CN109934392A - A kind of micro-capacitance sensor short-term load forecasting method based on deep learning - Google Patents
A kind of micro-capacitance sensor short-term load forecasting method based on deep learning Download PDFInfo
- Publication number
- CN109934392A CN109934392A CN201910153480.8A CN201910153480A CN109934392A CN 109934392 A CN109934392 A CN 109934392A CN 201910153480 A CN201910153480 A CN 201910153480A CN 109934392 A CN109934392 A CN 109934392A
- Authority
- CN
- China
- Prior art keywords
- data
- network
- lstm
- load
- cnn
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 30
- 238000013277 forecasting method Methods 0.000 title claims abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000010248 power generation Methods 0.000 claims abstract description 10
- 238000013527 convolutional neural network Methods 0.000 claims description 42
- 238000012549 training Methods 0.000 claims description 42
- 230000000694 effects Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000001413 cellular effect Effects 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000008859 change Effects 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims description 2
- 238000013499 data model Methods 0.000 claims description 2
- 238000011161 development Methods 0.000 claims description 2
- 238000011430 maximum method Methods 0.000 claims description 2
- 210000002569 neuron Anatomy 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims description 2
- 239000004744 fabric Substances 0.000 claims 1
- 230000005540 biological transmission Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of micro-capacitance sensor short-term load forecasting method based on deep learning, based on CNN and LSTM network, classification is accurately handled historical data by CNN deep learning method, LSTM network group is constructed by LSTM deep learning method, and be combined two kinds of deep learning methods eventually by selector, obtain CNN-LSTM model;Based on CNN-LSTM model, construct the load forecasting model based on deep learning, the model can not only carry out the prediction of higher precision to the power load in micro-grid load, the prediction of degree of precision, the final precision of prediction promoted to micro-capacitance sensor integral load can also be carried out to the distributed photovoltaic power generation load in micro-grid load.
Description
Technical field
The present invention relates to a kind of methods of micro-capacitance sensor short-term load forecasting, are base more particularly, to a kind of deep learning network
The load forecasting method of plinth.
Background technique
Micro-capacitance sensor is the important channel of power grid consumption distributed new, to the pre- of micro-capacitance sensor internal load and integral load
Survey has great importance.Currently, the algorithm for load prediction includes traditional algorithm and modern algorithm two major classes.
Since there are the higher distributed news of permeability inside micro-capacitance sensor, load variations and fluctuation are increasing,
It is bigger to predict that difficulty compares traditional big net load.Use tradition widely applied Load Forecast Algorithm such as BP neural network and support
The shallow-layers learning method such as vector machine is no longer satisfied the huge micro-grid load prediction of fluctuation.When input data amount is very big
When, for shallow-layer learning algorithm network due to inherently a kind of fully-connected network, the weight that will lead to whole network is too many, and performance is anxious
Play decline, finally will limit every layer of neuron number that can be accommodated up to, and then limits it and learn depth, can only rest on shallow-layer
Study.Shallow-layer study carries out Backward error propagation process with gradient transmission method and limits its study depth, when study depth increases
The error of added-time, reverse transfer can sharply die-off with the increase of depth, generate " gradient disappearance " phenomenon, the transmission effect of error
Fruit can only rest on shallow-layer, there is no any effect to the right value update of deep layer, and training effect levels off to nothing, even result in
The preceding layer parameter for closing on input layer can not optimize, and tend to be random.
Summary of the invention
The present invention mainly solves technical problem present in the prior art;It provides a kind of based on convolutional neural networks
(CNN) and the deep learning prediction technique of shot and long term memory network (LSTM) will more by the combination of CNN and LSTM network
Feature and relationship existing for the excavation load data of depth greatly improve the precision and reliability of load prediction.
Further object of the present invention is to solve technical problem present in the prior art;It provides a kind of based on CNN algorithm
Day genre classification methods load is carried out, in turn by Accurate classification by one day of deep learning network struction type sorter
It can be improved the precision of load prediction.
It is to solve technical problem present in the prior art that the present invention has a purpose again;It provides a kind of based on LSTM network
Load prediction network group obtain a series of network by being trained and data mining for different day type loads,
Network group is constituted, the precision of load prediction can be greatly improved.
Above-mentioned technical problem of the invention is mainly to be addressed by following technical proposals:
A kind of micro-capacitance sensor short-term load forecasting method based on deep learning, based on defined below:
Define 1, data block: data block is divided into two major parts: training block and test/prediction block;Training block includes training
Training data needed for network includes typical history power generation data and weather label (labels) for CNN network training
Data packet, for LSTM network, for the data for predicting required data and time point actual negative charge values comprising power generation data
Packet;Test/prediction block includes test/data to be predicted needed for being tested or carrying out load prediction, to based on CNN
Deep learning sorter network for, data packet format is consistent with format of the training block in CNN network;To with LSTM net
For deep learning load prediction network based on network group, the data of data packet format and training block in LSTM network
Packet format is consistent;
Define 2, CNN block: include in CNN block is the classification method based on the study of CNN network depth;It is defeated by data block
After entering training data information, it can train and generate CNN sorter network, inputting other data informations can be carried out category classification;
For distributed photovoltaic power generation loaded portion, category classification identification should be 4 classes: fine day, cloudy, cloudy, sleet;To electricity consumption
For loaded portion, category classification identification should be 2 classes: working day, festivals or holidays;
3 are defined, selection block: this block structure is simple, only one selector, for being only selective discrimination;Into
During row training network, the object that it is selected is the result of CNN network and the total data set that training data is constituted, and is acted on
In separating photovoltaic load according to 2 classes according to 4 classes or power load, then brings LSTM block into again and be trained;Carry out
When prediction, the object that it is selected is LSTM network group, and effect is that weather forecast information selects suitable LSTM net in LSTM groups
Network, and enter data into LSTM network and predicted;
Define 4, LSTM block: include in LSTM block is the prediction technique based on the study of LSTM network depth;It was training
Cheng Zhong, the different classes of data sent according to selection block, the different LSTM network of building training, these networks ultimately form
LSTM network group suitable for prediction;During prediction, according to selector, the LSTM network of suitable prediction data is recalled, into
Row load prediction, output load predicted value;
It specifically includes:
Step 1, historical data pre-processes: by historical load data and day information input model, it is established that raw data set,
It is carried out again using Matlab software except operations such as wrong and normalization, group, which is combined into, meets data arrangement required for data model inputs
Format obtains the data set for being able to carry out building load forecasting model;
Step 2, historical data classifier: historical load data collection is inputted into CNN network classifier, is successively taken in step 1
Historical data inputs CNN network, until fully entering until, to historical data progress deep learning, by the not number of type on the same day
It is separated and is reconfigured according to collection, obtain sorted history data set;
Step 3, selector chooses data: by the sorted data set input selector of step 2, by selector according to history
Data classification result is directed respectively into different LSTM networks, and acquisition can be by the history data set of different LSTM Network Recognitions;
Step 4, LSTM load prediction network group is constructed: by the data set after step 3 selection respectively according to the choosing of selector
It selects and inputs each LSTM network one by one, carry out deep learning respectively, LSTM load prediction network group is obtained after study;
Step 5, it carries out load prediction: will predict the selector of information needed input step 3, LSTM is selected by selector
Meet the load prediction network of load to be predicted in network group, and input information, obtains the result of load prediction.
In a kind of above-mentioned micro-capacitance sensor short-term load forecasting method based on deep learning, the step 2, history
The specific operation method is as follows for data sorter:
Step 2.1, input of each 10 days of the typical day CNN network under type on the same day is chosen not;
Step 2.2, initialization algorithm must parameter: the parameter of initialization is usually convolution kernel and bias, convolution kernel without
Special circumstances carry out random initializtion, and bias is initialized as 0 without special circumstances;General CNN network is learnt
When do not need that operation is normalized;
Step 2.3, convolutional calculation and pond: the result of convolution is not necessarily a value, is one in most cases
Matrix, this matrix are weight matrix, referred to as convolution kernel;Entire convolution step can be considered as to one to be weighted and ask
And the step of;Chi Hua, i.e. sub-sampling, main purpose is compressive features figure map and reduces dimension, uses maximum method pond;
Step 2.4, error back propagation and network training;Error carries out back-propagation process in each layer of network, be in order to
Change each layer weight and bias, and reduces final error by constantly changing them;
Step 2.5: historical data being fully entered into trained CNN network that is, in historical data classifier and is classified
Historical data afterwards.
In a kind of above-mentioned micro-capacitance sensor short-term load forecasting method based on deep learning, the step 4, building
The specific operation method is as follows for LSTM load prediction network group:
Step 4.1, by selector, separating for the sorted historical data independent isolating of CNN inputs different respectively
In LSTM network;
Step 4.2, data processing: mainly including data normalization, is packaged into cellular and is convenient for input, after being normalized,
Number of computations grade is reduced, and can accelerate the speed of service of whole network, therefore historical data is before entering LSTM network training,
Data cellular need to be obtained to historical data normalized;
Step 4.3, door state and network training the training of LSTM network: are calculated;Each input quantity is two parts,
It is the output of last moment hidden layer and the input of current time neural network respectively;Isolated data are passed through into nerve in LSTM
The state and error stream of first door are trained LSTM network, obtain LSTM network one by one, constitute LSTM network group.
In a kind of above-mentioned micro-capacitance sensor short-term load forecasting method based on deep learning, the step 3, selector
It carries out: when carrying out the building of LSTM network group, the classification results of historical data classifier being isolated, it is independent to input different LSTM
Network;
Also, when being predicted, according to the predictive information of input, suitable LSTM network is selected to be predicted.
In a kind of above-mentioned micro-capacitance sensor short-term load forecasting method based on deep learning, by radiating in the step 3
Electromagnetic wave high fdrequency component is 1GHz~3GHz superelevation frequency component.
Therefore, the present invention has the advantage that 1. use the classifier of CNN network struction, historical data can be excavated
Relevance accurately classifies to historical data;2., can be to not class on the same day using the prediction module of LSTM network group building
The load of type carries out Accurate Prediction;3. connection of the use selector to block, quickly can handle and select not type need on the same day
The network predicted.
Detailed description of the invention
Fig. 1 is the connection schematic diagram of modules of the present invention.
Fig. 2 a is the present invention with prediction result (the micro-capacitance sensor inside generation load to micro-capacitance sensor internal load under different situations
Fine day prediction result).
Fig. 2 b is the present invention with prediction result (the micro-capacitance sensor inside generation load to micro-capacitance sensor internal load under different situations
Cloudy prediction result).
Fig. 2 c is the present invention with prediction result (the micro-capacitance sensor inside generation load to micro-capacitance sensor internal load under different situations
Cloudy prediction result).
Fig. 2 d is the present invention with prediction result (the micro-capacitance sensor inside generation load to micro-capacitance sensor internal load under different situations
Sleet prediction result).
Fig. 2 e is the present invention with prediction result (the micro-capacitance sensor power inside load to micro-capacitance sensor internal load under different situations
Working day prediction result).
Fig. 2 f is the present invention with prediction result (the micro-capacitance sensor power inside load to micro-capacitance sensor internal load under different situations
Festivals or holidays prediction result).
Fig. 3 is micro-capacitance sensor integral load prediction result of the invention.
Specific embodiment
Below with reference to the embodiments and with reference to the accompanying drawing the technical solutions of the present invention will be further described.
Embodiment:
A kind of micro-capacitance sensor short-term load forecasting method based on deep learning, comprising the following steps:
Step 1, data block: data block is divided into two major parts: training block and test/prediction block.Training block includes training
Training data needed for network includes typical history power generation data and weather label (labels) for CNN network training
Data packet, for LSTM network, for the data for predicting required data and time point actual negative charge values comprising power generation data
Packet.Test/prediction block includes test/data to be predicted needed for being tested or carrying out load prediction, to based on CNN
Deep learning sorter network for, data packet format is consistent with format of the training block in CNN network;To with LSTM net
For deep learning load prediction network based on network group, the data packet of data packet format and training block in LSTM network
Format is consistent.
Step 2, CNN block: include in CNN block is the classification method based on the study of CNN network depth.It is defeated by data block
After entering training data information, it can train and generate CNN sorter network, inputting other data informations can be carried out category classification.
For distributed photovoltaic power generation loaded portion, category classification identification should be 4 classes: fine day, cloudy, cloudy, sleet.To electricity consumption
For loaded portion, category classification identification should be 2 classes: working day, festivals or holidays.
Step 3, selection block: this block structure is simple, only one selector, function is only selective discrimination.?
It is trained in network development process, the object that it is selected is the result of CNN network and the total data set that training data is constituted, effect
It is according to 4 classes or power load to separate photovoltaic load according to 2 classes, then brings LSTM block into again and be trained.Into
When row prediction, the object that it is selected is LSTM network group, and effect is that weather forecast information selects suitable LSTM in LSTM crowds
Network, and enter data into LSTM network and predicted.
Step 4, LSTM block: include in LSTM block is the prediction technique based on the study of LSTM network depth.It was training
Cheng Zhong, the different classes of data sent according to selection block, the different LSTM network of building training, these networks ultimately form
LSTM network group suitable for prediction.During prediction, according to selector, the LSTM network of suitable prediction data is recalled, into
Row load prediction, output load predicted value.
It in the present embodiment, include: including for counting using the module of detection method of partial discharge of electrical equipment above
According to the data block 2 of processing, carry out historical data classification CNN block 3, connection CNN block and LSMT block selector 4, predicted
LSTM block.
Specific embodiment described herein is only an example for the spirit of the invention.The neck of technology belonging to the present invention
The technical staff in domain can make various modifications or additions to the described embodiments or replace by a similar method
In generation, however, it does not deviate from the spirit of the invention or beyond the scope of the appended claims.
Although CNN, LSTM is used more herein, classifier, memory unit, input gate, out gate, forgets the arts such as door
Language, but it does not exclude the possibility of using other terms.The use of these items is only for be more convenient to describe and explain this
The essence of invention;Being construed as any additional limitation is disagreed with spirit of that invention.
Claims (5)
1. a kind of micro-capacitance sensor short-term load forecasting method based on deep learning, based on defined below:
Define 1, data block: data block is divided into two major parts: training block and test/prediction block;Training block includes training network
Required training data, the data comprising typical history power generation data and weather label (labels) for CNN network training
Packet, for LSTM network, for the data packet for predicting required data and time point actual negative charge values comprising power generation data;It surveys
Examination/prediction block includes test/data to be predicted needed for being tested or carrying out load prediction, to the depth based on CNN
It spends for learning classification network, data packet format is consistent with format of the training block in CNN network;It is to LSTM network group
For the deep learning load prediction network on basis, the data packet format one of data packet format and training block in LSTM network
It causes;
Define 2, CNN block: include in CNN block is the classification method based on the study of CNN network depth;It is inputted and is instructed by data block
After practicing data information, it can train and generate CNN sorter network, inputting other data informations can be carried out category classification;To point
For cloth photovoltaic power generation loaded portion, category classification identification should be 4 classes: fine day, cloudy, cloudy, sleet;To power load
For part, category classification identification should be 2 classes: working day, festivals or holidays;
3 are defined, selection block: this block structure is simple, only one selector, for being only selective discrimination;It is being instructed
Practice network development process in, it select object be CNN network result and training data constitute total data set, effect be by
Photovoltaic load is separated according to 4 classes or power load according to 2 classes, is then brought LSTM block into again and is trained;It is being predicted
When, the object that it is selected is LSTM network group, and effect is that weather forecast information selects suitable LSTM network in LSTM groups, and
LSTM network is entered data into be predicted;
Define 4, LSTM block: include in LSTM block is the prediction technique based on the study of LSTM network depth;In the training process,
The different classes of data sent according to selection block, the different LSTM network of building training, these networks, which ultimately form, to be suitable for
The LSTM network group of prediction;During prediction, according to selector, the LSTM network of suitable prediction data is recalled, carries out load
Prediction, output load predicted value;
It specifically includes:
Step 1, historical data pre-processes: by historical load data and day information input model, it is established that raw data set, then adopt
It is carried out removing the operations such as wrong and normalization with Matlab software, group, which is combined into, meets data arrangement lattice required for data model inputs
Formula obtains the data set for being able to carry out building load forecasting model;
Step 2, historical data classifier: historical load data collection is inputted into CNN network classifier, successively takes history in step 1
Data input CNN network, until fully entering until, to historical data progress deep learning, by the not data set of type on the same day
It is separated and is reconfigured, obtain sorted history data set;
Step 3, selector chooses data: by the sorted data set input selector of step 2, by selector according to historical data
Classification results are directed respectively into different LSTM networks, and acquisition can be by the history data set of different LSTM Network Recognitions;
Step 4, construct LSTM load prediction network group: by step 3 selection after data set respectively according to the selection of selector by
A each LSTM network of input, carries out deep learning respectively, LSTM load prediction network group is obtained after study;
Step 5, it carries out load prediction: will predict the selector of information needed input step 3, LSTM network is selected by selector
Meet the load prediction network of load to be predicted in group, and input information, obtains the result of load prediction.
2. a kind of micro-capacitance sensor short-term load forecasting method based on deep learning according to claim 1, which is characterized in that
In the step 2, the specific operation method is as follows for historical data classifier:
Step 2.1, input of each 10 days of the typical day CNN network under type on the same day is chosen not;
Step 2.2, initialization algorithm must parameter: the parameter of initialization is usually convolution kernel and bias, and convolution kernel is without special
Situation carries out random initializtion, and bias is initialized as 0 without special circumstances;When general CNN network is learnt not
Need to be normalized operation;
Step 2.3, convolutional calculation and pond: the result of convolution is not necessarily a value, is a matrix in most cases,
This matrix is weight matrix, referred to as convolution kernel;Entire convolution step can be considered as to one to be weighted and sum
Step;Chi Hua, i.e. sub-sampling, main purpose is compressive features figure map and reduces dimension, uses maximum method pond;
Step 2.4, error back propagation and network training;Error carries out back-propagation process in each layer of network, is to change
Each layer weight and bias, and reduce final error by constantly changing them;
Step 2.5: historical data being fully entered into trained CNN network that is, in historical data classifier and is obtained sorted
Historical data.
3. a kind of micro-capacitance sensor short-term load forecasting method based on deep learning according to claim 2, which is characterized in that
In the step 4, the specific operation method is as follows for building LSTM load prediction network group:
Step 4.1, by selector, by separating for the sorted historical data independent isolating of CNN, different LSTM is inputted respectively
In network;
Step 4.2, data processing: mainly including data normalization, is packaged into cellular convenient for input, after being normalized, calculates
The order of magnitude is reduced, and can accelerate the speed of service of whole network, therefore historical data is before entering LSTM network training, is needed pair
Historical data normalized obtains data cellular;
Step 4.3, door state and network training the training of LSTM network: are calculated;Each input quantity is two parts, respectively
It is the output of last moment hidden layer and the input of current time neural network;Isolated data are passed through into neuron door in LSTM
State and error stream LSTM network is trained, obtain LSTM network one by one, constitute LSTM network group.
4. a kind of micro-capacitance sensor short-term load forecasting method based on deep learning according to claim 1, which is characterized in that
In the step 3, selector is carried out: when carrying out the building of LSTM network group, the classification results of historical data classifier are isolated,
It is independent to input different LSTM networks;
Also, when being predicted, according to the predictive information of input, suitable LSTM network is selected to be predicted.
5. a kind of micro-capacitance sensor short-term load forecasting method based on deep learning according to claim 1, which is characterized in that
By in the step 3, radiated electromagnetic wave high fdrequency component is 1GHz~3GHz superelevation frequency component.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910153480.8A CN109934392A (en) | 2019-02-28 | 2019-02-28 | A kind of micro-capacitance sensor short-term load forecasting method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910153480.8A CN109934392A (en) | 2019-02-28 | 2019-02-28 | A kind of micro-capacitance sensor short-term load forecasting method based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109934392A true CN109934392A (en) | 2019-06-25 |
Family
ID=66986086
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910153480.8A Pending CN109934392A (en) | 2019-02-28 | 2019-02-28 | A kind of micro-capacitance sensor short-term load forecasting method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109934392A (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110153802A (en) * | 2019-07-04 | 2019-08-23 | 西南交通大学 | A kind of cutting-tool wear state discrimination method based on convolutional neural networks and long Memory Neural Networks conjunctive model in short-term |
CN110363360A (en) * | 2019-07-24 | 2019-10-22 | 广东工业大学 | A kind of short-term wind power forecast method, device and equipment |
CN110837934A (en) * | 2019-11-11 | 2020-02-25 | 四川大学 | Smart grid short-term residential load prediction method based on deep learning |
CN110991761A (en) * | 2019-12-13 | 2020-04-10 | 新奥数能科技有限公司 | Heat supply load prediction method and device |
CN112001486A (en) * | 2020-08-28 | 2020-11-27 | 河北工业大学 | Load decomposition method based on deep learning |
CN112001519A (en) * | 2020-05-20 | 2020-11-27 | 国网浙江省电力有限公司 | Power load prediction method based on deep neural network |
CN112508261A (en) * | 2020-12-03 | 2021-03-16 | 广东电网有限责任公司 | Neural network-based distribution transformer load hierarchical prediction method and device |
CN112598248A (en) * | 2020-12-16 | 2021-04-02 | 广东电网有限责任公司广州供电局 | Load prediction method, load prediction device, computer equipment and storage medium |
CN112633738A (en) * | 2020-12-30 | 2021-04-09 | 安徽广志电气有限公司 | Distribution control cabinet terminal load distribution method |
CN113962364A (en) * | 2021-10-22 | 2022-01-21 | 四川大学 | Multi-factor power load prediction method based on deep learning |
CN114118592A (en) * | 2021-11-30 | 2022-03-01 | 南京信息工程大学 | Short-term energy consumption prediction system for power consumption end of smart power grid |
CN114330902A (en) * | 2021-12-31 | 2022-04-12 | 河北工业大学 | Distribution line load prediction method based on decoupling mechanism |
CN114595878A (en) * | 2022-02-28 | 2022-06-07 | 武汉大学 | Load prediction method based on generalized load separation result |
CN114330902B (en) * | 2021-12-31 | 2024-10-18 | 河北工业大学 | Distribution line load prediction method based on decoupling mechanism |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108280551A (en) * | 2018-02-02 | 2018-07-13 | 华北电力大学 | A kind of photovoltaic power generation power prediction method using shot and long term memory network |
CN108416690A (en) * | 2018-01-19 | 2018-08-17 | 中国矿业大学 | Load Forecasting based on depth LSTM neural networks |
CN108537122A (en) * | 2018-03-07 | 2018-09-14 | 中国科学院西安光学精密机械研究所 | Image fusion acquisition system containing meteorological parameters and image storage method |
CN108537121A (en) * | 2018-03-07 | 2018-09-14 | 中国科学院西安光学精密机械研究所 | Self-adaptive remote sensing scene classification method based on meteorological environment parameter and image information fusion |
-
2019
- 2019-02-28 CN CN201910153480.8A patent/CN109934392A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108416690A (en) * | 2018-01-19 | 2018-08-17 | 中国矿业大学 | Load Forecasting based on depth LSTM neural networks |
CN108280551A (en) * | 2018-02-02 | 2018-07-13 | 华北电力大学 | A kind of photovoltaic power generation power prediction method using shot and long term memory network |
CN108537122A (en) * | 2018-03-07 | 2018-09-14 | 中国科学院西安光学精密机械研究所 | Image fusion acquisition system containing meteorological parameters and image storage method |
CN108537121A (en) * | 2018-03-07 | 2018-09-14 | 中国科学院西安光学精密机械研究所 | Self-adaptive remote sensing scene classification method based on meteorological environment parameter and image information fusion |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110153802B (en) * | 2019-07-04 | 2020-05-01 | 西南交通大学 | Tool wear state identification method based on convolution neural network and long-term and short-term memory neural network combined model |
CN110153802A (en) * | 2019-07-04 | 2019-08-23 | 西南交通大学 | A kind of cutting-tool wear state discrimination method based on convolutional neural networks and long Memory Neural Networks conjunctive model in short-term |
CN110363360A (en) * | 2019-07-24 | 2019-10-22 | 广东工业大学 | A kind of short-term wind power forecast method, device and equipment |
CN110837934A (en) * | 2019-11-11 | 2020-02-25 | 四川大学 | Smart grid short-term residential load prediction method based on deep learning |
CN110991761B (en) * | 2019-12-13 | 2022-06-28 | 新奥数能科技有限公司 | Heat supply load prediction method and device |
CN110991761A (en) * | 2019-12-13 | 2020-04-10 | 新奥数能科技有限公司 | Heat supply load prediction method and device |
CN112001519A (en) * | 2020-05-20 | 2020-11-27 | 国网浙江省电力有限公司 | Power load prediction method based on deep neural network |
CN112001486A (en) * | 2020-08-28 | 2020-11-27 | 河北工业大学 | Load decomposition method based on deep learning |
CN112001486B (en) * | 2020-08-28 | 2022-07-22 | 河北工业大学 | Load decomposition method based on deep learning |
CN112508261B (en) * | 2020-12-03 | 2021-09-03 | 广东电网有限责任公司 | Neural network-based distribution transformer load hierarchical prediction method and device |
CN112508261A (en) * | 2020-12-03 | 2021-03-16 | 广东电网有限责任公司 | Neural network-based distribution transformer load hierarchical prediction method and device |
CN112598248A (en) * | 2020-12-16 | 2021-04-02 | 广东电网有限责任公司广州供电局 | Load prediction method, load prediction device, computer equipment and storage medium |
CN112633738A (en) * | 2020-12-30 | 2021-04-09 | 安徽广志电气有限公司 | Distribution control cabinet terminal load distribution method |
CN113962364A (en) * | 2021-10-22 | 2022-01-21 | 四川大学 | Multi-factor power load prediction method based on deep learning |
CN113962364B (en) * | 2021-10-22 | 2023-04-18 | 四川大学 | Multi-factor power load prediction method based on deep learning |
CN114118592A (en) * | 2021-11-30 | 2022-03-01 | 南京信息工程大学 | Short-term energy consumption prediction system for power consumption end of smart power grid |
CN114330902A (en) * | 2021-12-31 | 2022-04-12 | 河北工业大学 | Distribution line load prediction method based on decoupling mechanism |
CN114330902B (en) * | 2021-12-31 | 2024-10-18 | 河北工业大学 | Distribution line load prediction method based on decoupling mechanism |
CN114595878A (en) * | 2022-02-28 | 2022-06-07 | 武汉大学 | Load prediction method based on generalized load separation result |
CN114595878B (en) * | 2022-02-28 | 2024-08-13 | 武汉大学 | Load prediction method based on generalized load separation result |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109934392A (en) | A kind of micro-capacitance sensor short-term load forecasting method based on deep learning | |
CN109829543A (en) | A kind of online method for detecting abnormality of data flow based on integrated study | |
CN109635928A (en) | A kind of voltage sag reason recognition methods based on deep learning Model Fusion | |
CN112685504B (en) | Production process-oriented distributed migration chart learning method | |
CN112232522B (en) | Intelligent recommendation and dynamic optimization method for deep roadway support scheme | |
CN109784488B (en) | Construction method of binary convolution neural network suitable for embedded platform | |
CN104881735A (en) | System and method of smart power grid big data mining for supporting smart city operation management | |
CN114676742A (en) | Power grid abnormal electricity utilization detection method based on attention mechanism and residual error network | |
CN112232577A (en) | Power load probability prediction system and method for multi-core intelligent meter | |
CN112990258A (en) | Fault diagnosis method and system for water chilling unit | |
CN114169249B (en) | Artificial intelligent identification method for high-resistance ground fault of power distribution network | |
CN115130375A (en) | Rock burst intensity prediction method | |
CN110059737A (en) | Distribution transformer connection relationship discrimination method based on integrated deep neural network | |
CN115481788B (en) | Phase change energy storage system load prediction method and system | |
CN115099296A (en) | Sea wave height prediction method based on deep learning algorithm | |
CN111311001B (en) | Bi-LSTM network short-term load prediction method based on DBSCAN algorithm and feature selection | |
CN117081063A (en) | Distributed charging load prediction method and system based on GCN-Crossformer model | |
CN117748495A (en) | Photovoltaic power station short-term generation power prediction method based on residual neural network | |
CN115713162A (en) | Distributed low-carbon energy station health state early warning prediction method | |
Chen et al. | Graph neural network-based wind farm cluster speed prediction | |
CN115734274A (en) | Cellular network fault diagnosis method based on deep learning and knowledge graph | |
CN114266676A (en) | Abnormal electricity utilization detection method of genetic optimization Bagging heterogeneous integration model | |
CN113222067A (en) | Intelligent island detection method based on SVM-Adaboost algorithm | |
CN113033898A (en) | Electrical load prediction method and system based on K-means clustering and BI-LSTM neural network | |
CN113282747A (en) | Text classification method based on automatic machine learning algorithm selection |
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: 20190625 |