CN111459697A - Excitation system fault monitoring method based on deep learning network - Google Patents

Excitation system fault monitoring method based on deep learning network Download PDF

Info

Publication number
CN111459697A
CN111459697A CN202010228034.1A CN202010228034A CN111459697A CN 111459697 A CN111459697 A CN 111459697A CN 202010228034 A CN202010228034 A CN 202010228034A CN 111459697 A CN111459697 A CN 111459697A
Authority
CN
China
Prior art keywords
excitation system
network
fault
excitation
data
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
Application number
CN202010228034.1A
Other languages
Chinese (zh)
Inventor
马宏忠
钱昆
王婧佳
张海强
司宇
彭晓晗
沈梦洁
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.)
Hohai University HHU
Original Assignee
Hohai University HHU
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 Hohai University HHU filed Critical Hohai University HHU
Priority to CN202010228034.1A priority Critical patent/CN111459697A/en
Publication of CN111459697A publication Critical patent/CN111459697A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00—Error detection; Error correction; Monitoring
    • G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
    • G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
    • G06F11/079—Root cause analysis, i.e. error or fault diagnosis
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Quality & Reliability (AREA)
  • Testing And Monitoring For Control Systems (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses an excitation system fault monitoring method based on a deep learning network, which comprises the steps of collecting historical data of active power, reactive power, excitation voltage and current in the operation process of an excitation system, forming a convolutional neural network training set and a test set after standardized processing, inputting the training set data into the constructed convolutional neural network, calculating the minimum value of a loss function of the network by a random gradient descent method, updating the weight of the network, testing the convolutional neural network by the test set until the required accuracy is reached, and finally obtaining whether the excitation system of a power plant is stably operated or not by the excitation system fault monitoring network based on the deep learning network. According to the invention, the deep learning convolutional neural network is constructed to monitor the power plant excitation system according to the multi-source historical data in the operation process of the excitation system, the method is simple, the result is accurate and reliable, and data support and theoretical basis are provided for the operation maintenance and on-line monitoring of the power plant excitation system.

Description

Excitation system fault monitoring method based on deep learning network
Technical Field
The invention belongs to the technical field of excitation system fault diagnosis, and particularly relates to an excitation system fault monitoring method based on a deep learning network.
Background
The excitation system is an important component of the synchronous generator, and the good excitation system can effectively improve the stability limit level and the operation technical and economic indexes of the power system and the generator. The excitation system plays an important role in maintaining the voltage of the generator terminal, reasonably distributing reactive power, guaranteeing the safe operation of power equipment and improving the stability of power. However, most of the existing excitation system equipment detection means are to manually extract fault characteristic information (automatic extraction is used as an auxiliary), and to perform fault analysis and positioning under the guidance of manual experience, so that the existing excitation system equipment detection means have fatal defects of offline diagnosis, high requirement on maintenance personnel and the like, and the real-time performance and accuracy of fault diagnosis are seriously influenced. Therefore, there is a need to develop a method for automatically examining the operation state and health condition of the generator excitation system.
Disclosure of Invention
Aiming at the problems, the invention provides an excitation system fault monitoring method based on a deep learning network, which is used for realizing the fault monitoring of the excitation system of a power plant and can accurately and quickly judge the fault type.
The technical scheme adopted by the invention is as follows:
a fault monitoring method for an excitation system based on a deep learning network comprises the following steps:
acquiring original sampling data of an excitation system in different fault operation processes;
carrying out standardization processing on original sampling data to form a convolutional neural network training set and a test set;
constructing a convolutional neural network;
inputting a convolutional neural network training set into the constructed convolutional neural network for network training;
testing the trained convolutional neural network by adopting a convolutional neural network test set until the required accuracy is reached to obtain an excitation system fault monitoring network based on a deep learning network;
and inputting the real-time operation data of the excitation system into an excitation system fault monitoring network based on a deep learning network, and monitoring the excitation system fault in real time.
Further, the acquiring of the original sampling data in the different fault operation processes of the excitation system includes:
active power, reactive power, excitation voltage and excitation current historical data in different fault operation processes of an excitation system are sampled at equal intervals through an auxiliary power monitoring system, and an original sampling data set is obtained:
Figure BDA0002428365350000021
wherein S isiA raw sample data set representing a type i fault,
Figure BDA0002428365350000022
and
Figure BDA0002428365350000023
respectively representing the jth active power and reactive power collected by the ith type of fault of the excitation system,
Figure BDA0002428365350000024
and
Figure BDA0002428365350000025
j is 1,2,3, … n, and n is the sample size.
Further, the different faults include: the excitation system normally operates, the power module fails, the excitation module fails, the regulation module fails and the de-excitation module fails.
Further, the normalizing the raw sampling data includes:
Figure BDA0002428365350000026
wherein the content of the first and second substances,
Figure BDA0002428365350000027
the average value and the variance of sample data R of the ith fault are obtained, and the sample data R represents active power, reactive power, excitation voltage and excitation current data of a group of excitation systems; riIs a normalized vector containing n normalized sample data.
Further, the constructing the convolutional neural network includes:
the input layer is used for sending the sample data after the standardization processing to the next layer of operation;
the convolution layer is used for performing convolution operation on input sample data and performing feature extraction:
Figure BDA0002428365350000028
wherein k istDenotes the t-th convolution kernel, b denotes the offset, ytRepresents TiThe t-th feature map after convolution operation, s represents the number of convolution kernels, MjRepresenting a fault class, f (-) is an excitation function;
the excitation layer is used to introduce nonlinear elements:
Figure BDA0002428365350000034
wherein x is an input amount,
Figure BDA0002428365350000031
gaussian noise with mean 0 and variance σ;
the pooling layer output is calculated as follows:
xl＝f(down(xl-1)+b)
wherein x isl-1For data transmitted from convolutional layers, xlFor the data after pooling sampling, down (-) is a sampling function;
the full connection is used for integrating the features extracted by the convolutional layers, sending the features into a softmax classifier, calculating the score values of input data in different fault classes, calculating the probability of the input data under each fault class, and outputting the fault class with the maximum fault probability:
Figure BDA0002428365350000032
wherein softmax(s)i) Indicating the i-th class failure probability, siThe score value of the model on the ith fault is represented.
Further, the inputting the convolutional neural network training set into the constructed convolutional neural network for network training includes:
calculating the minimum value of the network loss function by adopting a random gradient descent method, and updating the weight;
the network loss function is calculated as follows:
Figure BDA0002428365350000033
where m is the input sample size, WkIs the regularization weight of the k layer, y' represents the fault class actually corresponding to the input data, xiThe failure types identified by the network are shown, a represents a regularization coefficient, and L is the sum of the weights of all layers in the network.
Further, the method also comprises the following steps: after the convolutional neural network is established, random numbers with the mean value of 0 and the variance of 1 are generated, and parameters of the convolutional neural network are initialized.
Further, the accuracy is calculated as: and counting and identifying the correct number of samples, and dividing the number by the total number of samples to obtain the accuracy.
The invention has the beneficial effects that:
(1) the method provided by the invention realizes fault monitoring of the power plant excitation system by using the deep learning network, can accurately and rapidly judge the fault type, has strong mobility, and is suitable for practical application in many industrial fields.
(2) The method has an automatic learning function, records new faults of the system in real time operation, and improves the robustness and reliability of the fault diagnosis model.
(3) The method directly extracts the data characteristics so as to identify the fault type of the excitation system without abundant prior knowledge.
Drawings
Fig. 1 is a flow chart of an excitation system fault monitoring method based on a deep learning network.
FIG. 2 is a diagram of a deep learning convolutional neural network structure constructed in the present invention;
FIG. 3 is a schematic diagram of a deep learning convolutional neural network loss function training process in an embodiment of the present invention;
FIG. 4 is a diagram illustrating the classification result of the test set according to the embodiment of the present invention.
Detailed Description
The invention is further described below. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
The invention provides a fault monitoring method for an excitation system based on a deep learning network, which comprises the following steps:
step 1: acquiring active power, reactive power, exciting voltage, current and other historical data of an exciting system in different fault operation processes through an ECMS (electric control mechanical system) of a power plant;
step 2: the collected historical data are subjected to processing such as screening and standardization, a neural network training set is formed, a deep learning convolution neural network is built, and parameters are initialized;
and step 3: inputting the training set into a convolutional neural network, calculating the minimum value of a loss function of the convolutional neural network by a random gradient descent method, and updating the weight of the convolutional neural network;
and 4, step 4: reading the operating data of the excitation system through the ECMS of the power plant, inputting the operating data into the trained convolutional neural network, repeating the process of the step 3, and testing the trained convolutional neural network until the required accuracy is reached;
and 5: and (3) adopting a tested convolutional neural network to monitor whether the power plant excitation system operates stably on line.
As a preferred embodiment, a specific implementation procedure of the excitation system fault monitoring method based on the deep learning network of the present invention, referring to fig. 1, includes:
step 1, reading historical data of an excitation system in 5 fault types in an auxiliary power monitoring system (ECMS) of a 2 × 350MW thermal power plant in the operation process, wherein the historical data comprises active power, reactive power, excitation voltage and current to form an original data set, and the fault types comprise normal operation of the excitation system, fault of a power module, fault of an excitation initiating module, fault of a regulating module and fault of a field extinguishing module.
Further, an ECMS system is used for sampling each source signal of the excitation system at equal intervals to form an original data set:
Figure BDA0002428365350000051
each type of data contained in the original data set has n consecutive sample points,
Figure BDA0002428365350000052
and
Figure BDA0002428365350000053
respectively representing jth active power and reactive power collected by ith type of fault of the excitation system,
Figure BDA0002428365350000054
and
Figure BDA0002428365350000055
the j-th excitation voltage and current respectively represent the collection of the ith fault of the excitation system, i represents the ith fault, j is 1,2,3, … n, and i is 1,2,3,4 and 5.
Furthermore, the ECMS system station control layer adopts an IEC61850 standard dual Ethernet redundant structure, is provided with a database server, an electric operator station, an electric engineer station, a printer and a forwarding station for communicating with other systems, and is a monitoring and management center of the whole ECMS system.
Step 2: and respectively standardizing each data in the collected original data set to enable the mean value to be 0 and the variance to be 1, and forming a convolutional neural network training set.
Active power for first-class fault of 2 × 350MW thermal power plant excitation system
Figure BDA0002428365350000056
Normalization is as in equation (1), and other kinds of sample data are normalized by a similar method:
Figure BDA0002428365350000057
in the formula (I), the compound is shown in the specification,
Figure BDA0002428365350000058
the mean value and the variance of the active power of the first type of fault are obtained; p1Is a normalized vector comprising n normalized data.
After data standardization, a deep learning convolutional neural network is built, wherein the convolutional neural network is mainly divided into an input layer, a convolutional layer, an excitation layer, a pooling layer and a full-connection part, and the specific structure is shown in FIG. 2.
Network input layer: data set T of all fault classes after being standardizedi＝(Pi,Qi,Ui,Ii) And sending the next layer of operation.
And a convolution operation layer: the convolution layer performs convolution operation on input data according to the formula (2), and then performs corresponding feature extraction:
Figure BDA0002428365350000061
in the formula, ktDenotes the t-th convolution kernel, b denotes the offset, ytRepresents TiThe t-th feature map after convolution operation, s represents the number of convolution kernels, MjIndicating the fault class, f (-) is the stimulus function.
And (3) adopting an excitation function formula (3), correcting a linear unit function (Re L u), and introducing a nonlinear element into the whole network:
Figure BDA0002428365350000062
wherein, x is the input quantity,
Figure BDA0002428365350000063
gaussian noise with mean 0 and variance σ.
A pooling layer: if the first layer is a pooling layer and the first-1 layer is a convolutional layer, the formula for the first layer is as follows (4):
xl＝f(down(xl-1)+b) (4)
where down (-) is the sampling function, xl-1For data transmitted from convolutional layers, xlThe data after pooling sampling.
Fully connecting: the method mainly comprises a full connection layer and a softmax classifier. Each node of the fully connected layer is connected with all nodes of the previous layer and is used for integrating the extracted features. Sending the data of the full connection layer into a softmax classifier to obtain the score values of the data of 5 types of faults, and calculating the probability of the data under each type of fault through a formula (5), wherein the maximum probability is the fault to which the data belongs:
Figure BDA0002428365350000064
in the formula, softmax(s)i) Indicating the i-th class failure probability, siThe score value of the model on the ith fault is represented.
After the deep learning convolutional neural network of the corresponding structure is established, random numbers with the mean value of 0 and the variance of 1 are generated, and parameters of the convolutional neural network are initialized.
And step 3: inputting the data set subjected to standardization in the step 2 into a deep learning convolutional neural network as a training set, calculating the minimum value of a loss function by a random gradient descent method, and updating the weight Wk。
The loss function of the network is calculated as follows:
Figure BDA0002428365350000071
where m is input sample data, a set of active, reactive, voltage and current are one sample data, and W iskIs the regularization weight of the k layer, y' represents the fault category actually corresponding to the input sample data, xiThe failure types identified by the network are shown, a represents a regularization coefficient, and the invention takes 0.0001 and L as the sum of the weight numbers of each layer in the network.
The training process of the loss function is shown in figure 3, the curve of the loss function image in the figure is smooth, and the curve is finally converged to be about 0, which shows that the model training process is normal and the overall quality is good.
And 4, sampling the operation data of an excitation system in an auxiliary power monitoring system (ECMS) of the 2 × 350MW thermal power plant for multiple times, wherein the sampling frequency is 12khz, the sample capacity is 6000, and normalizing the sampled data to obtain a test set.
Inputting the test set into the trained neural network, calculating the fault category corresponding to each test data, and setting α by calculating the accuracy α of the networkmin90 percent when α is more than or equal to αminWhen the network is constructed, the network construction is finished; otherwise, repeat step 3.
The accuracy α of the network is calculated by counting the number of samples identified correctly and dividing by the total number of samples to obtain the accuracy.
Through repeated experiments, the accuracy rate of the excitation system fault monitoring network based on the deep learning network is 92.47%, and the final test set is classified and shown in fig. 4. The dots on the line in the graph indicate that the network has identified correctly, and the dots outside the line indicate that the network has identified incorrectly.
And 5: real-time operation data of an excitation system in an auxiliary power monitoring system (ECMS) is input into an excitation system fault monitoring network based on a deep learning network, and excitation system faults are monitored in real time.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.

Claims (8)

1. A fault monitoring method for an excitation system based on a deep learning network is characterized by comprising the following steps:
acquiring original sampling data of an excitation system in different fault operation processes;
carrying out standardization processing on original sampling data to form a convolutional neural network training set and a test set;
constructing a convolutional neural network;
inputting a convolutional neural network training set into the constructed convolutional neural network for network training;
testing the trained convolutional neural network by adopting a convolutional neural network test set until the required accuracy is reached to obtain an excitation system fault monitoring network based on a deep learning network;
and inputting the real-time operation data of the excitation system into an excitation system fault monitoring network based on a deep learning network, and monitoring the excitation system fault in real time.
2. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, wherein the obtaining of raw sampling data in different fault operation processes of the excitation system comprises:
active power, reactive power, excitation voltage and excitation current historical data in different fault operation processes of an excitation system are sampled at equal intervals through an auxiliary power monitoring system, and an original sampling data set is obtained:
Figure FDA0002428365340000011
wherein S isiRaw sample data set, P, representing a type i faultj iAnd
Figure FDA0002428365340000012
respectively representing the jth active power and reactive power collected by the ith type of fault of the excitation system,
Figure FDA0002428365340000013
and
Figure FDA0002428365340000014
j is 1,2,3,.. n, and n represents the sample size.
3. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, wherein the different faults comprise: the excitation system normally operates, the power module fails, the excitation module fails, the regulation module fails and the de-excitation module fails.
4. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, wherein the normalizing process of the raw sampling data comprises:
Figure FDA0002428365340000021
wherein the content of the first and second substances,
Figure FDA0002428365340000022
the average value and the variance of sample data R of the ith fault are obtained, and the sample data R represents active power, reactive power, excitation voltage and excitation current data of a group of excitation systems; riIs a normalized vector containing n targetsNormalized sample data.
5. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, wherein the constructing of the convolutional neural network comprises:
the input layer is used for sending the sample data after the standardization processing to the next layer of operation;
the convolution layer is used for performing convolution operation on input sample data and performing feature extraction:
Figure FDA0002428365340000023
wherein k istDenotes the t-th convolution kernel, b denotes the offset, ytRepresents TiThe t-th feature map after convolution operation, s represents the number of convolution kernels, MjRepresenting a fault class, f (-) is an excitation function;
the excitation layer is used to introduce nonlinear elements:
Figure FDA0002428365340000024
wherein x is an input amount,
Figure FDA0002428365340000025
gaussian noise with mean 0 and variance σ;
the pooling layer output is calculated as follows:
xl＝f(down(xl-1)+b)
wherein x isl-1For data transmitted from convolutional layers, xlFor the data after pooling sampling, down (-) is a sampling function; the full connection is used for integrating the features extracted by the convolutional layers, sending the features into a softmax classifier, calculating the score values of input data in different fault classes, calculating the probability of the input data under each fault class, and outputting the fault class with the maximum fault probability:
Figure FDA0002428365340000026
wherein softmax(s)i) Indicating the i-th class failure probability, siThe score value of the model on the ith fault is represented.
6. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, wherein the inputting of the convolutional neural network training set to the constructed convolutional neural network for network training comprises:
calculating the minimum value of the network loss function by adopting a random gradient descent method, and updating the weight;
the network loss function is calculated as follows:
Figure FDA0002428365340000031
where m is the input sample size, WkIs the regularization weight of the k layer, y' represents the fault class actually corresponding to the input data, xiThe failure types identified by the network are shown, a represents a regularization coefficient, and L is the sum of the weights of all layers in the network.
7. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, further comprising: after the convolutional neural network is established, random numbers with the mean value of 0 and the variance of 1 are generated, and parameters of the convolutional neural network are initialized.
8. The excitation system fault monitoring method based on the deep learning network as claimed in claim 1, wherein the accuracy is calculated as: and counting and identifying the correct number of samples, and dividing the number by the total number of samples to obtain the accuracy.
CN202010228034.1A 2020-03-27 2020-03-27 Excitation system fault monitoring method based on deep learning network Pending CN111459697A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010228034.1A CN111459697A (en) 2020-03-27 2020-03-27 Excitation system fault monitoring method based on deep learning network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010228034.1A CN111459697A (en) 2020-03-27 2020-03-27 Excitation system fault monitoring method based on deep learning network

Publications (1)

Publication Number Publication Date
CN111459697A true CN111459697A (en) 2020-07-28

Family

ID=71685730

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010228034.1A Pending CN111459697A (en) 2020-03-27 2020-03-27 Excitation system fault monitoring method based on deep learning network

Country Status (1)

Country Link
CN (1) CN111459697A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115712064A (en) * 2022-11-07 2023-02-24 华能威海发电有限责任公司 Excitation system fault diagnosis method based on LSTM-CNN hybrid neural network
CN116306095A (en) * 2023-01-16 2023-06-23 中国电建集团北京勘测设计研究院有限公司 Pumped storage unit fault diagnosis system and method based on edge calculation

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103996077A (en) * 2014-05-22 2014-08-20 中国南方电网有限责任公司电网技术研究中心 Electric equipment fault forecasting method based on multi-dimension time sequence
CN107271925A (en) * 2017-06-26 2017-10-20 湘潭大学 The level converter Fault Locating Method of modularization five based on depth convolutional network
CN108896296A (en) * 2018-04-18 2018-11-27 北京信息科技大学 A kind of wind turbine gearbox method for diagnosing faults based on convolutional neural networks
US20190379589A1 (en) * 2018-06-12 2019-12-12 Ciena Corporation Pattern detection in time-series data

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103996077A (en) * 2014-05-22 2014-08-20 中国南方电网有限责任公司电网技术研究中心 Electric equipment fault forecasting method based on multi-dimension time sequence
CN107271925A (en) * 2017-06-26 2017-10-20 湘潭大学 The level converter Fault Locating Method of modularization five based on depth convolutional network
CN108896296A (en) * 2018-04-18 2018-11-27 北京信息科技大学 A kind of wind turbine gearbox method for diagnosing faults based on convolutional neural networks
US20190379589A1 (en) * 2018-06-12 2019-12-12 Ciena Corporation Pattern detection in time-series data

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
周于杰: "同步发电机励磁装置分级递阶故障诊断研究", 《中国硕士学位论文全文数据库 信息科技辑》 *
许庆勇: "《基于深度学习理论的纹身图像识别与检测研究》", 31 December 2018 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115712064A (en) * 2022-11-07 2023-02-24 华能威海发电有限责任公司 Excitation system fault diagnosis method based on LSTM-CNN hybrid neural network
CN115712064B (en) * 2022-11-07 2024-02-06 华能威海发电有限责任公司 Excitation system fault diagnosis method based on LSTM-CNN hybrid neural network
CN116306095A (en) * 2023-01-16 2023-06-23 中国电建集团北京勘测设计研究院有限公司 Pumped storage unit fault diagnosis system and method based on edge calculation

Similar Documents

Publication Publication Date Title
CN111507376B (en) Single-index anomaly detection method based on fusion of multiple non-supervision methods
CN111504676B (en) Equipment fault diagnosis method, device and system based on multi-source monitoring data fusion
CN111539550B (en) Method, device, equipment and storage medium for determining working state of photovoltaic array
US8868985B2 (en) Supervised fault learning using rule-generated samples for machine condition monitoring
CN104732276B (en) One kind metering production facility on-line fault diagnosis method
CN109344517A (en) A kind of high-voltage isulation method for diagnosing faults of new-energy automobile
CN106021771A (en) Method and device for diagnosing faults
CN107909118A (en) A kind of power distribution network operating mode recording sorting technique based on deep neural network
CN108304567B (en) Method and system for identifying working condition mode and classifying data of high-voltage transformer
CN111580506A (en) Industrial process fault diagnosis method based on information fusion
CN112464995A (en) Power grid distribution transformer fault diagnosis method and system based on decision tree algorithm
CN111459697A (en) Excitation system fault monitoring method based on deep learning network
CN108287327A (en) Metering automation terminal fault diagnostic method based on Bayes's classification
CN117074839B (en) Electromagnetic valve fault diagnosis method and system
CN108709744A (en) Motor bearings method for diagnosing faults under a kind of varying load operating mode
CN113391239A (en) Transformer abnormality monitoring method and system based on edge calculation
CN110580492A (en) Track circuit fault precursor discovery method based on small fluctuation detection
CN115954879A (en) Power distribution network line variable relation accurate identification method based on AO algorithm optimization PNN
CN114814501A (en) On-line diagnosis method for capacitor breakdown fault of capacitor voltage transformer
CN113010394B (en) Machine room fault detection method for data center
CN111190072A (en) Centralized meter reading system diagnosis model establishing method, fault diagnosis method and fault diagnosis device
CN117347796A (en) Intelligent gateway-based switching equipment partial discharge diagnosis system and method
CN109784777B (en) Power grid equipment state evaluation method based on time sequence information fragment cloud similarity measurement
CN116578922A (en) Valve cooling system fault diagnosis method and device based on multichannel convolutional neural network
CN116338545A (en) Method, system, equipment and medium for identifying metering error state of current transformer

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200728