CN110866840A - Database modeling method for power load characteristic quantity training based on knowledge graph - Google Patents
Database modeling method for power load characteristic quantity training based on knowledge graph Download PDFInfo
- Publication number
- CN110866840A CN110866840A CN201911098420.7A CN201911098420A CN110866840A CN 110866840 A CN110866840 A CN 110866840A CN 201911098420 A CN201911098420 A CN 201911098420A CN 110866840 A CN110866840 A CN 110866840A
- Authority
- CN
- China
- Prior art keywords
- data
- database
- load characteristic
- matrix
- power
- 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
- 238000012549 training Methods 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 title claims abstract description 21
- 239000011159 matrix material Substances 0.000 claims abstract description 68
- 230000000694 effects Effects 0.000 claims abstract description 12
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 9
- 230000009466 transformation Effects 0.000 claims abstract description 4
- 238000003064 k means clustering Methods 0.000 claims abstract 3
- 230000006870 function Effects 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 5
- 238000005728 strengthening Methods 0.000 claims description 3
- 239000002131 composite material Substances 0.000 abstract description 6
- 230000008569 process Effects 0.000 abstract description 4
- 238000004458 analytical method Methods 0.000 abstract description 3
- 238000005070 sampling Methods 0.000 description 10
- 238000013527 convolutional neural network Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000009825 accumulation Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Public Health (AREA)
- Mathematical Physics (AREA)
- Water Supply & Treatment (AREA)
- Probability & Statistics with Applications (AREA)
- Biophysics (AREA)
- Human Resources & Organizations (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a database modeling method for power load characteristic quantity training based on a knowledge graph, which comprises the following steps: acquiring the use condition of an electric appliance; obtaining waveform data processed under a steady state; executing a waveform decomposition algorithm on the waveform data to acquire load characteristic matrix data; importing the load characteristic matrix data acquired by executing the waveform decomposition algorithm in S3 on the processed waveform data into a database; calling a K-Means clustering algorithm to classify the data subjected to wavelet transformation in S4; performing CNN algorithm on the load characteristic matrix data for identification; and removing the influence of the edge matrix through Gaussian blur, searching a central submatrix which is equal to the input matrix in the database, giving confidence coefficient and outputting the central submatrix. The invention effectively solves the problem that the composite load is difficult to classify due to the harmonic effect generated in the analysis process of the power system.
Description
Technical Field
The invention relates to the technical field of electrical informatization, in particular to a database modeling method for power load characteristic quantity training based on a knowledge graph.
Background
The electric load is also called an electric load, and the sum of electric power taken by electric equipment of an electric energy user to an electric power system at a certain moment is called an electric load. The characteristic quantities of the electrical load include, but are not limited to, information of current, power, peak value, phase angle variation of voltage and current, power factor, harmonic component, and the like.
The effective steady-state characteristics of the electric appliance can be uniquely identified by extracting from waveforms of voltage, current, power, peak value, phase angle change of voltage and current, power factor, harmonic component and the like, the total power consumption is decomposed into the power consumption of each electric device and collected and stored, and the current sensor facilities used for electrical monitoring do not have storage and self-learning functions for the electric load, and particularly, the electric devices are usually in a working mode of series-parallel connection and common-end operation of multiple electric devices, so that the difficulty is brought to the characteristic quantity extraction and analysis of the electric load.
In order to solve the problems, close attention is paid to research on a method for carrying out non-invasive power load identification by utilizing various learning algorithms derived based on a neural network, the neural network algorithm is seriously developed in the field of artificial intelligence, particularly face identification and voice identification, but in the face of a complex actual running state of multiple devices in parallel in a multi-state and multi-connection mode of power equipment, the traditional neural network algorithm cannot achieve good effects on identifying and analyzing complicated power loads and on identifying and analyzing power flow.
Disclosure of Invention
The invention aims to provide a database modeling method for power load characteristic quantity training based on a knowledge graph, which aims to solve the problems in the background technology and solve the problem of difficulty in extracting the power load characteristic quantity, and enables the power load identification to have a self-learning function, an associative storage function, a self-excitation high-order database function and an evolutionary database function through a training algorithm of a neural network, thereby realizing the intelligent analysis of the power load effect.
In order to achieve the purpose, the invention provides the following technical scheme:
a database modeling method for power load characteristic quantity training based on a knowledge graph comprises the following steps:
s1, collecting information such as user voltage, current, power, peak value, phase angle change of voltage and current, power factor, harmonic component and the like on a user electric meter through the Internet of things to obtain the service condition of the electric appliance;
s2, initializing waveform data, denoising for the first time, and carrying out Gaussian blurring on multi-period waveforms to obtain waveform data processed under a steady state;
s3, executing a waveform decomposition algorithm on the waveform data to acquire load characteristic matrix data;
and S4, importing the load feature matrix data acquired by executing the waveform decomposition algorithm in S3 on the processed waveform data into a database, and executing a K-Means algorithm on the feature matrix data in the database called by the input data for identification. If yes, returning a load identification result; if not, outputting the database which does not exist, selectively importing the database, and deleting the load characteristic matrix data if not importing;
s5, calling a CNN algorithm to classify the data subjected to the wavelet transformation in the S4, and generating multi-component power equipment interference-removing load characteristic matrix data to be imported into a database;
s6, executing a K-Means algorithm to the load characteristic matrix data for identification, if the load characteristic matrix data exist, returning a load identification result, and if the load characteristic matrix data do not exist, analyzing the matrix;
s7, removing the influence of the edge matrix through Gaussian blur, strengthening the offset effect of the central matrix identified in the CNN algorithm, searching a central sub-matrix in the database, which is equal to the input matrix, giving confidence and outputting the central sub-matrix.
As a further scheme of the invention: in S1, the Internet of things means that information of the power equipment is collected in real time through a sensor device and technology.
As a further scheme of the invention: in S4, the load characteristic matrix data is matrix data composed of characteristic quantities of the power load.
As a further scheme of the invention: in S5, the CNN algorithm is an algorithm in which multiple layers of convolution kernels jointly affect binary classification of data.
As a still further scheme of the invention: in S6, the K-Means algorithm is to classify the sample data set using probability statistics, and to find similar data between the experimental input data and the sample data set to give confidence.
Compared with the prior art, the invention has the beneficial effects that: the method for creating training data based on the relational database in the self-learning mode and training the database based on the self-excitation of the neural network solves the problems of rare samples, high sampling cost, high sampling technical requirement and high sampling difficulty of part of special power equipment through a self-training framework. Particularly, under the conditions that the initial data collection amount of the database is small and the recognition result is not accurate enough, the over-fitting negative effect generated in the process of training the algorithm is effectively reduced.
Drawings
FIG. 1 is a flow chart of a database modeling method for power load characteristic quantity training based on knowledge graph.
FIG. 2 is a schematic diagram of a neural network in a knowledge-graph-based database modeling method for power load characteristic quantity training.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1 and fig. 2, in an embodiment of the present invention, a method for modeling a database trained on power load characteristic quantities based on a knowledge graph includes the following steps:
s1, collecting information such as user voltage, current, power, peak value, phase angle change of voltage and current, power factor, harmonic component and the like on a user electric meter through the Internet of things to obtain the service condition of the electric appliance;
in S1, the Internet of things means that the information of the power equipment is collected in real time through a sensor device and technology, and the collected sampling frequency is greater than that of the power equipmentThe Nyquist sampling theorem is satisfied,for the highest frequencies in the signal, the sampling frequency should be increased appropriately when analyzing the multicomponent power device.
S2, initializing waveform data, denoising for the first time, and carrying out Gaussian blurring on multi-period waveforms to obtain waveform data processed under a steady state;
s3, executing a waveform decomposition algorithm on the waveform data to acquire load characteristic matrix data;
and S4, importing the load feature matrix data acquired by executing the waveform decomposition algorithm in S3 on the processed waveform data into a database, and executing a K-Means algorithm on the feature matrix data in the database called by the input data for identification. If yes, returning a load identification result; if not, outputting the database which does not exist, selectively importing the database, and deleting the load characteristic matrix data if not importing;
if the input initial state data is multi-component power equipment interference data, the signals may be polluted by random noise to different degrees in the processes of excitation, transmission and detection, and the noise interference is particularly serious in small signal acquisition and measurement. Further denoising should be performed at this time to obtain data closer to the true value by distinguishing the effective signal from the gaussian noise using wavelet transform.
In S4, the load characteristic matrix data is matrix data composed of characteristic quantities of the power load.
S5, calling a CNN algorithm to classify the data subjected to the wavelet transformation in the S4, and generating multi-component power equipment interference-removing load characteristic matrix data to be imported into a database;
in S5, the CNN algorithm is an algorithm in which multivariate kernel functions jointly affect binary classification of data.
S6, executing a K-Means algorithm to the load characteristic matrix data for identification, if the load characteristic matrix data exist, returning a load identification result, and if the load characteristic matrix data do not exist, analyzing the matrix; the K-Means algorithm is that probability statistics is used for classifying sample data sets, and confidence is given by searching similar data of experiment input data and the sample data sets.
In order to more fully and accurately utilize the characteristic quantity of the power load, and to consider that a considerable part of the characteristic quantity of the power load cannot be accurately predicted through linear accumulation, and outliers existing in a small quantity of the characteristic quantity of the power load can influence the prediction effect of power load identification, on the basis of screening the characteristic quantity of the power load, characteristic attributes under series-parallel connection are set for all the characteristic quantity of the power load, a weight judgment criterion is defined at the same time, the weight judgment of the characteristic quantity of the power load is realized, and a self-training framework is introduced to improve the data utilization rate of a characteristic matrix of the power load so as to improve the prediction effect and the application range of a model after training and order increase.
The characteristic attribute of the composite power load characteristic quantity is a power load characteristic quantity which defines voltage, current, power, peak value, phase angle change of voltage and current, power factor, harmonic component and the like of more than two different power devices passing through the same node in physics (the unit of the characteristic quantity of two waveform accumulation when two different power devices are connected in series and parallel uses an international standard unit). The corresponding relation between the composite power load characteristic quantity and the two or more power load characteristic quantities is as follows:
whereinA matrix of characteristic quantities representative of the composite electrical load,a power load characteristic quantity matrix representing the nth power device,characteristic quantity of nth power load representing nth power equipmentWhich represents the operation of the join operation,representing the connection operation relationship between two electric power devices.
The weight judgment criterion is to set per unit value weight, perform weighted calculation on each power load characteristic quantity matrix data, and distinguish the influence of the power load characteristic quantity on different power equipment and power networks with different components.
The self-training framework comprises two training algorithms with operations not interfering with each other, wherein the evolutionary algorithm takes single-power-equipment interference-free data as a sample, the sample differentiation is realized through an excitation function, and a hidden layer selects a sparse evolutionary sample meeting conditions to return to the sub-database. The training order-increasing algorithm realizes the data self of the composite electric power load characteristic matrix by calling the non-interference load characteristic matrix data of the single electric power equipment in the databaseGenerating, namely DA conversion of different states is carried out on data of a composite power load characteristic matrix, the obtained waveform data is denoised, then a waveform decomposition algorithm is executed to obtain data of a multi-component power equipment interference-free load characteristic matrix, the training result is similar to experimental test data, sequential increasing order is executed to be second order, CPU operational capability is judged, if the operational capability is insufficient to increase order to third order or above in set clock time, sequential increasing order training is stopped, reverse increasing order is executed to n-1 order, and similar is repeatedAnd finishing the full-stage training.
S7, removing the influence of the edge matrix through Gaussian blur, strengthening the offset effect of the central matrix identified in the CNN algorithm, searching a central sub-matrix in the database, which is equal to the input matrix, giving confidence and outputting the central sub-matrix.
The analytical matrix determines that matrix data in a database is logically larger than input matrix data, so that a matrix similar to the input data is selected as a database matrix center, the influence of an edge matrix is removed through Gaussian blur, the offset effect of the center matrix identified in a CNN algorithm is strengthened, a center submatrix equal to the input matrix in the database is searched, confidence is given, and the center submatrix is output.
According to the invention, voltage, current and power information is collected through the Internet of things, and characteristic quantities such as current, power, peak value, phase angle change of voltage and current, power factor and the like are extracted; denoising and abnormal values of current, voltage and power signals are removed, and phase information is ensured to be accurate through voltage and current phase correction; identifying the initial state data after data processing to calculate and identify the accuracy rate, and adjusting an algorithm model; and then carrying out analog signal digital parameter input data setting parameters and parameters needing debugging on harmonic components meeting the Nyquist sampling rate until the accuracy is greater than a set threshold value, and debugging according to the parameters needing debugging: learning rate, hidden layer number, hidden layer node number, activation function and the like, and then outputting a power equipment load identification result and confidence coefficient thereof.
Especially, due to the development of the electrical big data technology, a large number of label-free samples are more and more easily obtained, and the obtaining cost of the label samples is still high, so that the label samples are few in the field of some special power equipment, and the prediction effect of the model is difficult to ensure by the traditional modeling method when the label samples are few. The self-training framework solves the problems of rare samples, high sampling cost, high sampling technical requirement and high sampling difficulty of part of special power equipment. Particularly, under the conditions that the initial data collection amount of the database is small and the recognition result is not accurate enough, the over-fitting negative effect generated in the process of training the algorithm is effectively reduced.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.
Claims (5)
1. A database modeling method for power load characteristic quantity training based on a knowledge graph is characterized by comprising the following steps:
s1, collecting user voltage, current, power, peak value, phase angle change of voltage and current, power factor and harmonic component information on a user electric meter through the Internet of things to obtain the service condition of the electric appliance;
s2, initializing waveform data, denoising for the first time, and carrying out Gaussian blurring on multi-period waveforms to obtain waveform data processed under a steady state;
s3, executing a waveform decomposition algorithm on the waveform data to acquire load characteristic matrix data;
s4, importing the load characteristic matrix data acquired by executing the waveform decomposition algorithm in S3 on the processed waveform data into a database, and executing a K-Means clustering algorithm on the characteristic matrix data in the database called by the input data for identification; if yes, returning a load identification result; if not, outputting the database which does not exist, selectively importing the database, and deleting the load characteristic matrix data if not importing;
s5, calling a CNN algorithm to classify the data subjected to the wavelet transformation in the S4, and generating multi-component power equipment interference-removing load characteristic matrix data to be imported into a database;
s6, performing K-Means clustering algorithm on the load characteristic matrix data for identification, if the load characteristic matrix data exist, returning a load identification result, and if the load characteristic matrix data do not exist, analyzing the matrix;
s7, removing the influence of the edge matrix through Gaussian blur, strengthening the offset effect of the central matrix identified in the CNN algorithm, searching a central sub-matrix in the database, which is equal to the input matrix, giving confidence and outputting the central sub-matrix.
2. The knowledge-graph-based database modeling method for power load characteristic quantity training according to claim 1, wherein in S1, the internet of things means that information of power equipment is collected in real time through sensor devices and technologies.
3. The method for modeling a database based on training of characteristic quantities of power loads according to claim 2, wherein in S4, the load characteristic matrix data is matrix data composed of characteristic quantities of power loads.
4. The knowledge-graph-based database modeling method for power load characteristic quantity training according to claim 3, wherein in S5, the CNN algorithm is an algorithm in which multiple kernel functions jointly influence binary classification of data.
5. The method for modeling a database based on power load characteristic quantity training of knowledge-graph according to claim 1 or 4, wherein in S6, the K-Means algorithm is to classify the sample data set by using probability statistics and to find the similar data of the experimental input data and the sample data set to give confidence.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911098420.7A CN110866840A (en) | 2019-11-12 | 2019-11-12 | Database modeling method for power load characteristic quantity training based on knowledge graph |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911098420.7A CN110866840A (en) | 2019-11-12 | 2019-11-12 | Database modeling method for power load characteristic quantity training based on knowledge graph |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110866840A true CN110866840A (en) | 2020-03-06 |
Family
ID=69653582
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911098420.7A Pending CN110866840A (en) | 2019-11-12 | 2019-11-12 | Database modeling method for power load characteristic quantity training based on knowledge graph |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110866840A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112183878A (en) * | 2020-10-13 | 2021-01-05 | 东北大学 | Power load prediction method combining knowledge graph and neural network |
CN113687164A (en) * | 2021-08-25 | 2021-11-23 | 天津大学 | Non-invasive load event two-stage self-adaptive detection method |
CN117113159A (en) * | 2023-10-23 | 2023-11-24 | 国网山西省电力公司营销服务中心 | Deep learning-based power consumer side load classification method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103177188A (en) * | 2013-04-02 | 2013-06-26 | 东南大学 | Electric system load dynamic characteristic classifying method based on characteristic mapping |
CN103777091A (en) * | 2013-12-13 | 2014-05-07 | 国家电网公司 | High-speed rail electric energy quality monitoring data classification method based on K mean value |
CN108564546A (en) * | 2018-04-18 | 2018-09-21 | 厦门美图之家科技有限公司 | Model training method, device and photo terminal |
CN110277086A (en) * | 2019-06-25 | 2019-09-24 | 中国科学院自动化研究所 | Phoneme synthesizing method, system and electronic equipment based on dispatching of power netwoks knowledge mapping |
-
2019
- 2019-11-12 CN CN201911098420.7A patent/CN110866840A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103177188A (en) * | 2013-04-02 | 2013-06-26 | 东南大学 | Electric system load dynamic characteristic classifying method based on characteristic mapping |
CN103777091A (en) * | 2013-12-13 | 2014-05-07 | 国家电网公司 | High-speed rail electric energy quality monitoring data classification method based on K mean value |
CN108564546A (en) * | 2018-04-18 | 2018-09-21 | 厦门美图之家科技有限公司 | Model training method, device and photo terminal |
CN110277086A (en) * | 2019-06-25 | 2019-09-24 | 中国科学院自动化研究所 | Phoneme synthesizing method, system and electronic equipment based on dispatching of power netwoks knowledge mapping |
Non-Patent Citations (2)
Title |
---|
ABDUL QAYYUM等: "Classification of EEG Learning and Resting States using 1D-Convolutional Neural Network for Cognitive Load Assesment", 《2018 IEEE STUDENT CONFERENCE ON RESEARCH AND DEVELOPMENT (SCORED)》, 13 May 2019 (2019-05-13), pages 1 - 5 * |
杨茂等: "基于数据驱动的超短期风电功率预测综述", 《电力***保护与控制》, vol. 47, no. 13, 1 July 2019 (2019-07-01), pages 171 - 186 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112183878A (en) * | 2020-10-13 | 2021-01-05 | 东北大学 | Power load prediction method combining knowledge graph and neural network |
CN113687164A (en) * | 2021-08-25 | 2021-11-23 | 天津大学 | Non-invasive load event two-stage self-adaptive detection method |
CN117113159A (en) * | 2023-10-23 | 2023-11-24 | 国网山西省电力公司营销服务中心 | Deep learning-based power consumer side load classification method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109975673B (en) | Method for identifying fault arc at direct current side of photovoltaic microgrid | |
CN110866840A (en) | Database modeling method for power load characteristic quantity training based on knowledge graph | |
Gao et al. | Power quality disturbance classification under noisy conditions using adaptive wavelet threshold and DBN-ELM hybrid model | |
Yuan et al. | A multiscale feature learning scheme based on deep learning for industrial process monitoring and fault diagnosis | |
CN112881942B (en) | Abnormal current diagnosis method and system based on wavelet decomposition and empirical mode decomposition | |
CN111008726A (en) | Class image conversion method in power load prediction | |
CN110458189A (en) | Compressed sensing and depth convolutional neural networks Power Quality Disturbance Classification Method | |
Sindi et al. | An adaptive deep learning framework to classify unknown composite power quality event using known single power quality events | |
Liu et al. | Drought prediction method based on an improved CEEMDAN-QR-BL model | |
CN114548586A (en) | Short-term power load prediction method and system based on hybrid model | |
CN112085111A (en) | Load identification method and device | |
Liu et al. | A novel dual-attention optimization model for points classification of power quality disturbances | |
Sun et al. | The CEEMD-LSTM-ARIMA model and its application in time series prediction | |
CN113627539A (en) | Rolling bearing composite fault diagnosis method and system, storage medium and computing equipment | |
CN114036977A (en) | Fan power quality disturbance monitoring method based on two-dimensional image coding features | |
CN113987910A (en) | Method and device for identifying load of residents by coupling neural network and dynamic time planning | |
CN117421994A (en) | Edge application health monitoring method and system | |
CN117669656A (en) | TCN-Semi PN-based direct-current micro-grid stability real-time monitoring method and device | |
CN110633741B (en) | Time sequence classification method based on improved impulse neural network | |
CN116776245A (en) | Three-phase inverter equipment fault diagnosis method based on machine learning | |
CN111090679A (en) | Time sequence data representation learning method based on time sequence influence and graph embedding | |
CN116318925A (en) | Multi-CNN fusion intrusion detection method, system, medium, equipment and terminal | |
CN115936196A (en) | Monthly rainfall model prediction method based on time sequence convolution network | |
Chiam et al. | Global attention-based LSTM for noisy power quality disturbance classification | |
CN116933119A (en) | Signal data trend removal method based on convolutional 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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200306 |
|
RJ01 | Rejection of invention patent application after publication |