CN110110915A - A kind of integrated prediction technique of the load based on CNN-SVR model - Google Patents
A kind of integrated prediction technique of the load based on CNN-SVR model Download PDFInfo
- Publication number
- CN110110915A CN110110915A CN201910350144.2A CN201910350144A CN110110915A CN 110110915 A CN110110915 A CN 110110915A CN 201910350144 A CN201910350144 A CN 201910350144A CN 110110915 A CN110110915 A CN 110110915A
- Authority
- CN
- China
- Prior art keywords
- load
- data
- cnn
- svr
- 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 34
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000010606 normalization Methods 0.000 claims abstract description 9
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 239000011159 matrix material Substances 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 description 13
- 230000006870 function Effects 0.000 description 8
- 230000005611 electricity Effects 0.000 description 7
- 238000003066 decision tree Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000013277 forecasting method Methods 0.000 description 3
- 230000007787 long-term memory Effects 0.000 description 3
- 230000001537 neural effect Effects 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 2
- 240000002853 Nelumbo nucifera Species 0.000 description 2
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of loads based on CNN-SVR model to integrate prediction technique, this method first initializes the daily load data of all users in the influence factor of somewhere influence load, somewhere, calculate Pearson correlation coefficient, then data are subjected to extreme value normalization and cluster, obtain different classes of user data label, according to step packet label, user data is merged, as training input data;Secondly model training being carried out after building CNN-SVR load forecasting model and input data being pre-processed, data are input in the CNN-SVR model after training and are predicted, obtain prediction result, by result renormalization, obtain final multiple groups prediction load;Finally multiple groups load is summed, somewhere is obtained and finally predicts load daily, if prediction somewhere month load, load is monthly summed.Compared with prior art, the present invention a large number of users in somewhere can be carried out Ensemble classifier, have many advantages, such as independently to extract data characteristics, precision of prediction it is high.
Description
Technical field
The present invention relates to Load Prediction In Power Systems fields, more particularly, to a kind of load collection based on CNN-SVR model
At prediction technique.
Background technique
Traditional Methods of electric load forecasting have regression analysis, wavelet analysis method, decision tree, random forest, support to
Amount machine, neural network scheduling algorithm, wherein support vector regression is often used in clock synchronization as the mutation of algorithm of support vector machine
Between sequence data prediction, be a kind of algorithms most in use with preferable quasi- precision of prediction, but since its precision of prediction is joined by itself
It is several and input feature vector data to be affected, therefore the step of input data is frequently necessary to by feature extraction.Neural network is calculated
Method is popular algorithm in recent years, has the autonomous ability for extracting nonlinear characteristic well, can be very good to carry out non-linear
Fitting.But there are many hyper parameter of neural network algorithm, model training process complexity is cumbersome.
With the rapid development of China's economic level, when the river rises the boat goes up for demand of the China to electricity consumption, and electricity consumption increases continuously and healthily
It is long.In this case, reinforce demand side management, analyze the electricity consumption behavior of user, the accuracy of load prediction is improved, to enhancing
The economy of electric system, stabilization, intelligent operation have realistic meaning.The behind of electricity consumption sustainable growth is that the electricity consumption of flood tide is used
Family faces a large amount of electricity consumption user when carrying out regional load prediction, and most methods are only to divide user's cluster
Class, but how load prediction is carried out to it after not analyzing user's classification;Or it proposes only for some user or a certain whole
Load forecasting method, and the load prediction situation of the user of a large amount of different electricity consumption behaviors is not analyzed.Therefore, being badly in need of one kind being capable of needle
The method that regional load is predicted in the case where there is a large number of users to a certain area.
Summary of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide one kind can be to somewhere
A large number of users carry out Ensemble classifier, can independently extract data characteristics and with preferable precision of prediction based on CNN-SVR mould
The load of type integrates prediction technique.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of integrated prediction technique of the load based on CNN-SVR model, comprising the following steps:
Step 1: on every daily load of all users in the influence factor of somewhere influence power system load, somewhere
Data are initialized.
Step 2: calculate each user data the moon load, moon load mean value, median, standard deviation and influence factor with
The Pearson correlation coefficient of load data, calculation formula are as follows:
In formula, γmIndicate the Pearson correlation coefficient of different influence factors, aiIndicate load number of certain user at i-th day
According to,Indicate the m influence factors such as temperature, weather, date type.
Step 3: the data in step 1 and step 2 are carried out extreme value normalized according to the following formula:
In formula, x ' is the data after normalization, and x is input data, xminIt is input data minimum value, xmaxIt is input data
Maximum value.
Step 4: the data after normalization are clustered, and obtain different classes of user data label.
Step 5: merged the data of user in step 1 according to the packet label in step 4, it is defeated as training
Enter data.
Step 6: building CNN-SVR load forecasting model (convolutional neural networks-support vector regression load prediction mould
Type), which is made of two layers of convolutional layer and two layers of full articulamentum.
Step 7: according to the packet data of step 4, station work CNN-SVR load forecasting model, specific training process
Are as follows:
1) dummy variable is constructed to the training data in step 5.
2) dummy variable data are subjected to matrixing processing.
Matrixing is that the one-dimensional vector for the m 1 × n that will be originally inputted is converted into mMatrix form.If
It is not integer, then to former vector data zero padding, until n can be by radication of integer.
3) data of step 2) are passed sequentially through into two layers of convolutional layer and two layers of full articulamentum carries out convolution algorithm.
4) CNN (Convolutional Neural Networks, convolutional neural networks) model is trained, training
Number is not less than ten times, and after training, the data of the first full articulamentum are input to SVR (Support Vector
Machine, support vector regression) it is trained in model.
Preferably, CNN model training number is 10 times, and learning rate is set as selecting RBF kernel function in 0.01, SVR model
(Radial Basis Function, radial basis function) is as full articulamentum and ReLU layers of primary function, hyper parameter punishment system
Number C takes 50, γ to take 0.01.
Step 8: building prediction input data, i.e., will need the weather, temperature and date type of forecast date to normalize,
Construct dummy variable, row matrix of going forward side by side.
It is predicted Step 9: the data in step 8 are input in trained CNN-SVR model, obtains prediction knot
Fruit simultaneously saves.
Step 10: the result in step 9 is carried out renormalization, final multiple groups prediction load is obtained.Renormalization is public
Formula are as follows:
yre=ypre(rmax-rmin)+rmin
In formula: yreIt is that load, y are predicted obtained by renormalizationpreIt is model prediction as a result, rminFor the minimum in input data
Value, rmaxFor the maximum value of input data.
Step 11: multiple groups load is summed, obtains somewhere and finally predict load daily.If predicting somewhere month load,
Then load is monthly summed.
Compared with prior art, the invention has the following advantages that
1) the method for the present invention creates convolutional neural networks-support vector regression load characteristics clustering integrated predictive model, right
Cnn-svr model is reused after data progress clustering ensemble to be predicted, can be grouped integrated number of users effectively to user grouping
According to rear, user data greatly reduces, therefore model quickly can carry out load prediction to the area for having a large number of users, quickly high
Effect;
2) the CNN-SVR model that the present invention creates combines the advantages of CNN and SVR model, when convolutional neural networks are trained
Between it is long, difficulty is big, support vector regression requires initial data and hyper parameter high, and the present invention uses shallow-layer convolutional neural networks
It is fast that requirement and shallow Model training speed of the support vector regression to data are met as data characteristics extractor, are used
The support vector regression of RBF kernel function is to load prediction, and training is simple, predetermined speed is fast, improves the overall load of model
Precision of prediction;
3) decision-tree model based on the clustering ensemble load forecasting method of CNN-SVR model relative to conventional method, branch
Holding vector machine model, convolutional neural networks model, shot and long term memory models has higher load prediction precision, high reliablity.
Detailed description of the invention
Fig. 1 is the flow diagram of the method for the present invention;
Fig. 2 is the training flow chart of CNN-SVR model in the method for the present invention.
Specific embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.Obviously, described embodiment is this
A part of the embodiment of invention, rather than whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art exist
Every other embodiment obtained under the premise of creative work is not made, all should belong to the scope of protection of the invention.
As shown in Figure 1, the present invention relates to a kind of loads based on CNN-SVR model to integrate prediction technique, specifically include down
Column step:
S1. the daily of all users in the influence factors such as daily height temperature, weather, date type and somewhere is initialized to bear
Lotus data, if the number of users in somewhere is K.
S2. calculate each user data the moon load, moon load mean value, median, standard deviation and influence factor and load
The Pearson correlation coefficient of data.The formula of Pearson correlation coefficient is as follows:
In formula, γmIndicate the Pearson correlation coefficient of different influence factors.aiIndicate load number of certain user at i-th day
According to,Indicate the m influence factors such as temperature, weather, date type.
S3. the data in step S1 and S2 are subjected to extreme value normalized, normalization formula is as follows:
In formula, x ' is the data after normalization, and x is input data, xminIt is input data minimum value, xmaxIt is input data
Maximum value.
S4. the data after normalization are clustered, setting clustering method is the hierarchical clustering method using ward connection, is gathered
Class group range is 22~26, and the selected cluster group of the present embodiment is 24, and end of clustering obtains 24 groups of different classes of users
Data label.
S5. according to the packet label in S4, the data integration of script K group user in S1 is merged into 24 groups of data as instruction
Practice input data.
S6. convolutional neural networks-support vector regression load forecasting model is constructed, model is complete by level 2 volume lamination and 2 layers
Articulamentum composition.
S7. dummy variable is constructed to the training data in S5.
S8. the data in S7 are subjected to matrixing processing.Using the load data of the one day of some user as a training
Sample, data are the one-dimensional vector form of 1 × n.Matrixing is exactly that the one-dimensional vector for the m 1 × n that will be originally inputted is converted into m
It is aMatrix form.IfIt is not integer, then to former vector data zero padding, until n can be by radication of integer.
S9. station work CNN-SVR load forecasting model, training process are shown in Fig. 2.Wherein CNN model training number is 10
Secondary, learning rate is set as selecting RBF kernel function (Radial Basis Function, radial basis function) in 0.01, SVR model, surpasses
Parameter penalty coefficient C takes 50, γ to take 0.01.
S10. building prediction input data, i.e., will need the weather, temperature and date type of forecast date to normalize, and construct
Dummy variable, row matrix of going forward side by side.
S11. load prediction is carried out, the data in S10 are input to train and are predicted in 24 groups of CNN-SVR models,
It obtains 24 groups of prediction results and saves.
S12. the result in S11 is subjected to renormalization, obtains 24 groups of final prediction loads.Renormalization formula is such as
Under:
yre=ypre(rmax-rmin)+rmin
In formula: yreIt is that load, y are predicted obtained by renormalizationpreIt is model prediction as a result, rminFor the minimum in input data
Value, rmaxFor the maximum value of input data.
S13. 24 groups of loads are summed, obtains somewhere and finally predicts load daily.If predicting somewhere month load, will bear
Lotus is monthly summed.
For the validity and reliability for proving CNN-SVR model in the method for the present invention, the present embodiment takes real data pair
Decision-tree model, supporting vector machine model (Support Vector Machine, SVM), convolutional neural networks model
(Convolutional Neural Networks, CNN), shot and long term memory models (Long Short Term Memory,
LSTM) prediction with CNN-SVR model in the method for the present invention is compared, and test result is as shown in table 1.
Table 1 uses the test result of distinct methods
By table 1 it is found that relative to the decision-tree model of conventional method, supporting vector machine model, convolutional neural networks mould
Type, shot and long term memory models, clustering ensemble load forecasting method predetermined speed based on CNN-SVR model is fast, rapidly and efficiently, tool
There are higher load prediction precision, high reliablity.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
The staff for being familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (7)
1. a kind of load based on CNN-SVR model integrates prediction technique, which is characterized in that this method includes the following steps:
1) the daily load data of all users in the influence factor of somewhere influence power system load, somewhere is carried out just
Beginningization;
2) calculate each user data the moon load, moon load mean value, median, standard deviation and influence factor and load data
Pearson correlation coefficient;
3) data in step 1) and step 2) are subjected to extreme value normalized, and the data after normalization are clustered,
Obtain different classes of user data label;
4) according to the packet label in step 3), the data of user in step 1) are merged, as training input data;
5) CNN-SVR load forecasting model is constructed, and according to the packet data in step 3), station work CNN-SVR load is pre-
Survey model;
6) prediction input data is built, dummy variable, row matrix of going forward side by side are constructed;
7) data in step 6) are input in the CNN-SVR model after training and are predicted, obtain prediction result and saved;
8) result in step 7), is subjected to renormalization, obtains final multiple groups prediction load;
9) multiple groups load is summed, obtains somewhere and finally predicts load daily, if prediction somewhere month load, monthly by load
Summation.
2. a kind of load based on CNN-SVR model according to claim 1 integrates prediction technique, which is characterized in that step
It is rapid 2) in, each user data the moon load, moon load mean value, median, standard deviation and influence factor and load data skin
The expression formula of your inferior related coefficient are as follows:
In formula, γmFor the Pearson correlation coefficient of different influence factors, aiLoad data for certain user at i-th day,For
The m influence factor such as temperature, weather, date type.
3. a kind of load based on CNN-SVR model according to claim 1 integrates prediction technique, which is characterized in that step
It is rapid 3) in, the expression formula of extreme value normalized are as follows:
In formula, x ' is the data after normalization, and x is input data, xminFor input data minimum value, xmaxFor input data maximum
Value.
4. a kind of load based on CNN-SVR model according to claim 3 integrates prediction technique, which is characterized in that step
It is rapid 3) in, the data after normalization are clustered using the hierarchical clustering method of ward connection.
5. a kind of load based on CNN-SVR model according to claim 1 integrates prediction technique, which is characterized in that institute
The CNN-SVR load forecasting model stated is made of two layers of convolutional layer and two layers of full articulamentum.
6. a kind of load based on CNN-SVR model according to claim 5 integrates prediction technique, which is characterized in that
The training process of CNN-SVR load forecasting model specifically includes the following steps:
A1 dummy variable) is constructed to the training data in step 4);
A2 dummy variable data) are subjected to matrixing processing;
A3) by the data of step a2) pass sequentially through CNN-SVR load forecasting model two layers of convolutional layer and two layers of full articulamentum into
Row convolution algorithm;
A4) CNN model is trained, frequency of training is not less than ten times, after training, the data of the first full articulamentum are defeated
Enter into SVR model and is trained.
7. a kind of load based on CNN-SVR model according to claim 1 integrates prediction technique, which is characterized in that step
It is rapid 8) in, the calculating formula of renormalization are as follows:
yre=ypre(rmax-rmin)+rmin
In formula: yreTo predict load, y obtained by renormalizationpreFor model prediction as a result, rm i nFor the minimum value in input data,
rmaxFor the maximum value of input data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910350144.2A CN110110915A (en) | 2019-04-28 | 2019-04-28 | A kind of integrated prediction technique of the load based on CNN-SVR model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910350144.2A CN110110915A (en) | 2019-04-28 | 2019-04-28 | A kind of integrated prediction technique of the load based on CNN-SVR model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110110915A true CN110110915A (en) | 2019-08-09 |
Family
ID=67487267
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910350144.2A Pending CN110110915A (en) | 2019-04-28 | 2019-04-28 | A kind of integrated prediction technique of the load based on CNN-SVR model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110110915A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110618610A (en) * | 2019-08-29 | 2019-12-27 | 杭州华电江东热电有限公司 | Gas turbine NOx standard exceeding early warning method and device based on LSTM deep learning |
CN110956306A (en) * | 2019-10-23 | 2020-04-03 | 广东电网有限责任公司 | Load prediction method based on load clustering |
CN111461462A (en) * | 2020-04-29 | 2020-07-28 | 南京工程学院 | Daily load prediction method based on TrellisNet-L STM |
CN111581581A (en) * | 2020-04-23 | 2020-08-25 | 大唐环境产业集团股份有限公司 | Method and system for detecting NOx concentration at SCR inlet under multi-boundary condition |
CN113067383A (en) * | 2021-03-31 | 2021-07-02 | 广东博力威科技股份有限公司 | Method, apparatus, electronic device and medium for optimizing user battery charging habits |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130110756A1 (en) * | 2011-10-31 | 2013-05-02 | Siemens Corporation | Short-term Load Forecast Using Support Vector Regression and Feature Learning |
CN104778503A (en) * | 2015-03-16 | 2015-07-15 | 国家电网公司 | Method for forecasting public building air conditioner short-time base wire load with consideration of real-time weather factors |
CN106952181A (en) * | 2017-03-08 | 2017-07-14 | 深圳市景程信息科技有限公司 | Electric Load Prediction System based on long Memory Neural Networks in short-term |
CN108846530A (en) * | 2018-09-28 | 2018-11-20 | 国网上海市电力公司 | One kind being based on the short-term load forecasting method of " cluster-recurrence " model |
CN109063911A (en) * | 2018-08-03 | 2018-12-21 | 天津相和电气科技有限公司 | A kind of Load aggregation body regrouping prediction method based on gating cycle unit networks |
-
2019
- 2019-04-28 CN CN201910350144.2A patent/CN110110915A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130110756A1 (en) * | 2011-10-31 | 2013-05-02 | Siemens Corporation | Short-term Load Forecast Using Support Vector Regression and Feature Learning |
CN104778503A (en) * | 2015-03-16 | 2015-07-15 | 国家电网公司 | Method for forecasting public building air conditioner short-time base wire load with consideration of real-time weather factors |
CN106952181A (en) * | 2017-03-08 | 2017-07-14 | 深圳市景程信息科技有限公司 | Electric Load Prediction System based on long Memory Neural Networks in short-term |
CN109063911A (en) * | 2018-08-03 | 2018-12-21 | 天津相和电气科技有限公司 | A kind of Load aggregation body regrouping prediction method based on gating cycle unit networks |
CN108846530A (en) * | 2018-09-28 | 2018-11-20 | 国网上海市电力公司 | One kind being based on the short-term load forecasting method of " cluster-recurrence " model |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110618610A (en) * | 2019-08-29 | 2019-12-27 | 杭州华电江东热电有限公司 | Gas turbine NOx standard exceeding early warning method and device based on LSTM deep learning |
CN110956306A (en) * | 2019-10-23 | 2020-04-03 | 广东电网有限责任公司 | Load prediction method based on load clustering |
CN111581581A (en) * | 2020-04-23 | 2020-08-25 | 大唐环境产业集团股份有限公司 | Method and system for detecting NOx concentration at SCR inlet under multi-boundary condition |
CN111461462A (en) * | 2020-04-29 | 2020-07-28 | 南京工程学院 | Daily load prediction method based on TrellisNet-L STM |
CN111461462B (en) * | 2020-04-29 | 2023-11-21 | 南京工程学院 | Daily load prediction method based on trellis Net-LSTM |
CN113067383A (en) * | 2021-03-31 | 2021-07-02 | 广东博力威科技股份有限公司 | Method, apparatus, electronic device and medium for optimizing user battery charging habits |
CN113067383B (en) * | 2021-03-31 | 2024-01-02 | 广东博力威科技股份有限公司 | Method, device, electronic equipment and medium for optimizing charging habit of user battery |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110110915A (en) | A kind of integrated prediction technique of the load based on CNN-SVR model | |
Yu et al. | LSTM-EFG for wind power forecasting based on sequential correlation features | |
CN111929748B (en) | Meteorological element forecasting method and system | |
CN106909933B (en) | A kind of stealing classification Forecasting Methodology of three stages various visual angles Fusion Features | |
CN110674407B (en) | Hybrid recommendation method based on graph convolution neural network | |
CN109325547A (en) | Non-motor vehicle image multi-tag classification method, system, equipment and storage medium | |
CN110245709A (en) | Based on deep learning and from the 3D point cloud data semantic dividing method of attention | |
CN111696101A (en) | Light-weight solanaceae disease identification method based on SE-Inception | |
CN104573879A (en) | Photovoltaic power station output predicting method based on optimal similar day set | |
CN108090510A (en) | A kind of integrated learning approach and device based on interval optimization | |
CN105512676A (en) | Food recognition method at intelligent terminal | |
CN111008640A (en) | Image recognition model training and image recognition method, device, terminal and medium | |
CN105893609A (en) | Mobile APP recommendation method based on weighted mixing | |
CN109165819B (en) | Active power distribution network reliability rapid evaluation method based on improved AdaBoost. M1-SVM | |
CN111898703B (en) | Multi-label video classification method, model training method, device and medium | |
CN108345908A (en) | Sorting technique, sorting device and the storage medium of electric network data | |
KR102593835B1 (en) | Face recognition technology based on heuristic Gaussian cloud transformation | |
CN105046323B (en) | Regularization-based RBF network multi-label classification method | |
CN105701572A (en) | Photovoltaic short-term output prediction method based on improved Gaussian process regression | |
CN113344288B (en) | Cascade hydropower station group water level prediction method and device and computer readable storage medium | |
Mori et al. | Hybrid intelligent method of relevant vector machine and regression tree for probabilistic load forecasting | |
CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
CN107506792A (en) | A kind of semi-supervised notable method for checking object | |
CN102163285A (en) | Cross-domain video semantic concept detection method based on active learning | |
CN112926645A (en) | Electricity stealing detection method based on edge calculation |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190809 |