CN110543942A - Multi-space-time long and short memory depth network accurate prediction method - Google Patents
Multi-space-time long and short memory depth network accurate prediction method Download PDFInfo
- Publication number
- CN110543942A CN110543942A CN201910801527.7A CN201910801527A CN110543942A CN 110543942 A CN110543942 A CN 110543942A CN 201910801527 A CN201910801527 A CN 201910801527A CN 110543942 A CN110543942 A CN 110543942A
- Authority
- CN
- China
- Prior art keywords
- load
- space
- time
- data
- network
- 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
- 230000015654 memory Effects 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000013528 artificial neural network Methods 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims description 16
- 230000007787 long-term memory Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 claims description 2
- 238000004880 explosion Methods 0.000 claims 1
- 238000010606 normalization Methods 0.000 claims 1
- 230000002688 persistence Effects 0.000 claims 1
- 238000007781 pre-processing Methods 0.000 claims 1
- 230000000306 recurrent effect Effects 0.000 claims 1
- 230000006403 short-term memory Effects 0.000 claims 1
- 238000010801 machine learning Methods 0.000 abstract 1
- 238000013179 statistical model Methods 0.000 abstract 1
- 238000010248 power generation Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 7
- 230000004913 activation Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 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)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Tourism & Hospitality (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Business, Economics & Management (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Development Economics (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 invention provides a multi-space-time length memory depth network accurate prediction method which can solve the problem that the load prediction precision of the conventional power system is low due to large power load fluctuation of different time scales and space scales. The invention provides a thought of predicting by combining with the multi-space-time distribution characteristic of the load of the power system, and learns the power load data with different time scales and space scales by utilizing a long and short memory deep neural network, wherein the neural network is a time cycle network, has strong time sequence memory capacity, and is very suitable for predicting the load of the power system. The multi-space-time length memory depth network accurate prediction method provided by the invention can change the input and output numbers of the prediction model according to the difference of the load distribution characteristics of the actually required prediction area, and accurately predict the power loads with different time scales and space scales. Can replace the traditional statistical model and the common machine learning model used for load prediction.
Description
Technical Field
the invention belongs to the field of load prediction of power systems, relates to a method for replacing traditional unified time scale and space scale load prediction, and is suitable for load prediction of power systems.
background
With the proposition and development of the energy internet, the power system combines a plurality of internet frontier information technologies, such as advanced sensors, software application programs and the like, to form a complex multi-network-flow power system which takes distributed energy as a main multi-network system and is closely coupled, and in addition, with the proposition of the concept of the internet of things, the development of the sharing concept and the large-scale practice of the shared commodities in the social range enable the load to show the characteristics of space discreteness and time randomness. Therefore, the difficulty of load prediction of the power system is certainly greatly improved, and the load prediction certainly plays an important role in the future energy internet, so that accurate prediction of the load becomes particularly important.
in the load of the power system, the load prediction is difficult due to the huge load region, the uncertainty is larger with the increase of the prediction region, and in order to cope with the change of the load, the power system now adopts the "unit combination", "economic dispatch", "automatic generation control" and "unit power distribution" to control the system frequency. The uniform time scale cannot rapidly and accurately cope with the change of the load.
In order to solve the problems that load prediction is inaccurate due to the fact that a load area is too large and a frequency control method with a unified time scale cannot timely cope with load changes, the invention provides a multi-space-time accurate prediction method. In order to solve the problem of prediction of huge time series data in a power system, the invention provides a long-short memory deep neural network algorithm applied to multi-time multi-space load distribution. The traditional system frequency control method unifies the automatic power generation control of 4 seconds scale, the economic dispatching of 15 minutes and the unit combination of 1 day time scale, and finally the unit output is a time scale. Aiming at the defect that the traditional frequency control cannot quickly and accurately cope with the change of the load, the invention designs a multi-space-time long-short memory depth network accurate prediction method, which divides a load area into a plurality of sub-areas according to the characteristics of the load area, uses a long-short memory depth neural network to respectively perform accurate prediction, divides the unified time scale of the traditional frequency modulation control into a plurality of time scales according to the characteristics of each link, and uses a deep neural network algorithm to perform training.
At present, the deep learning algorithm is developed rapidly, and the cyclic neural network and the improved algorithm thereof have a great position in the processing of time series data such as classification and regression, so that the method can be applied to the load prediction problem of a power system to improve the accuracy of load prediction.
Disclosure of Invention
The invention provides a multi-space-time length memory depth network accurate prediction method, which is different from a traditional load prediction and frequency control framework. Based on the long and short memory deep network and the deep neural network, the time sequence data are learned, and the model can accurately predict the load and send the data to the frequency control module so as to realize real-time frequency control and adjustment.
The neural network designed in the multi-space-time accurate prediction method can memorize the previously input data for a long time and allow the information to be persistent, and the memory can reach a long distance.
The long and short memory neural network is formed by connecting individual cells, the input is a time sequence data set, the input xt at a certain time point and the state ht-1 at the previous time point are respectively multiplied by the weight value and then added to obtain the state ht at the next time point. The state h is propagated forward in sequence, and the principle is as follows:
h＝xw+hw (1)
where wxh is the connection weight for input xt and last-in-time state quantity ht-1.
In addition, the long and short memory depth network can be stacked up and down in multiple layers, and the state h not only propagates forwards, but also upwards as the input of the next layer. The principle is as follows:
The output of the long and short memory network is all time sequences of the last layer state of the spatial layer, namely:
the long-term memory of the long-short memory deep network is realized by controlling the long-term state c through three gates in the network, namely a memory gate, an input gate and an output gate. The gate is actually an activation function, the input is a vector, the output is a real number vector between 0 and 1, when the gate is 0, it means that no vector can pass, and when the gate is 1, it means that any vector can pass. Therefore, by controlling the opening of the door, useful information can be input, useless information can be forgotten, and required information can be output.
The basic algorithm principle is as follows:
the memory gate is a sigmoid activation function, and the output ft is:
f＝σ[w(xw+hw)+b] (4)
the input gates are sigmoid activation function and tanh activation function, and the sum of the output it is:
i＝σ[w(xw+hw)+b] (5)
the output gate is a sigmoid activation function, and the output ot is as follows:
o＝σ[w(xw+hw)+b] (7)
The cell state ct is:
the output state ht is:
h＝o×tanh(c) (9)
The neural network in the method for accurately predicting the multi-space-time distribution long and short memory deep network has higher requirements on the division of input samples, namely, each time sequence of the long and short memory neural network needing to be trained has the distribution characteristics of the load of the region, and each trained network accurately predicts the load according to the load distribution characteristics of a training region. The data with distribution characteristics can be obtained by simply dividing the geographical area, and can also be divided according to the characteristics of the electricity utilization type. Therefore, in order to characterize each network characteristic as much as possible, data for training the long and short memory deep network can be specially generated as required.
Drawings
FIG. 1 is a schematic diagram of a long-short memory deep network according to the method of the present invention.
FIG. 2 is a schematic diagram of each cell of the long-short memory deep network according to the method of the present invention.
FIG. 3 is a schematic diagram of a multi-time and multi-space scale load prediction real-time scheduling and control framework of the method of the present invention.
Detailed Description
the invention provides a multi-space-time length memory depth network accurate prediction method, which is described in detail in the following with reference to the attached drawings;
FIG. 1 is a schematic diagram of a long-short memory deep network according to the method of the present invention. The input is a section of continuous time sequence data, network state quantities h and c are transmitted forward on a time axis from left to right and a space axis from bottom to top through a plurality of hidden layers, finally, data label output is output, and weights of all layers of the network are adjusted through error backward transmission to achieve the purpose of network training. The use of multiple neural networks of this type trains loads of different distribution characteristics, enabling the network to accurately predict loads of different regions or different types. The specific prediction method is to input current time sequence data of a section of load into the network and obtain the load value sequence of the next time at the output position of the network.
FIG. 2 is a schematic diagram of each cell of the long-short memory deep network according to the method of the present invention. . In the cell, there is an input xt, which is a time series, with two network state quantities h and c, whose changes are controlled via memory gates, input gates and output gates. The use of three gates is the core characteristic of the cell, wherein a memory gate is used for determining what information is discarded from the cell state, the gate reads the state quantity h of the last timestamp and the input quantity x of the current timestamp, and outputs a value between 0 and 1 to the cell state quantity c, wherein 1 represents completely reserved information, and 0 represents completely forgotten information; the entry gate determines what new information is stored in the cell state, wherein the sigmoid layer determines what value to update, and the tanh layer causes a new candidate vector to be added to the cell state; finally, the output gate determines what value needs to be output, where the sigmoid layer determines which portions of the cell state are to be output, and then the cell state is processed by the tanh layer, which ultimately determines the portions of the output.
FIG. 3 is a schematic diagram of a multi-time and multi-space scale load prediction real-time scheduling and control framework of the method of the present invention. The real-time power generation dispatching and controller adopts a multi-time scale frame, is an optimized algorithm frame of a traditional combined algorithm of unit combination, economic dispatching, automatic power generation control and power generation power distribution, and adopts a deep neural network to respectively train the unit combination, the economic dispatching, the automatic power generation control and the power generation power distribution. The multi-time scale deep neural network algorithm can not only improve the optimization control capability through offline pre-training, but also continuously adapt to new environmental changes for updating and training through online learning. In the schematic diagram of the multi-time and multi-space scale load prediction real-time scheduling and control framework of the method, by taking Guangxi Zhuang autonomous region as an example, the load prediction area of the Guangxi Zhuang autonomous region is divided into load prediction areas with multi-space scales according to the urban region range, each area is provided with a set of multi-time scale frequency control algorithm combination, and the framework of the algorithm combination is formed by four parts of unit combination, economic scheduling, automatic power generation control and power generation power distribution. The time scale of each link in the combination is specifically as follows: starting the algorithm of the unit combination problem once every 0 day, and starting the algorithm again every other day; starting an algorithm of an economic scheduling problem every 15 minutes; starting an automatic power generation control algorithm every 4 seconds; the generated power allocation algorithm is initiated every 4 seconds. Each algorithm is represented by a deep neural network, wherein the input of the unit combination neural network is PDi, t, and the output is ui, j, t and Pj, t; the input of the economic dispatch network is PDi, and the output is Pi, j; the input of the automatic power generation control network is ei and delta fi, and the output is delta Pi; the input of the generated power distribution network is Δ Pi, and the output is Δ Pi, j.
Claims (6)
1. A multi-space-time length and short memory depth network accurate prediction method is characterized in that loads can be accurately predicted according to load distribution characteristics of power systems in different space and time, and the method mainly comprises the following steps in the using process:
(1) Dividing the load of a target power system into a plurality of sub-system loads according to the space division distribution characteristics;
(2) Collecting historical data of each subsystem load, and preprocessing the data to generate a training data set;
(3) Determining the input quantity and the output quantity of a corresponding training model by using the characteristics of training data;
(4) Training by using the training data obtained in the step (3) and adopting a long and short memory deep neural network;
(5) And calculating in the trained model by utilizing the real-time load data, and accurately predicting each space-time load.
2. The method for accurately predicting the multi-space-time-length memory depth network as claimed in claim 1, wherein in the step (1), the load of the power system is divided into a plurality of sub-areas according to space and time, and the influence of time and space on load variation is fully considered, so that more characteristic load training data is obtained, the load prediction can be performed in different areas and different time periods, and the prediction result is more accurate.
3. the method as claimed in claim 1, wherein the load training data collected in step (2) is preprocessed by bad data recognition, normalization, one-hot coding, etc., so that the periodicity regularity of the data can be better utilized.
4. the method for accurately predicting the multi-space-time long and short memory depth network as claimed in claim 1, wherein the step (3) of analyzing the data characteristics can better select the input amount and the sample data required by the training, determine a suitable training iteration mode, and determine a suitable model order by analyzing the autocorrelation coefficients of the training samples.
5. The method as claimed in claim 1, wherein the long and short memory deep neural network in step (4) solves the problem of gradient explosion of the conventional recurrent neural network to some extent, and has strong long-distance dependence processing capability, so that the network can memorize the early data of a time sequence with more loads, thereby predicting the later data of the sequence more accurately.
6. the method as claimed in claim 1, wherein the deep neural network in step (4) can memorize the previously inputted data for a long time, allow information persistence, and realize a long distance, and the deep neural network has two states h and c, wherein c is used for memorizing the long-term memory, and wherein h is used for memorizing the short-term memory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910801527.7A CN110543942A (en) | 2019-08-28 | 2019-08-28 | Multi-space-time long and short memory depth network accurate prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910801527.7A CN110543942A (en) | 2019-08-28 | 2019-08-28 | Multi-space-time long and short memory depth network accurate prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110543942A true CN110543942A (en) | 2019-12-06 |
Family
ID=68712253
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910801527.7A Pending CN110543942A (en) | 2019-08-28 | 2019-08-28 | Multi-space-time long and short memory depth network accurate prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110543942A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680838A (en) * | 2020-06-08 | 2020-09-18 | 中国电力科学研究院有限公司 | Air conditioner load aggregated power prediction method and system |
| CN112784965A (en) * | 2021-01-28 | 2021-05-11 | 广西大学 | Large-scale multi-element time series data abnormity detection method oriented to cloud environment |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN107612045A (en) * | 2017-10-09 | 2018-01-19 | 国网安徽省电力公司 | A kind of Power Plant generated energy intelligent dispensing system and method |
| CN107622329A (en) * | 2017-09-22 | 2018-01-23 | 深圳市景程信息科技有限公司 | The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales |
| CN107765347A (en) * | 2017-06-29 | 2018-03-06 | 河海大学 | A kind of Gaussian process returns and the short-term wind speed forecasting method of particle filter |
| CN108830487A (en) * | 2018-06-21 | 2018-11-16 | 王芊霖 | Methods of electric load forecasting based on long neural network in short-term |
| CN108960488A (en) * | 2018-06-13 | 2018-12-07 | 国网山东省电力公司经济技术研究院 | A kind of accurate prediction technique of saturation loading spatial distribution based on deep learning and Multi-source Information Fusion |
| CN109149659A (en) * | 2018-10-08 | 2019-01-04 | 广西大学 | A kind of active Real Time Economic regulation relaxation depth learning method of unified time scale |
| CN109255505A (en) * | 2018-11-20 | 2019-01-22 | 国网辽宁省电力有限公司经济技术研究院 | A kind of short-term load forecasting method of multi-model fused neural network |
| CN109754113A (en) * | 2018-11-29 | 2019-05-14 | 南京邮电大学 | Load forecasting method based on dynamic time warping Yu length time memory |
| CN110009132A (en) * | 2019-03-04 | 2019-07-12 | 三峡大学 | A kind of short-term electric load fining prediction technique based on LSTM deep neural network |
| US20190228481A1 (en) * | 2018-01-22 | 2019-07-25 | Pason Power Inc. | Intelligent energy management system for distributed energy resources and energy storage systems using machine learning |
-
2019
- 2019-08-28 CN CN201910801527.7A patent/CN110543942A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN107765347A (en) * | 2017-06-29 | 2018-03-06 | 河海大学 | A kind of Gaussian process returns and the short-term wind speed forecasting method of particle filter |
| CN107622329A (en) * | 2017-09-22 | 2018-01-23 | 深圳市景程信息科技有限公司 | The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales |
| CN107612045A (en) * | 2017-10-09 | 2018-01-19 | 国网安徽省电力公司 | A kind of Power Plant generated energy intelligent dispensing system and method |
| US20190228481A1 (en) * | 2018-01-22 | 2019-07-25 | Pason Power Inc. | Intelligent energy management system for distributed energy resources and energy storage systems using machine learning |
| CN108960488A (en) * | 2018-06-13 | 2018-12-07 | 国网山东省电力公司经济技术研究院 | A kind of accurate prediction technique of saturation loading spatial distribution based on deep learning and Multi-source Information Fusion |
| CN108830487A (en) * | 2018-06-21 | 2018-11-16 | 王芊霖 | Methods of electric load forecasting based on long neural network in short-term |
| CN109149659A (en) * | 2018-10-08 | 2019-01-04 | 广西大学 | A kind of active Real Time Economic regulation relaxation depth learning method of unified time scale |
| CN109255505A (en) * | 2018-11-20 | 2019-01-22 | 国网辽宁省电力有限公司经济技术研究院 | A kind of short-term load forecasting method of multi-model fused neural network |
| CN109754113A (en) * | 2018-11-29 | 2019-05-14 | 南京邮电大学 | Load forecasting method based on dynamic time warping Yu length time memory |
| CN110009132A (en) * | 2019-03-04 | 2019-07-12 | 三峡大学 | A kind of short-term electric load fining prediction technique based on LSTM deep neural network |
Non-Patent Citations (8)
| Title |
|---|
| LINFEIYIN等: "Relaxed deep learning for real-time economic generation dispatch and control with unified time scale", 《ENERGY》 * |
| 孙大帅等: "基于复杂系统的短期负荷预测研究及系统设计", 《上海理工大学学报》 * |
| 林艳艳等: "多微网系统二次频率控制策略研究", 《机电工程》 * |
| 肖白等: "空间电力负荷预测方法综述与展望", 《中国电机工程学报》 * |
| 贾逸伦等: "基于有效度和相对熵的短期电力负荷组合预测方法", 《电力科学与技术学报》 * |
| 钟庆等: "基于负荷空间分布预测的电网规划方法研究(II):电力系统负荷分区分布预测", 《国际电力》 * |
| 陈皓勇等: "含高渗透率风电的微网系统鲁棒经济调度方法", 《控制理论与应用》 * |
| 马晓东等: "基于区域划分的分布式短期负荷预测方案设计", 《自动化与仪器仪表》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111680838A (en) * | 2020-06-08 | 2020-09-18 | 中国电力科学研究院有限公司 | Air conditioner load aggregated power prediction method and system |
| CN112784965A (en) * | 2021-01-28 | 2021-05-11 | 广西大学 | Large-scale multi-element time series data abnormity detection method oriented to cloud environment |
| CN112784965B (en) * | 2021-01-28 | 2022-07-29 | 广西大学 | Large-scale multi-element time series data anomaly detection method oriented to cloud environment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108280551B (en) | Photovoltaic power generation power prediction method utilizing long-term and short-term memory network | |
| CN109685252B (en) | Building energy consumption prediction method based on cyclic neural network and multi-task learning model | |
| CN110766212B (en) | Ultra-short-term photovoltaic power prediction method for historical data missing electric field | |
| CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
| CN104820877A (en) | Photovoltaic system generation power prediction method based on cloud adaptive PSO-SNN | |
| Jiang et al. | Day‐ahead renewable scenario forecasts based on generative adversarial networks | |
| CN113919545A (en) | Photovoltaic power generation power prediction method and system with integration of multiple data models | |
| CN110543942A (en) | Multi-space-time long and short memory depth network accurate prediction method | |
| Sen | Time series prediction based on improved deep learning | |
| Chen et al. | A multivariate grey RBF hybrid model for residual useful life prediction of industrial equipment based on state data | |
| Tavares et al. | Comparison of PV power generation forecasting in a residential building using ANN and DNN | |
| Alharbi et al. | Short-term wind speed and temperature forecasting model based on gated recurrent unit neural networks | |
| CN116865343B (en) | Model-free self-adaptive control method, device and medium for distributed photovoltaic power distribution network | |
| Prema et al. | LSTM based Deep Learning model for accurate wind speed prediction | |
| Wei et al. | A combination forecasting method of grey neural network based on genetic algorithm | |
| Bolstad et al. | Day-ahead load forecasting using explainable artificial intelligence | |
| CN116880201A (en) | Water network channel state control system based on fuzzy neural network | |
| CN114372615A (en) | Short-term photovoltaic power prediction method and system considering spatial correlation | |
| Huang et al. | Short-term load forecasting based on similar day approach and intelligent algorithm using analytic hierarchy process | |
| CN113935600A (en) | Deep learning-based adaptive economic dispatching system and method | |
| Kondaiah et al. | Short-term load forecasting with deep learning | |
| CN112183814A (en) | Short-term wind speed prediction method | |
| Guillén-Asensio et al. | Battery state prediction in photovoltaic standalone installations | |
| Sun et al. | Reliability growth prediction method based on GA-Elman neural network | |
| Yang et al. | Electricity load forecasting based on long and short-term memory neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191206 |
|
| WD01 | Invention patent application deemed withdrawn after publication |