CN112560898B - Load space-time prediction method based on deep learning - Google Patents
Load space-time prediction method based on deep learning Download PDFInfo
- Publication number
- CN112560898B CN112560898B CN202011338730.4A CN202011338730A CN112560898B CN 112560898 B CN112560898 B CN 112560898B CN 202011338730 A CN202011338730 A CN 202011338730A CN 112560898 B CN112560898 B CN 112560898B
- Authority
- CN
- China
- Prior art keywords
- load
- dimensional
- space
- correlation
- deep learning
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000013135 deep learning Methods 0.000 title claims abstract description 31
- 238000012216 screening Methods 0.000 claims abstract description 8
- 238000005457 optimization Methods 0.000 claims abstract description 4
- 230000009466 transformation Effects 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 5
- CIWBSHSKHKDKBQ-JLAZNSOCSA-N Ascorbic acid Chemical compound OC[C@H](O)[C@H]1OC(=O)C(O)=C1O CIWBSHSKHKDKBQ-JLAZNSOCSA-N 0.000 claims description 4
- 238000005314 correlation function Methods 0.000 claims description 4
- 238000011176 pooling Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 2
- 230000001419 dependent effect Effects 0.000 claims 1
- 238000011426 transformation method Methods 0.000 claims 1
- 238000012549 training Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000005611 electricity Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 239000002699 waste material Substances 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06N3/08—Learning methods
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Marketing (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Development Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a load space-time prediction method based on deep learning, and belongs to the field of power load prediction. The method performs load space-time feature screening, and constructs gridding cells according to the space coordinates of the load to serve as input features of a prediction model; on the basis, a load space-time prediction model based on deep learning, which is suitable for load space-time prediction, is established, and the load space-time prediction model based on the deep learning is obtained after optimization by training cells of space load distribution and taking a minimized prediction error as a training target, so that the load space-time prediction is realized, and the load space-time prediction model has strong flexibility and adaptability.
Description
Technical Field
The invention relates to a load space-time prediction method based on deep learning, and belongs to the field of power load prediction.
Background
With the rapid development of the power grid technology in each country, the electricity consumption is continuously increased, the requirements of each electricity consumption unit on the grid structure and the electric energy quality are also continuously increased, and the space load prediction method is highly valued by the electric power department. The traditional load prediction method only predicts the load, does not give fine load space distribution, and cannot meet the power system planning requirement of lean conversion. Therefore, the space load prediction result is an important basis for the power distribution network partition planning and the power grid construction, and is a key link for safe and reliable operation of the power grid.
When the predicted load value is far greater than the actual value of the load, the power department builds power plants and power networks on a large scale, and the power network income is possibly smaller than the construction investment, so that the resource and fund waste of the power department is caused; in contrast, when the predicted load value is far smaller than the actual load value, the power department reduces the construction force to the power grid due to misleading of inaccurate load data, so that the power equipment is supplied insufficiently, the power system power supply capacity is far smaller than the electric quantity required by society, and the requirements of power users cannot be met. Therefore, accurate and efficient spatial load prediction is of paramount importance. The existing space load prediction methods at present are more than dozens of methods, and the methods comprise a land simulation method based on a rough set theory, a land simulation method based on load components and SVM technology, a regression analysis method, a Markov method and the like. Compared with the traditional load prediction, the prediction results obtained by the method have obvious space-time distribution characteristics and have important significance for power system planning and load prediction work, but the method often only utilizes shallow information existing in data, but cannot establish a complex and profound prediction model, so that a load space-time prediction model based on deep learning is necessary to be established.
Disclosure of Invention
Aiming at the limitations of the related background technology, the invention provides a load space-time prediction method based on deep learning, which combines a maximum correlation minimum redundancy algorithm to obtain the correlation factor which is most suitable for being used as model output, constructs and establishes a multi-layer neural network model, and realizes the space-time distribution prediction of the load.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
a load space-time prediction method based on deep learning comprises the following steps:
1) Screening to obtain load related factors, and removing redundant related factors so as to improve prediction accuracy;
2) Dividing a load into cells according to longitude and latitude coordinates and grids, and establishing a plurality of two-dimensional cell matrixes according to different load types to form three-dimensional cells;
3) Taking the three-dimensional cells obtained in the step 2) and the load related factors obtained in the screening in the step 1) as model inputs, taking a cell result to be predicted as output, and establishing a load space-time prediction deep learning model based on deep learning;
4) And predicting the power load by using the established load space-time prediction model.
In the above technical solution, further, the step 1) specifically includes:
the maximum correlation minimum redundancy algorithm is adopted, the correlation between the load correlation factors and the load data is maximized, the correlation between the load correlation factors is minimized, and the expression is as follows:
maxD(S,x 2 )
minR(S)
wherein S is a feature subset, x 1i 、x 1j Features that are feature subset S, i.e., load-related factors; x is x 2 For load data, I () is a correlation function; in order to consider both maximizing the correlation of the load-related factors with the load data and minimizing the correlation between the load-related factors, an operator φ (D, R) is defined, i.e., the above formula is converted into:
maxφ(D,R)
φ(D,R)=D-R
according to the algorithm, the optimal feature subset of the load related factors is obtained and is used as the input of a load space-time prediction model based on deep learning.
The correlation function between the data is specifically:
wherein I is correlation, A and B are calculated correlation data, wherein A is a load correlation factor, and B is a load correlation factor or load data; pr () is probability.
Further, in step 2), the load is divided into cells according to the longitude and latitude coordinates, and a plurality of two-dimensional cell matrixes are established according to different load types to form three-dimensional cells, specifically: and (3) dividing the area according to longitude and latitude meshing, establishing a two-dimensional matrix for the load data of each land type, wherein the numerical value of each element in the matrix is the load data of the corresponding land type in the area corresponding to the position of the element, and if the load data of the land type is not present in the area, the element is 0. For different land types, a plurality of two-dimensional matrixes with the same size are established, the two-dimensional matrixes are used as two-dimensional cells, and the two-dimensional cells are spliced to form a three-dimensional cell C. Three-dimensional cell C wherein the first two dimensions are space coordinates and the third dimension is of the type of land, C ··r Load data representing class r land, C pq· Representing the various loads at coordinates (p, q).
The method is adopted to process load data at different moments to obtain three-dimensional cells, a plurality of three-dimensional cells are arranged in time sequence to obtain a time sequence of the three-dimensional cells, and stability verification is carried out on the time sequence of the three-dimensional cells. If the stability is not satisfied, the stability of the newly generated time sequence is ensured by adopting technologies such as difference, logarithmic transformation and the like.
Further, the step 3) specifically includes: and (3) inputting the historical data of the load related factors into a full-connection layer for nonlinear transformation, inputting the historical three-dimensional cells into a convolution layer for convolution calculation to form a hidden layer of a load space-time prediction model based on deep learning, inputting the hidden layer into the full-connection layer again, and predicting to obtain a space load prediction result. And carrying out dimension reduction summation on a third dimension of the three-dimensional cell at the moment to be predicted as a true value of the prediction result at the moment to be predicted, calculating a loss function of the space load prediction result, and carrying out iterative optimization of network parameters with the aim of reducing the loss function.
Solving a load space-time prediction model based on deep learning, iteratively updating network parameters by using an Adam optimization algorithm, and solving the load space-time prediction model based on deep learning.
Further, the fully connected layer is subjected to nonlinear transformation, which is represented by linearly transforming input data and activating the input data through an activation function, so that no linear relation exists between an output result and original data:
z 1 =ωξ 1 +b
a 1 =σ(z 1 )
in xi 1 For input data, ω is input weight, b is input bias, σ () is activation function, z 1 As intermediate variable, a 1 To output the result.
Further, the three-dimensional cells are input into a convolution layer to carry out convolution calculation, specifically: performing nonlinear variation on the convolution layer, wherein the process comprises convolution transformation and pooling transformation; for convolution transforms, the output results are expressed as:
a 2 =σ(z 2 )
in xi 2 [ω 1 ,ω 2 ,ω 3 ]Is a three-dimensional cell in (omega) 1 ,ω 2 ,ω 3 ) The element value at the coordinates, F is the convolution layer window, F is the window size, nc is the three-dimensional cell channel number,in coordinates for intermediate variables->Element value at a 2 To output the result. For the pooled transformation, the output results are expressed as:
wherein a is 3 To output the result, pool () is a pooling function, taking the average or maximum value of the input matrix as the output.
The invention has the beneficial effects that:
the invention provides a load space-time prediction model based on deep learning, which selects load related factors and space-time distribution data of three-dimensional cells as the input of the model, and takes a minimized prediction error as a training target to obtain an optimized load space-time prediction model based on deep learning, thereby realizing load space-time prediction so as to guide power grid planning and having strong flexibility and adaptability; extracting input features suitable for a prediction model by using a maximum correlation minimum redundancy algorithm; and a convolutional neural network is adopted to learn the space-time distribution rule of the load.
Drawings
FIG. 1 is an inventive deep learning based load spatiotemporal prediction flow;
FIG. 2 is a deep learning predictive model framework.
Detailed Description
The invention is further described below with reference to the drawings and examples of implementation.
The flow of the load space-time prediction method based on deep learning provided by the invention is shown in figure 1.
A load space-time prediction method based on deep learning comprises the following steps:
(1) Firstly, screening load related factors and removing redundancy related factors, specifically: the maximum correlation minimum redundancy algorithm is adopted, the correlation between the load correlation factors and the load data is maximized, and the correlation between the load correlation factors is minimized, so that the purposes of screening the correlation factor information and removing the redundancy correlation factors are realized:
maxD(S,x 2 )
minR(S)
wherein S is a feature subset, x 1i 、x 1j Features that are feature subset S, i.e., load-related factors; x is x 2 For load data, I () is a correlation function; in order to consider both maximizing the correlation of the load-related factors with the load data and minimizing the correlation between the load-related factors, phi (D, R) is defined:
maxφ(D,R)
φ(D,R)=D-R
according to the algorithm, an optimal feature subset of the load related factors is obtained and used as input of a load space-time prediction model based on deep learning;
the correlation between the data is calculated, specifically:
wherein I is a correlation; a and B are calculation correlation data, wherein A is a load correlation factor, and B is a load correlation factor or load data; p () is a probability.
(2) Dividing the load into cells according to the longitude and latitude coordinates and grids, and establishing a plurality of two-dimensional cell matrixes according to different load types to form three-dimensional cells, wherein the method specifically comprises the following steps of: dividing an area according to longitude and latitude meshing, establishing a two-dimensional matrix for the load data of each land type, wherein the numerical value of each element in the matrix is the moment load data of the corresponding land type in the area corresponding to the element position, and if the load data of the land type is not present in the area, the element is 0; for different land types, a plurality of two-dimensional matrixes with the same size are established, the two-dimensional matrixes are taken as two-dimensional cells, and the two-dimensional cells are spliced to form a three-dimensional cell C, wherein the first two dimensions are space coordinates, and the third dimension is the land type, namely C ··r Load data representing class r land, C pq· Representing the various loads at coordinates (p, q). The method is adopted to process load data at different moments to obtain three-dimensional cells, a plurality of three-dimensional cells are arranged in time sequence to obtain a time sequence of the three-dimensional cells, and stability verification is carried out on the time sequence of the three-dimensional cells. Typical data were selected and tested for time series stationarity of three-dimensional cells using an Augmented Dickey-Fuller (ADF). And carrying out differential processing on the condition that the stability is not met, and carrying out ADF (automatic frequency correction) inspection on the data after the differential processing again until the data stability is met.
(3) Dividing samples into training and testing sets, the samples including t= { (T) 1i ,t 2 ,y)} N Wherein t is 1i Is three-dimensional cell after difference, t 2 The load related factors obtained by screening comprise an air temperature value, week information and the like, wherein the air temperature value, the week information and the like are explanatory variables, y is a target variable, the target variable is a differential two-dimensional cell to be predicted, namely a space load, and N is the number of samples.
(4) The load space-time prediction model based on deep learning as shown in fig. 2 is established, three-dimensional cells are input into a convolution network, load related factors are input into a fully-connected network, the outputs of the two are spliced and input into the fully-connected network, and a prediction result is output. And iteratively updating network parameters by taking the root mean square of the prediction result error as a loss function until the model converges.
(5) And obtaining the predicted value of the load to be predicted by adding the predicted result of the differential data and the historical value at the last moment.
(6) And evaluating the prediction effect based on the test set. The error evaluation indexes of the prediction results include average absolute ratio error (mean absolute percentage error, MAPE), root mean square error (root mean square error, RMSE), average absolute error (mean absolute error, MAE):
Claims (6)
1. The load space-time prediction method based on deep learning is characterized by comprising the following steps of:
1) Screening to obtain load related factors, and removing redundant related factors to improve prediction accuracy;
2) Dividing a load into cells according to longitude and latitude coordinates and grids, and establishing a plurality of two-dimensional cell matrixes according to different load types to form three-dimensional cells;
3) Taking three-dimensional cells and load related factors obtained by screening in the step 1) as model input, taking cell results to be predicted as output, and establishing a load space-time prediction model based on deep learning;
4) Predicting the power load by using the established load space-time prediction model;
in the step 2), the load is divided into cells according to longitude and latitude coordinates and grids, and a plurality of two-dimensional cell matrixes are established according to different load types to form three-dimensional cells, specifically:
dividing an area according to longitude and latitude meshing, establishing a two-dimensional matrix for the load data of each land type, wherein the numerical value of each element in the matrix is the moment load data of the corresponding land type in the area corresponding to the element position, and if the load data of the land type is not present in the area, the element is 0; for different land types, a plurality of two-dimensional matrixes with the same size are established, the two-dimensional matrixes are used as two-dimensional cells, the two-dimensional cells are spliced to form a three-dimensional cell C, wherein the first two dimensions are space coordinates, and the third dimension is the landType, i.e. C ··r Load data representing class r land, C pq· Representing various types of loads at coordinates (p, q);
processing load data at different moments by adopting the method to obtain three-dimensional cells, arranging a plurality of three-dimensional cells in time sequence to obtain a time sequence of the three-dimensional cells, and performing stability verification on the time sequence of the three-dimensional cells;
the step 3) is specifically as follows: inputting the historical data of the load related factors into a full-connection layer for nonlinear transformation, inputting the historical three-dimensional cells into a convolution layer for convolution calculation to form a hidden layer of a load space-time prediction model based on deep learning, inputting the hidden layer into the full-connection layer again, and predicting to obtain a space load prediction result; and carrying out dimension reduction summation on a third dimension of the three-dimensional cells at the moment to be predicted, taking the third dimension as a true value of a predicted result at the moment to be predicted, calculating a loss function of a space load predicted result, and carrying out iterative optimization of network parameters with the aim of reducing the loss function.
2. The load space-time prediction method based on deep learning according to claim 1, wherein the step 1) specifically comprises: the maximum correlation minimum redundancy algorithm is adopted, the correlation between the load correlation factors and the load data is maximized, the correlation between the load correlation factors is minimized, and the expression is as follows:
maxD(S,x 2 )
minR(S)
wherein S is a feature subset, x 1i 、x 1j Features that are feature subset S, i.e., load-related factors; x is x 2 For load data, I () is a correlation function; to simultaneously consider the maximization of the load-related factors andthe correlation of the load data and the correlation between the minimized load-dependent factors define the operator phi (D, R):
maxφ(D,R)
φ(D,R)=D-R
according to the algorithm, the optimal feature subset of the load related factors is obtained and is used as the input of a load space-time prediction model based on deep learning.
3. The deep learning-based load spatiotemporal prediction method of claim 2, wherein the correlation between the calculated data is specifically:
wherein I is a correlation; a and B are calculation correlation data, wherein A is a load correlation factor, and B is a load correlation factor or load data; pr () is probability.
4. The deep learning-based load space-time prediction method according to claim 1, wherein the time series of the three-dimensional cells is subjected to stationarity verification, and if the stationarity is not satisfied, a differential or logarithmic transformation method is adopted to ensure the stationarity of the time series of the three-dimensional cells.
5. The deep learning based load spatiotemporal prediction method of claim 1, wherein the full-connected layer is subjected to nonlinear transformation, which is represented by linear transformation of input data and activation by an activation function, so that there is no linear relationship between the output result and the original data:
z 1 =ωξ 1 +b
a 1 =σ(z 1 )
in xi 1 For input data, ω is input weight, b is input bias, σ () is activation function, z 1 As intermediate variable, a 1 To output the result.
6. The deep learning-based load space-time prediction method according to claim 1, wherein the three-dimensional cells are input into a convolution layer to perform convolution calculation, specifically: performing nonlinear variation on the convolution layer, wherein the process comprises convolution transformation and pooling transformation; for convolution transforms, the output results are expressed as:
a 2 =σ(z 2 )
in xi 2 [ω 1 ,ω 2 ,ω 3 ]Is a three-dimensional cell in (omega) 1 ,ω 2 ,ω 3 ) The element value at the coordinates, F is the convolution layer window, F is the window size, nc is the three-dimensional cell channel number,in coordinates for intermediate variables->Element value at a 2 To output a result; for the pooled transformation, the output results are expressed as:
wherein a is 3 To output the result, pool () is a pooling function, taking the average or maximum value of the input matrix as the output.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011338730.4A CN112560898B (en) | 2020-11-25 | 2020-11-25 | Load space-time prediction method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011338730.4A CN112560898B (en) | 2020-11-25 | 2020-11-25 | Load space-time prediction method based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112560898A CN112560898A (en) | 2021-03-26 |
CN112560898B true CN112560898B (en) | 2023-04-28 |
Family
ID=75043597
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011338730.4A Active CN112560898B (en) | 2020-11-25 | 2020-11-25 | Load space-time prediction method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112560898B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107169878A (en) * | 2017-04-12 | 2017-09-15 | 华中科技大学 | A kind of method based on information independence collection space load basic data of increasing income |
CN108182484A (en) * | 2017-12-04 | 2018-06-19 | 国网吉林省电力有限公司长春供电公司 | Spatial Load Forecasting method based on gridding technology and BP neural network |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9312698B2 (en) * | 2012-12-19 | 2016-04-12 | Robert Bosch Gmbh | System and method for energy distribution |
US9960637B2 (en) * | 2015-07-04 | 2018-05-01 | Sunverge Energy, Inc. | Renewable energy integrated storage and generation systems, apparatus, and methods with cloud distributed energy management services |
US10186889B2 (en) * | 2015-10-08 | 2019-01-22 | Taurus Des, Llc | Electrical energy storage system with variable state-of-charge frequency response optimization |
CN108960488B (en) * | 2018-06-13 | 2022-03-04 | 国网山东省电力公司经济技术研究院 | Saturated load spatial distribution accurate prediction method based on deep learning and multi-source information fusion |
CN109345013A (en) * | 2018-09-25 | 2019-02-15 | 广西电网有限责任公司电力科学研究院 | A kind of distribution user power utilization behavior prediction method |
CN109615119B (en) * | 2018-11-23 | 2022-09-20 | 东北电力大学 | Space load prediction method based on rank set pair analysis theory |
CN111445078A (en) * | 2020-03-31 | 2020-07-24 | 国网河北省电力有限公司 | Comprehensive energy system multi-element load prediction method based on long-term and short-term memory neural network |
-
2020
- 2020-11-25 CN CN202011338730.4A patent/CN112560898B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107169878A (en) * | 2017-04-12 | 2017-09-15 | 华中科技大学 | A kind of method based on information independence collection space load basic data of increasing income |
CN108182484A (en) * | 2017-12-04 | 2018-06-19 | 国网吉林省电力有限公司长春供电公司 | Spatial Load Forecasting method based on gridding technology and BP neural network |
Also Published As
Publication number | Publication date |
---|---|
CN112560898A (en) | 2021-03-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110070226B (en) | Photovoltaic power prediction method and system based on convolutional neural network and meta-learning | |
Wu et al. | A short‐term load forecasting method based on GRU‐CNN hybrid neural network model | |
CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
CN112348292B (en) | Short-term wind power prediction method and system based on deep learning network | |
CN110070228B (en) | BP neural network wind speed prediction method for neuron branch evolution | |
CN110738344A (en) | Distributed reactive power optimization method and device for load prediction of power system | |
CN114693064B (en) | Building group scheme generation performance evaluation method | |
CN110705727A (en) | Photovoltaic power station shadow shielding diagnosis method and system based on random forest algorithm | |
CN113468817A (en) | Ultra-short-term wind power prediction method based on IGOA (optimized El-electric field model) | |
CN116307291B (en) | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition | |
CN116011608A (en) | Photovoltaic power ultra-short-term prediction method and device, electronic equipment and storage medium | |
CN110009125A (en) | Short-term power load prediction model establishment method based on EMD-VMD-PSO-BPNN | |
CN111898316A (en) | Construction method and application of super-surface structure design model | |
CN114169251A (en) | Ultra-short-term wind power prediction method | |
CN114970952A (en) | Photovoltaic output short-term prediction method and system considering environmental factors | |
CN112560898B (en) | Load space-time prediction method based on deep learning | |
Liao et al. | A novel stacked generalization ensemble-based hybrid PSVM-PMLP-MLR model for energy consumption prediction of copper foil electrolytic preparation | |
CN112836876A (en) | Power distribution network line load prediction method based on deep learning | |
CN117154690A (en) | Photovoltaic power generation power prediction method and system based on neural network | |
CN117132132A (en) | Photovoltaic power generation power prediction method based on meteorological data | |
CN115577856A (en) | Method and system for predicting construction cost and controlling balance of power transformation project | |
CN113722951B (en) | Scatterer three-dimensional finite element grid optimization method based on neural network | |
CN114638421A (en) | Method for predicting requirement of generator set spare parts | |
CN114418194B (en) | Tower damage prediction method and device based on data driving and model driving | |
CN113675844B (en) | Multi-node power load prediction method |
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 |