CN111008584A - Electric energy quality measurement deficiency repairing method of fuzzy self-organizing neural network - Google Patents

Electric energy quality measurement deficiency repairing method of fuzzy self-organizing neural network Download PDF

Info

Publication number
CN111008584A
CN111008584A CN201911197788.9A CN201911197788A CN111008584A CN 111008584 A CN111008584 A CN 111008584A CN 201911197788 A CN201911197788 A CN 201911197788A CN 111008584 A CN111008584 A CN 111008584A
Authority
CN
China
Prior art keywords
data
missing
dimensional
electric energy
steps
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.)
Granted
Application number
CN201911197788.9A
Other languages
Chinese (zh)
Other versions
CN111008584B (en
Inventor
杨挺
何周泽
盆海波
李扬
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tianjin University
Original Assignee
Tianjin University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tianjin University filed Critical Tianjin University
Priority to CN201911197788.9A priority Critical patent/CN111008584B/en
Publication of CN111008584A publication Critical patent/CN111008584A/en
Application granted granted Critical
Publication of CN111008584B publication Critical patent/CN111008584B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08—Feature extraction
    • G06F2218/10—Feature extraction by analysing the shape of a waveform, e.g. extracting parameters relating to peaks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications

Abstract

The invention discloses a method for repairing the missing of electric energy quality measurement data based on a fuzzy self-organizing neural network, which is executed by a computer program and comprises the following steps: the method comprises the following steps: 1) inputting a missing electric energy quality one-dimensional measurement data set; 2) carrying out segmented interception on the sampled one-dimensional waveform data according to N electric energy quality sampling periods; 3) converting the matrix x into an image by a graying method; 4) extracting characteristic values of the two-dimensional harmonic gray scale image and performing normalization processing; 5) determining the optimal clustering number of the power grid harmonic data; 6) constructing an objective function by taking the weighted square sum of the distances from each sample to all the clustering centers as a target; 7) completing the repair of the two-dimensional gray scale image; 8) the repaired data of each layer are fused, and through experimental data analysis, under the condition of random deletion or continuous deletion, compared with the existing algorithm, the method provided by the invention has lower repairing error and higher signal-to-noise ratio under the conditions of low data loss rate and high data loss rate.

Description

Electric energy quality measurement deficiency repairing method of fuzzy self-organizing neural network
Technical Field
The invention relates to a method for repairing measurement data loss, in particular to a method for repairing power quality measurement loss data, and particularly relates to a method for repairing power quality measurement data loss of a fuzzy self-organizing neural network.
Background
The universal power Internet of Things (UEIoT) realizes comprehensive perception and intelligent measurement of a power system, and provides strong information support for safe, stable and economic operation of a power grid. The ubiquitous awareness big data underlying the UEIoT is the basis for overall system situational awareness and state recognition. The power grid harmonic monitoring data are the key points for mastering a harmonic rule, realizing harmonic treatment and improving the power quality. However, no matter the sampling mode of classical Nyquist sampling or compressed sensing is adopted, the problem that part of the acquired harmonic signals are lost is often caused by faults of sensors, transmission equipment, conversion equipment and the like; or the phenomenon of data loss caused by the interference of the channel during the propagation of the communication channel, such as the power line carrier. Due to the non-repeatability of the power grid data acquisition, under the condition of insufficient redundancy, the missing harmonic data is used for analysis, and the conclusion obtained undoubtedly has larger deviation from the correct rule; when a signal is reconstructed by compression sampling, because each sampling point contains a large amount of information, the loss of each sampling point can cause great influence on signal reconstruction. Therefore, how to accurately and effectively repair the missing data and recover the original appearance of the acquired data is the key point of harmonic waveform data management.
The repair strategy provided by the invention is used for repairing the missing data through the similarity relation between different data according to the obvious measurement time sequence characteristic of the power quality data of the power grid and the autocorrelation and the regularity of the harmonic variation of the data, thereby greatly reducing the repair error.
Disclosure of Invention
The technical problem to be solved by the invention is to provide a method for repairing the power quality measurement data loss of the fuzzy self-organizing neural network, wherein under the conditions of random loss or continuous loss, the method has lower repairing errors and higher signal-to-noise ratio under the conditions of low data loss rate and high data loss rate.
The technical scheme adopted by the invention is as follows: a repair method suitable for power quality measurement data loss comprises the following steps:
1) inputting a missing electric energy quality one-dimensional measurement data set;
2) the one-dimensional waveform data obtained by sampling is segmented and intercepted according to N electric energy quality sampling periods (N is preferably 20 through experimental analysis), and the mapping rule x (i, j) ═ x (N) | N ═ x (i-1) N of a two-dimensional matrix x is usedl+j j≤nlMapping the signal into a row or a column in a two-dimensional matrix x, and performing two-dimensional truncation and recombination on the one-dimensional signal;
3) calculating a space gray value P (i, j) [ (255- (254 × (m-x (i, j)/(m-n)) ], and converting the matrix x into an image by a graying method;
4) extracting characteristic value X of two-dimensional harmonic gray imagej＝[X1,j,X2,j,…,Xm+l,j]And carrying out normalization treatment;
5) determining an optimal clustering number of the power grid harmonic data, comprising:
(1) calculating coefficients of aggregation level
Figure BDA0002295090920000011
And coefficients characterizing the degree of dispersion from class to class
Figure BDA0002295090920000012
(2) α is the index lambda for calculating the overall clustering effectmax+(1-βmin)；
(3) When | | lambda (k) -lambda (k-1) | < epsilon, judging that the total times of clustering iteration before convergence is the optimal classification number k of clustering; otherwise, returning to the step (1);
6) and (3) constructing an objective function by taking the weighted square sum of the distances from each sample to all the cluster centers as an objective:
Figure BDA0002295090920000021
and constructing a membership matrix U ═ Uij]To do so by
Figure BDA0002295090920000022
And training an optimal clustering mode for constraint conditions. The updating of the parameters in the objective function comprises the following steps:
(1) updating degree of membership
Figure BDA0002295090920000023
And fuzzy index
Figure BDA0002295090920000024
(2) Updating learning efficiency
Figure BDA0002295090920000025
(3) Updating neuron node weights
Figure BDA0002295090920000026
(4) Updating weight vectors
Figure BDA0002295090920000027
If | | Δ w | | non-calculation2＝||w(t+1)-w(t)||2>When epsilon is larger, return to (1); otherwise, the loop is ended.
7) Completing the repair of the two-dimensional gray scale image, comprising:
(1) traversing all data, and searching and recording the position sequence and the layering sequence of each missing point;
(2) searching the same missing value in all layers, finding out the maximum value of the number of the information available around the missing value, and extracting the position information of the missing point and the information of the layer where the missing point is located;
(3) the missing point with the largest number of available information around is repaired firstly in the layer;
(4) deleting the position information and the layer information of the repaired point, and if missing data exist, entering the step (2) to search again; otherwise go to 8).
8) And fusing the repaired data of each layer, and restoring the image information into a waveform signal.
Advantageous effects
The invention relates to a repairing method suitable for electric energy quality measurement data loss, which has the following characteristics:
the electric measurement data borne by the ubiquitous power internet of things are interfered in all links such as collection, transmission and conversion, so that data loss is caused, and the state estimation precision and the stable operation of the system are influenced. Aiming at the defect that the traditional restoration strategy only considers the transverse distribution rule of one-dimensional measurement data to cause lower data restoration precision, the invention fully considers the neighborhood data of the measurement data missing point of the power system and the periodic change rule of the measurement data, and provides the electric energy quality measurement data missing restoration method based on the fuzzy self-organizing neural network. According to the invention, the time-space correlation analysis among data is improved by mapping the one-dimensional measurement data of the power quality into a two-dimensional gray image. And clustering the original data by adopting an artificial intelligence FSOM neural network algorithm at the later stage, analyzing and constructing multilayer characteristic values of the data, and performing layered restoration on the clustered data. Through experimental data analysis, under the condition of random deletion or continuous deletion, compared with the existing algorithm, the FSOM restoration algorithm provided by the invention has lower restoration error and higher signal-to-noise ratio under the conditions of low data loss rate and high data loss rate.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention. In the drawings:
FIG. 1 is a two-dimensional truncated rebinning of one-dimensional data;
FIG. 2 is an FSOM neural network map;
FIG. 3 is an algorithm implementation flow;
FIG. 4 is a comparison of the repairing effect of voltage ramp-up and voltage ramp-down defect measurement;
FIG. 5 is a comparison of the repairing effect of voltage harmonic missing measurement data.
Detailed Description
The following detailed description of embodiments of the present invention is provided in connection with the accompanying drawings and examples. The following examples are intended to illustrate the invention but are not intended to limit the scope of the invention.
According to the method for repairing the power quality measurement data loss, the power quality one-dimensional measurement data are mapped into the two-dimensional gray image in the early stage, and time-space correlation analysis among the data is improved. And then, clustering the original data by adopting an artificial intelligence FSOM neural network algorithm, analyzing and constructing multilayer characteristic values of the data, and finally, performing layered restoration on the clustered data.
The method comprises the following steps: inputting a missing electric energy quality one-dimensional measurement data set
Step two: two-dimensional truncated reconstruction of waveform data
And carrying out high-frequency sampling on the voltage or current waveform of the monitoring point to obtain a one-dimensional signal. By using the truncation and recombination method shown in fig. 1, the sampled one-dimensional waveform data is segmented and intercepted according to N power quality sampling periods (it is more appropriate to take 20 by experimental analysis N), and the data is mapped into rows or columns in a two-dimensional matrix, so that the two-dimensional truncation and recombination of signals is realized. The mapping rules are as follows:
x(i,j)＝x(n)|n＝(i-1)nl+j j≤nl(1)
step three: two-dimensional matrix spatial graying
And converting the matrix into an image by a graying method, and mapping the original two-dimensional matrix value of the power quality signal into a space gray value. The mapping rules are as follows:
Figure BDA0002295090920000031
step four: extracting characteristic value of two-dimensional harmonic gray image
And extracting time domain characteristic values of the waveform from each sampling period, and establishing a characteristic matrix according to the statistical characteristics. The eigenvalue matrix can be represented as:
Figure BDA0002295090920000032
the selected time domain characteristic value indexes are as follows:
TABLE 1 time-domain eigenvalue index
Figure BDA0002295090920000041
Respectively calculating time domain characteristic value T in each harmonic signal period1～T6And normalized.
Step five: clustering waveform data by using an FSOM neural network, and clustering all samples into k layers, wherein a data set P can be expressed as: p ═ P1∪P2∪...∪Pk,
Figure BDA0002295090920000042
i≠j。
1) Carrying out harmonic missing data set clustering parameter initialization, comprising the following steps: number of nodes in output layer, fuzzy index dtThe number of initialization iterations T is 1, and the maximum number of iterations TmaxInitial neuron node weight vt。
2) Carrying out harmonic missing data set clustering, and constructing an objective function by taking the weighted square sum of the distances from each sample to all clustering centers as a target:
Figure BDA0002295090920000043
3) constructing a membership matrix U ═ Uij]To express the relation between each electric energy quality cycle wave data and the cluster waveform data, wherein the membership degree matrix needs to satisfy
Figure BDA0002295090920000044
4) Determining the optimal clustering number of the power grid harmonic data, comprising the following steps:
(1) coefficient for calculating aggregation degree of power grid harmonic data set
Figure BDA0002295090920000045
And coefficients characterizing the degree of dispersion from class to class
Figure BDA0002295090920000046
(2) Meterα is the index lambda of the integral clustering effect of the harmonic wave data set of the calculation power gridmax+(1-βmin)。
(3) When | | lambda (k) -lambda (k-1) | < epsilon, judging that the total times of clustering iteration before convergence is the optimal classification number k of clustering; otherwise, return to (1).
5) Optimizing the objective function using the Lagrange multiplier method to
Figure BDA0002295090920000047
For constraint, the updating of the parameters in the objective function comprises the following steps:
(1) updating membership matrix
Figure BDA0002295090920000048
And fuzzy index
Figure BDA0002295090920000049
(2) Updating learning efficiency
Figure BDA00022950909200000410
(3) Updating neuron node weights
Figure BDA00022950909200000411
(4) Updating weight vectors
Figure BDA0002295090920000051
If | | Δ w | | non-calculation2＝||w(t+1)-w(t)||2>When epsilon is larger, return to (1); otherwise, the loop is ended.
Step six: completing the repair of the two-dimensional gray scale image, comprising:
(1) traversing all data, searching the position sequence x (i, j) and the hierarchical sequence q of the missing pointr(r is the number of cluster levels).
(2) Searching the same missing value in all layers to find out the maximum value S of the number of the information available around the missing valuemaxPosition information x ' (i, j) and location layer information q ' of the missing point are extracted 'r。
(3) The missing point with the largest amount of information available around is first in the layer according to equation (3)
Figure BDA0002295090920000052
And (5) repairing.
(4) Delete the location information x ' (i, j) and the current layer information q ' of the repaired point 'rIf the data is missing, the step (2) is entered for searching again; otherwise, entering the step (5).
(5) Pressing each layer of repaired data again (4)
Figure BDA0002295090920000053
Fusion is performed.
In order to verify the effectiveness of the electric energy quality measurement data loss repair method based on the fuzzy self-organizing neural network, the method is applied to the original harmonic loss measurement data for data repair effect analysis.
And clustering the power grid harmonic data sets according to the characteristic values of the waveforms by using the FSOM neural network mapping method shown in FIG. 2. The algorithm flow chart is shown in fig. 4.
The original data set was established using the power quality standard signal and the disturbance signal model in table 2 below. Experiments prove that the original signal-to-noise ratio after Gaussian white noise is mixed is 17db by repairing the loss of abnormal power quality data containing voltage temporary rise and temporary drop. The second experiment data is harmonic voltage loss. The maximum sampling frequency is 20kHZ, and each group of experimental data loss modes are divided into two modes, namely continuous loss and random loss.
Meter 2 electric energy quality standard signal and disturbance signal model
Figure BDA0002295090920000054
The invention selects a plurality of evaluation indexes to evaluate the quality of the data restoration effect, and comprises the following steps: mean absolute error (MAD), which can avoid the problem of mutual error cancellation and better reflect the actual situation of error repair; signal-to-noise ratio (SNR), which reflects the repair accuracy of a noisy signal waveform; root Mean Square Error (RMSE), which reflects the degree of dispersion of the repair results. Calculating the formula:
Figure BDA0002295090920000055
Figure BDA0002295090920000056
Figure BDA0002295090920000061
by using the electric energy quality measurement data missing repair method based on the fuzzy self-organizing neural network, the missing measurement data are repaired, and the voltage temporary rising and temporary falling missing measurement repair effect pair is shown in fig. 4 and the voltage harmonic missing measurement data repair effect pair is shown in fig. 5.
The abscissa in fig. 4 and 5 is the data missing ratio. Under the condition of 30% data loss rate of a continuous deletion mode, compared with a MARS algorithm, the average absolute error of the algorithm provided by the invention is reduced by 60.71%; the signal to noise ratio is improved by 47.3 percent; the root mean square error is reduced by 56.76%. Under the condition of random deletion or continuous deletion, the FSOM restoration algorithm provided by the invention has lower restoration error and higher signal-to-noise ratio than a time dynamic matrix decomposition method, a multivariate self-adaptive regression spline method and a KNN restoration method.

Claims (6)

1. A method for repairing missing electric energy quality measurement data of a fuzzy self-organizing neural network is implemented by a computer program and comprises the following steps: the method comprises the following steps:
1) inputting a missing electric energy quality one-dimensional measurement data set;
2) segmenting and intercepting the sampled one-dimensional waveform data according to N power quality sampling periods, and mapping rules x (i, j) ═ x (N) | N ═ i-1) N of a two-dimensional matrix xl+j j≤nlMapping it to a row in a two-dimensional matrix xOr, performing two-dimensional truncation recombination on the one-dimensional signals;
3) calculating a space gray value P (i, j) [ (255- (254 × (m-x (i, j)/(m-n)) ], and converting the matrix x into an image by a graying method;
4) extracting characteristic value X of two-dimensional harmonic gray imagej＝[X1,j,X2,j,…,Xm+l,j]And carrying out normalization treatment;
5) determining the optimal clustering number of the power grid harmonic data;
6) and (3) constructing an objective function by taking the weighted square sum of the distances from each sample to all the cluster centers as an objective:
Figure FDA0002295090910000011
and constructing a membership matrix U ═ Uij]To do so by
Figure FDA0002295090910000012
Training an optimal clustering mode for constraint conditions;
7) completing the repair of the two-dimensional gray scale image;
8) and fusing the repaired data of each layer, and restoring the image information into a waveform signal.
2. The method of claim 1, wherein the method comprises the steps of: the step 5) of determining the optimal clustering number of the power grid harmonic data comprises the following steps:
(1) calculating coefficients of aggregation level
Figure FDA0002295090910000013
And coefficients characterizing the degree of dispersion from class to class
Figure FDA0002295090910000014
(2) α is the index lambda for calculating the overall clustering effectmax+(1-βmin)；
(3) When | | lambda (k) -lambda (k-1) | < epsilon, judging that the total times of clustering iteration before convergence is the optimal classification number k of clustering; otherwise, return to (1).
3. The method of claim 1, wherein the method comprises the steps of: the updating of the parameters in the objective function in the step 6) comprises the following steps:
(1) updating degree of membership
Figure FDA0002295090910000015
And fuzzy index
Figure FDA0002295090910000016
(2) Updating learning efficiency
Figure FDA0002295090910000017
(3) Updating neuron node weights
Figure FDA0002295090910000018
(4) Updating weight vectors
Figure FDA0002295090910000019
If | | Δ w | | non-calculation2＝||w(t+1)-w(t)||2>When epsilon is larger, return to (1); otherwise, the loop is ended.
4. The method of claim 1, wherein the method comprises the steps of: the step 7) of completing the repair of the two-dimensional gray scale image comprises the following steps:
(1) traversing all data, searching and recording the position sequence x (i, j) and the hierarchical sequence q of each missing pointr
(2) Searching the same missing value in all layers, finding out the maximum value of the number of the information available around the missing value, and extracting the position information of the missing point and the information of the layer where the missing point is located;
(3) the missing point with the largest number of available information around is repaired firstly in the layer;
(4) deleting the position information and the layer information of the repaired point, and if missing data exist, entering the step (2) to search again; otherwise go to 8).
5. The method for repairing the missing data of the electric energy quality measurement data based on the fuzzy self-organizing neural network as claimed in claim 1, wherein in the step 7) of repairing the two-dimensional gray scale map, if the corresponding data is not missing, the data is not processed.
6. The method for repairing the missing of the electric energy quality measurement data based on the fuzzy self-organizing neural network as claimed in claim 1, wherein the final clustering number k determined in the step 5) is substituted into the step 6) to perform similarity clustering.
CN201911197788.9A 2019-11-29 2019-11-29 Power quality measurement missing repair method for fuzzy self-organizing neural network Active CN111008584B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911197788.9A CN111008584B (en) 2019-11-29 2019-11-29 Power quality measurement missing repair method for fuzzy self-organizing neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911197788.9A CN111008584B (en) 2019-11-29 2019-11-29 Power quality measurement missing repair method for fuzzy self-organizing neural network

Publications (2)

Publication Number Publication Date
CN111008584A true CN111008584A (en) 2020-04-14
CN111008584B CN111008584B (en) 2023-09-08

Family

ID=70112527

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911197788.9A Active CN111008584B (en) 2019-11-29 2019-11-29 Power quality measurement missing repair method for fuzzy self-organizing neural network

Country Status (1)

Country Link
CN (1) CN111008584B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111597925A (en) * 2020-04-29 2020-08-28 山东卓文信息科技有限公司 Power system transient signal analysis method based on DCNN

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110026850A1 (en) * 2009-07-30 2011-02-03 Marcelo Weinberger Context-cluster-level control of filtering iterations in an iterative discrete universal denoiser
CN107291765A (en) * 2016-04-05 2017-10-24 南京航空航天大学 The clustering method of processing missing data is planned based on DC
CN109584260A (en) * 2018-11-27 2019-04-05 烟台中科网络技术研究所 A kind of liver imaging dividing method and system
WO2019218263A1 (en) * 2018-05-16 2019-11-21 深圳大学 Extreme learning machine-based extreme ts fuzzy inference method and system

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110026850A1 (en) * 2009-07-30 2011-02-03 Marcelo Weinberger Context-cluster-level control of filtering iterations in an iterative discrete universal denoiser
CN107291765A (en) * 2016-04-05 2017-10-24 南京航空航天大学 The clustering method of processing missing data is planned based on DC
WO2019218263A1 (en) * 2018-05-16 2019-11-21 深圳大学 Extreme learning machine-based extreme ts fuzzy inference method and system
CN109584260A (en) * 2018-11-27 2019-04-05 烟台中科网络技术研究所 A kind of liver imaging dividing method and system

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
崔迎宾;张卫欣;单增礼;李祯祥;王林林;杨挺;: "低压电力线载波通信的系统化测试方法" *
陈阳: "拓扑梯度耦合FCMC的全自动图像修复优化算法" *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111597925A (en) * 2020-04-29 2020-08-28 山东卓文信息科技有限公司 Power system transient signal analysis method based on DCNN

Also Published As

Publication number Publication date
CN111008584B (en) 2023-09-08

Similar Documents

Publication Publication Date Title
CN112327219B (en) Distribution transformer fault diagnosis method with automatic feature mining and parameter automatic optimizing functions
CN106055918B (en) Method for identifying and correcting load data of power system
CN110929847A (en) Converter transformer fault diagnosis method based on deep convolutional neural network
CN110212528B (en) Power distribution network measurement data missing reconstruction method
CN110443724B (en) Electric power system rapid state estimation method based on deep learning
CN106656357B (en) Power frequency communication channel state evaluation system and method
CN110087207B (en) Method for reconstructing missing data of wireless sensor network
CN106912067B (en) WSN wireless communication module fault diagnosis method based on fuzzy neural network
CN111397901A (en) Rolling bearing fault diagnosis method based on wavelet and improved PSO-RBF neural network
CN113688869B (en) Photovoltaic data missing reconstruction method based on generation countermeasure network
CN114266301A (en) Intelligent power equipment fault prediction method based on graph convolution neural network
CN110458189A (en) Compressed sensing and depth convolutional neural networks Power Quality Disturbance Classification Method
CN115983374A (en) Cable partial discharge database sample expansion method based on optimized SA-CACGAN
CN113203914A (en) Underground cable early fault detection and identification method based on DAE-CNN
CN115575125A (en) Bearing fault diagnosis method based on GADF-GAN-AVOA-CNN
CN111008584B (en) Power quality measurement missing repair method for fuzzy self-organizing neural network
CN113379116A (en) Cluster and convolutional neural network-based line loss prediction method for transformer area
CN116819423A (en) Method and system for detecting abnormal running state of gateway electric energy metering device
CN115081551A (en) RVM line loss model building method and system based on K-Means clustering and optimization
CN112766537B (en) Short-term electric load prediction method
CN114881176A (en) Non-invasive load identification method based on self-adaptive optimization random forest
Jin et al. Deep Machine Learning-Based Asset Management Approach for Oil-Immersed Power Transformers Using Dissolved Gas Analysis
CN114630207A (en) Multi-sensing-node perception data collection method based on noise reduction self-encoder
CN114048819A (en) Power distribution network topology identification method based on attention mechanism and convolutional neural network
CN113536508A (en) Method and system for classifying manufacturing network nodes

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant