ1)gt
(7)
θ＝θ+Δθ
(12)
t＝t+1
(13)
Wherein, { x'(1),x'(2),…,x'(m)Y is a set of a plurality of sample data in the data to be input acquired in step S221(1),Y(2),…,Y(m)The power is a set of the first total photovoltaic power corresponding to each sample data, gtIs a gradient value, r is a first order moment estimated value, s is a second order moment estimated value,
as a result of the first order moment deviation correction,
as a result of the second-order moment deviation correction,
is the correction coefficient of the first moment estimated value of the t-th round,
and the coefficient is the correction coefficient of the second moment estimated value of the t-th round, delta theta is a parameter correction value, theta is a model parameter value, and t is the iterative optimization times.
Referring to fig. 2, a schematic structural diagram of a system for constructing a wind-solar power joint prediction model according to an embodiment of the present invention includes a data obtaining module 1 and a model constructing module 2, where each module is as follows:
the data acquisition module 1 is used for respectively observing a plurality of wind power plants and a plurality of photovoltaic power plants in a preset area in real time to obtain a sample data set; the method comprises the steps that a sample data set comprises a plurality of wind power plant data corresponding to each wind power plant, a plurality of photovoltaic power plant data corresponding to each photovoltaic power plant, first total wind power of a preset area at different moments and first total photovoltaic power of the preset area at different moments;
the model construction module 2 is used for performing iterative optimization on model parameter values of the pre-built fully-connected neural network model by using the sample data set according to an Adam algorithm until the optimization times reach preset times or a combined prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and then taking the current fully-connected neural network model as a wind-solar power combined prediction model; and the combined prediction loss value is calculated according to a preset algorithm and by combining the wind power prediction loss value and the photovoltaic power prediction loss value.
As a preferred scheme, a system for constructing a wind-solar power joint prediction model further includes:
the loss analysis module 3 is used for preprocessing all wind power plant data and all photovoltaic power station data in the sample data set to obtain corresponding data to be input; inputting data to be input into a current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to a preset area; according to a preset algorithm, calculating to obtain a wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain a photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power; and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain a combined prediction loss value.
As a preferred scheme, the loss analysis module 3 specifically includes a preprocessing unit 31 and a loss analysis unit 32, and each unit specifically includes the following:
the preprocessing unit 31 is configured to utilize all wind farm data and all photovoltaic power station data in the sample data set to form a corresponding first variable matrix; performing decentralized processing on the first variable matrix to obtain a corresponding second variable matrix, and calculating to obtain a corresponding covariance matrix according to the second variable matrix; calculating a plurality of eigenvalues corresponding to the covariance matrix and eigenvectors corresponding to the eigenvalues by adopting an eigenvalue decomposition method; arranging all the eigenvalues in a descending order, selecting the first k eigenvalues according to the arrangement result, and performing projection transformation on the first variable matrix by using eigenvectors corresponding to the first k eigenvalues to obtain corresponding data to be input, thereby realizing the dimension reduction optimization processing of the first variable matrix;
the loss analysis unit 32 is configured to input data to be input into the current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area; according to a preset algorithm, calculating to obtain a wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain a photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power; and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain a combined prediction loss value.
As a preferred scheme, the model building module 2 specifically includes a data processing unit 21 and a model building optimization unit 22, and each unit specifically includes the following:
the data processing unit 21 is configured to calculate, according to an Adam algorithm, to obtain a parameter correction value corresponding to the fully-connected neural network model by combining all the wind farm data, all the photovoltaic power station data, all the first total wind power, and all the first total photovoltaic power in the sample data set;
the model construction optimization unit 22 is used for performing iterative optimization processing on model parameter values of the fully-connected neural network model by using the parameter correction values, updating the model parameter values of the fully-connected neural network model according to the current parameter correction values during each iterative optimization processing, and acquiring a combined prediction loss value corresponding to the current fully-connected neural network model until the optimization times reach preset times or the current combined prediction loss value meets a preset range, and then taking the current fully-connected neural network model as a wind-solar power combined prediction model; the fully-connected neural network model comprises an input layer, an output layer and a plurality of hidden layers, and the model parameter values comprise network weights and bias vectors corresponding to the hidden layers.
It can be clearly understood by those skilled in the art that, for convenience and simplicity of description, the specific working process of the system described above may refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.
Compared with the prior art, the embodiment of the invention has the following beneficial effects:
the invention provides a method and a system for constructing a wind-solar power joint prediction model, wherein in the process of iterative optimization of model parameter values of a pre-built fully-connected neural network model, a wind power prediction loss value and a photovoltaic power prediction loss value are combined, and a joint prediction loss value corresponding to the current fully-connected neural network model is obtained through calculation and serves as an approximate target value of iterative optimization, so that the model can learn the interactive coupling relation between wind power and photovoltaic power, and the prediction precision of the model can be optimized for multiple times.
Considering that the detection accuracy of the wind power plant data and the detection accuracy of the photovoltaic power station data are different, and the wind power plant data and the photovoltaic power station data have different influence degrees on wind and light power combined prediction, the wind power prediction loss value and the photovoltaic power prediction loss value are weighted respectively, so that different weight coefficients are given to different prediction loss values, the contribution degree of key data to model parameter values is increased, and the prediction accuracy of the wind and light power combined prediction model is further improved.
Referring to fig. 3, a schematic flow diagram of a wind-solar power joint prediction method provided in an embodiment of the present invention is shown, where the method includes steps S3 to S4, and each step is specifically as follows:
s3, acquiring a plurality of wind power plant data corresponding to a plurality of wind power plants in an area to be predicted and a plurality of photovoltaic power plant data corresponding to a plurality of photovoltaic power plants, and preprocessing all the wind power plant data and all the photovoltaic power plant data to obtain corresponding matrixes to be input; the data of all the wind power plants are the wind speed, irradiance and temperature of each wind power plant at different moments, and the data of all the photovoltaic power stations are the wind speed, irradiance and temperature of each photovoltaic power station at different moments.
In this embodiment, since the wind and light power is mainly related to meteorological factors, the main influence factor of the wind power is wind speed, the main influence factor of the photovoltaic power is irradiance, and the temperature affects the wind power and the photovoltaic power together, the wind speed, the irradiance and the temperature of each wind farm at different moments are used as wind farm data, the wind speed, the irradiance and the temperature of each photovoltaic power station at different moments are used as photovoltaic power station data, and the wind and light power joint prediction in the area to be predicted can be predicted better.
As a preferred scheme, the pre-processing flow of all wind farm data and all photovoltaic power station data specifically includes steps S31 to S32, and each step specifically includes the following steps:
and S31, forming a corresponding variable matrix by using all wind power plant data and all photovoltaic power station data in the sample data set.
And S32, performing dimension reduction optimization processing on the variable matrix by adopting a principal component analysis method to obtain a corresponding matrix to be input.
It should be noted that, for the specific execution process of step S3, reference may be made to the specific working process of the aforementioned method for constructing a wind-solar power joint prediction model, and details are not described here again.
S4, inputting the matrix to be input into the wind-solar power combined prediction model, outputting a power prediction vector corresponding to the region to be predicted, and acquiring the total wind power and the total photovoltaic power of the region to be predicted at different moments according to the power prediction vector to realize wind-solar power combined prediction of the region to be predicted; the wind-solar power joint prediction model is obtained by the construction method of the wind-solar power joint prediction model.
The above-mentioned embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, and it should be understood that the above-mentioned embodiments are only examples of the present invention and are not intended to limit the scope of the present invention. It should be understood that any modifications, equivalents, improvements and the like, which come within the spirit and principle of the invention, may occur to those skilled in the art and are intended to be included within the scope of the invention.
Claims (10)
1. A method for constructing a wind-solar power joint prediction model is characterized by comprising the following steps:
respectively observing a plurality of wind power plants and a plurality of photovoltaic power stations in a preset area in real time to obtain a sample data set; the sample data set comprises a plurality of wind power plant data corresponding to each wind power plant, a plurality of photovoltaic power plant data corresponding to each photovoltaic power plant, first total wind power of the preset area at different moments and first total photovoltaic power of the preset area at different moments;
according to an Adam algorithm, performing iterative optimization on model parameter values of a pre-built fully-connected neural network model by using the sample data set until the optimization times reach preset times or a combined prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and taking the current fully-connected neural network model as a wind-solar power combined prediction model; and the combined prediction loss value is calculated according to a preset algorithm and by combining the wind power prediction loss value and the photovoltaic power prediction loss value.
2. The method for constructing the wind-solar power joint prediction model according to claim 1, wherein the obtaining of the joint prediction loss value specifically comprises:
preprocessing all the wind power plant data and all the photovoltaic power station data in the sample data set to obtain corresponding data to be input;
inputting the data to be input into the current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area;
according to a preset algorithm, calculating to obtain the wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain the photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power;
and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain the combined prediction loss value.
3. The method for constructing the wind-solar-power joint prediction model according to claim 2, wherein the preprocessing is performed on all the wind farm data and all the photovoltaic power plant data in the sample data set to obtain corresponding data to be input, specifically:
forming a corresponding first variable matrix by using all the wind power plant data and all the photovoltaic power station data in the sample data set;
performing decentralized processing on the first variable matrix to obtain a corresponding second variable matrix, and calculating to obtain a corresponding covariance matrix according to the second variable matrix;
calculating a plurality of eigenvalues corresponding to the covariance matrix and eigenvectors corresponding to the eigenvalues by adopting an eigenvalue decomposition method;
and arranging all the eigenvalues in a descending order, selecting the first k eigenvalues according to an arrangement result, and performing projection transformation on the first variable matrix by using the eigenvectors corresponding to the first k eigenvalues to obtain the corresponding data to be input, thereby realizing the dimension reduction optimization processing of the first variable matrix.
4. The method for constructing the wind-solar power joint prediction model according to claim 1, wherein according to an Adam algorithm, model parameter values of a pre-constructed fully-connected neural network model are iteratively optimized by using the sample data set until the optimization times reach preset times or a joint prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and then the current fully-connected neural network model is used as the wind-solar power joint prediction model, specifically:
according to an Adam algorithm, all the wind power plant data, all the photovoltaic power station data, all the first total wind power and all the first total photovoltaic power in the sample data set are combined, and a parameter correction value corresponding to the full-connection neural network model is obtained through calculation;
performing iterative optimization processing on the model parameter values of the fully-connected neural network model by using the parameter correction values, updating the model parameter values of the fully-connected neural network model according to the current parameter correction values during each iterative optimization processing, and acquiring the joint prediction loss value corresponding to the current fully-connected neural network model until the optimization times reach preset times or the current joint prediction loss value meets a preset range, and taking the current fully-connected neural network model as the wind-light power joint prediction model;
the fully-connected neural network model comprises an input layer, an output layer and a plurality of hidden layers, and the model parameter values comprise network weights and bias vectors corresponding to the hidden layers.
5. A construction system of a wind-solar power joint prediction model is characterized by comprising the following steps:
the data acquisition module is used for respectively observing a plurality of wind power plants and a plurality of photovoltaic power plants in a preset area in real time to obtain a sample data set; the sample data set comprises a plurality of wind power plant data corresponding to each wind power plant, a plurality of photovoltaic power plant data corresponding to each photovoltaic power plant, first total wind power of the preset area at different moments and first total photovoltaic power of the preset area at different moments;
the model construction module is used for carrying out iterative optimization on model parameter values of a pre-built fully-connected neural network model by utilizing the sample data set according to an Adam algorithm until the optimization times reach preset times or a combined prediction loss value corresponding to the current fully-connected neural network model meets a preset range, and then taking the current fully-connected neural network model as a wind-light power combined prediction model; and the combined prediction loss value is calculated according to a preset algorithm and by combining the wind power prediction loss value and the photovoltaic power prediction loss value.
6. The system for constructing the wind-solar-power joint prediction model of claim 5, further comprising:
the loss analysis module is used for preprocessing all the wind power plant data and all the photovoltaic power plant data in the sample data set to obtain corresponding data to be input; inputting the data to be input into the current fully-connected neural network model to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area; according to a preset algorithm, calculating to obtain the wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain the photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power; and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain the combined prediction loss value.
7. The system for constructing the wind-solar-power joint prediction model according to claim 6, wherein the loss analysis module specifically comprises:
the preprocessing unit is used for forming a corresponding first variable matrix by utilizing all the wind power plant data and all the photovoltaic power plant data in the sample data set; performing decentralized processing on the first variable matrix to obtain a corresponding second variable matrix, and calculating to obtain a corresponding covariance matrix according to the second variable matrix; calculating a plurality of eigenvalues corresponding to the covariance matrix and eigenvectors corresponding to the eigenvalues by adopting an eigenvalue decomposition method; arranging all the eigenvalues in a descending order, selecting the first k eigenvalues according to an arrangement result, and performing projection transformation on the first variable matrix by using the eigenvectors corresponding to the first k eigenvalues to obtain the corresponding data to be input, thereby realizing the dimension reduction optimization processing of the first variable matrix;
the loss analysis unit is used for inputting the data to be input into the current fully-connected neural network model so as to obtain a plurality of second total wind power and a plurality of second total photovoltaic power corresponding to the preset area; according to a preset algorithm, calculating to obtain the wind power prediction loss value according to all the first total wind power and all the second total wind power, and calculating to obtain the photovoltaic power prediction loss value according to all the first total photovoltaic power and all the second total photovoltaic power; and respectively carrying out weighting processing on the wind power prediction loss value and the photovoltaic power prediction loss value, and adding a weighting result corresponding to the wind power prediction loss value and a weighting result corresponding to the photovoltaic power prediction loss value to obtain the combined prediction loss value.
8. The system for constructing the wind-solar-power joint prediction model according to claim 5, wherein the model construction module specifically comprises:
the data processing unit is used for calculating to obtain a parameter correction value corresponding to the fully-connected neural network model by combining all the wind power plant data, all the photovoltaic power station data, all the first total wind power and all the first total photovoltaic power in the sample data set according to an Adam algorithm;
the model construction optimization unit is used for performing iterative optimization processing on model parameter values of the fully-connected neural network model by using the parameter correction values, updating the model parameter values of the fully-connected neural network model according to the current parameter correction values during each iterative optimization processing, acquiring the joint prediction loss value corresponding to the current fully-connected neural network model, and taking the current fully-connected neural network model as the wind and light power joint prediction model until the optimization times reach preset times or the current joint prediction loss value meets a preset range; the fully-connected neural network model comprises an input layer, an output layer and a plurality of hidden layers, and the model parameter values comprise network weights and bias vectors corresponding to the hidden layers.
9. A wind-solar power joint prediction method is characterized by comprising the following steps:
acquiring a plurality of wind power plant data corresponding to a plurality of wind power plants and a plurality of photovoltaic power plant data corresponding to a plurality of photovoltaic power plants in an area to be predicted, and preprocessing all the wind power plant data and all the photovoltaic power plant data to obtain corresponding matrixes to be input;
inputting the matrix to be input into a wind-solar power joint prediction model, outputting a power prediction vector corresponding to the region to be predicted, and acquiring the total wind power and the total photovoltaic power of the region to be predicted at different moments according to the power prediction vector to realize wind-solar power joint prediction of the region to be predicted;
the wind power plant data are wind speed, irradiance and temperature of each wind power plant at different moments, the photovoltaic power plant data are wind speed, irradiance and temperature of each photovoltaic power plant at different moments, and the wind and light power combined prediction model is constructed by using the construction method of the wind and light power combined prediction model according to any one of claims 1 to 5.
10. The wind-solar power joint prediction method according to claim 9, wherein the preprocessing is performed on all the wind farm data and all the photovoltaic power plant data to obtain corresponding matrices to be input, specifically:
forming a corresponding variable matrix by using all the wind power plant data and all the photovoltaic power station data in the sample data set;
and performing dimension reduction optimization processing on the variable matrix by adopting a principal component analysis method to obtain the corresponding matrix to be input.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210851575.9A CN115271191A (en) | 2022-07-18 | 2022-07-18 | Method for constructing wind-light power combined prediction model System and prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210851575.9A CN115271191A (en) | 2022-07-18 | 2022-07-18 | Method for constructing wind-light power combined prediction model System and prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115271191A true CN115271191A (en) | 2022-11-01 |
Family
ID=83767477
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210851575.9A Pending CN115271191A (en) | 2022-07-18 | 2022-07-18 | Method for constructing wind-light power combined prediction model System and prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115271191A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115860247A (en) * | 2022-12-19 | 2023-03-28 | 广西电网有限责任公司 | Method and device for training fan loss power prediction model in extreme weather |
| CN116742626A (en) * | 2023-08-14 | 2023-09-12 | 长江三峡集团实业发展(北京)有限公司 | Wind-solar combined power set prediction method and device considering atmospheric chaos characteristics |
| CN117992751A (en) * | 2024-02-26 | 2024-05-07 | 湖北交投新能源投资有限公司 | Method, device, equipment and medium for predicting power generation data of distributed new energy |
| CN115860247B (en) * | 2022-12-19 | 2024-10-25 | 广西电网有限责任公司 | Training method and device for fan loss power prediction model in extreme weather |
-
2022
- 2022-07-18 CN CN202210851575.9A patent/CN115271191A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115860247A (en) * | 2022-12-19 | 2023-03-28 | 广西电网有限责任公司 | Method and device for training fan loss power prediction model in extreme weather |
| CN115860247B (en) * | 2022-12-19 | 2024-10-25 | 广西电网有限责任公司 | Training method and device for fan loss power prediction model in extreme weather |
| CN116742626A (en) * | 2023-08-14 | 2023-09-12 | 长江三峡集团实业发展(北京)有限公司 | Wind-solar combined power set prediction method and device considering atmospheric chaos characteristics |
| CN116742626B (en) * | 2023-08-14 | 2023-10-20 | 长江三峡集团实业发展(北京)有限公司 | Wind-solar combined power set prediction method and device considering atmospheric chaos characteristics |
| CN117992751A (en) * | 2024-02-26 | 2024-05-07 | 湖北交投新能源投资有限公司 | Method, device, equipment and medium for predicting power generation data of distributed new energy |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059878B (en) | Photovoltaic power generation power prediction model based on CNN LSTM and construction method thereof | |
| CN109117951B (en) | BP neural network-based probability load flow online calculation method | |
| Wang et al. | Research and application of a combined model based on multi-objective optimization for multi-step ahead wind speed forecasting | |
| CN111310889B (en) | Evaporation waveguide profile estimation method based on deep neural network | |
| CN115271191A (en) | Method for constructing wind-light power combined prediction model System and prediction method | |
| CN108205717A (en) | A kind of photovoltaic generation power Multiple Time Scales Forecasting Methodology | |
| JP5888640B2 (en) | Photovoltaic power generation prediction apparatus, solar power generation prediction method, and solar power generation prediction program | |
| CN106503867A (en) | A kind of genetic algorithm least square wind power forecasting method | |
| CN110942194A (en) | Wind power prediction error interval evaluation method based on TCN | |
| CN108090621B (en) | Short-term wind speed prediction method and system based on staged overall optimization | |
| CN110942205B (en) | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM | |
| CN114970952B (en) | Photovoltaic output short-term prediction method and system considering environmental factors | |
| CN112561058A (en) | Short-term photovoltaic power prediction method based on Stacking-ensemble learning | |
| CN112381282B (en) | Photovoltaic power generation power prediction method based on width learning system | |
| Chitsazan et al. | Wind speed forecasting using an echo state network with nonlinear output functions | |
| CN112215478B (en) | Power coordination control method and device for optical storage station and storage medium | |
| CN116187540B (en) | Wind power station ultra-short-term power prediction method based on space-time deviation correction | |
| CN115275991A (en) | Active power distribution network operation situation prediction method based on IEMD-TA-LSTM model | |
| CN106446440A (en) | Short-term photovoltaic generation power prediction method based on online sequential extreme learning machine | |
| CN113468817A (en) | Ultra-short-term wind power prediction method based on IGOA (optimized El-electric field model) | |
| CN112149883A (en) | Photovoltaic power prediction method based on FWA-BP neural network | |
| CN114330100A (en) | Short-term photovoltaic power probability interval prediction method | |
| CN114362175A (en) | Wind power prediction method and system based on depth certainty strategy gradient algorithm | |
| CN114298140A (en) | Wind power short-term power prediction correction method considering unit classification | |
| CN111488974B (en) | Ocean wind energy downscaling method based on deep learning neural network |
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 |