CN111553112A - Power system fault identification method and device based on deep belief network - Google Patents
Power system fault identification method and device based on deep belief network Download PDFInfo
- Publication number
- CN111553112A CN111553112A CN202010183621.3A CN202010183621A CN111553112A CN 111553112 A CN111553112 A CN 111553112A CN 202010183621 A CN202010183621 A CN 202010183621A CN 111553112 A CN111553112 A CN 111553112A
- Authority
- CN
- China
- Prior art keywords
- fault
- power system
- data
- training
- characteristic 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000012549 training Methods 0.000 claims abstract description 67
- 238000010606 normalization Methods 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims abstract description 17
- 238000002372 labelling Methods 0.000 claims abstract description 13
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 230000001133 acceleration Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000013075 data extraction Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 8
- 238000013473 artificial intelligence Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000012360 testing method Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000003745 diagnosis Methods 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000002847 impedance measurement Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/045—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
- G06N3/088—Non-supervised learning, e.g. competitive learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a power system fault identification method based on a deep belief network, which comprises the following steps: acquiring a training data set in power system time domain simulation software, and carrying out fault information labeling on the training data set; extracting characteristic data representing the state of the power system based on the obtained data in the training data set, and performing normalization processing on the characteristic data; inputting the extracted feature data and the sample fault labeling information into a deep confidence network to be trained, and training the deep confidence network to be trained to obtain a trained deep confidence network; and when a fault occurs, collecting the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault. In the implementation of the invention, the method and the device can realize the identification of the fault area and the fault position of the power system.
Description
Technical Field
The invention relates to the technical field of operation and control of an electric power system, in particular to a method and a device for identifying faults of the electric power system based on a deep belief network.
Background
In an electric power system, accurate fault detection and identification are the basis for safe and stable operation of a smart grid. With the increasing complexity of power system topologies and operations, fast and accurate grid fault identification has not been well achieved.
At present, fault identification methods can be roughly classified into three types based on impedance, traveling wave technology and artificial intelligence application. In the method based on impedance measurement, the method is divided into a vector domain and a time domain. In the method using the vector field, the purpose of fault identification is to perform vector calculation by current and voltage data. And the time domain method is to establish a differential equation according to the transmission line to solve. This method depends on the transmission line characteristics and is susceptible to fault resistance values. The fault identification method based on the traveling wave theory is to analyze the fault by analyzing a position time chart of current or voltage waveform and utilizing methods such as wavelet transformation, FFT and the like. Such methods are not affected by system topology and fault resistance, but generally require data with a higher sampling frequency.
In recent years, power system fault detection and diagnosis research based on artificial intelligence has been developed primarily, neural networks, support vector machines and random forest methods are common methods in fault diagnosis, and the traditional artificial intelligence methods have high fault identification correctness in some scenes, but cannot be widely applied because the traditional artificial intelligence methods have simple structures and cannot better extract power grid characteristic quantities. Compared with the traditional artificial intelligence method, the deep learning method can mine deep complex association relation in data, so that the algorithm precision is improved.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, and provides a power system fault identification method based on a deep confidence network, which can realize the identification of a power system fault area and a fault position.
In order to solve the above technical problem, an embodiment of the present invention provides a method for identifying a fault of an electrical power system based on a deep belief network, where the method includes:
acquiring a training data set in power system time domain simulation software, and carrying out fault information labeling on the training data set;
extracting characteristic data representing the state of the power system based on the obtained data in the training data set, and performing normalization processing on the characteristic data;
inputting the extracted feature data and the sample fault labeling information into a deep confidence network to be trained, and training the deep confidence network to be trained to obtain a trained deep confidence network;
and when a fault occurs, collecting the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault.
Optionally, the obtaining a training data set in the power system time domain simulation software includes:
in the time domain simulation software of the power system, fault data are obtained by setting different fault positions, fault types and fault duration;
acquiring historical fault data of the power system in power system time domain simulation software;
and combining the fault data with the historical fault data of the power system to obtain a training data set.
Optionally, the extracting the feature data representing the power system state includes:
the method comprises the following steps of (1) characteristic data of a system active load level at the initial moment of a system, characteristic data of a system reactive load level at the initial moment of the system, characteristic data of a system active output level at the initial moment of the system and characteristic data of a system reactive output level at the initial moment of the system;
the characteristic data of the maximum rotation angle difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum angular speed difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum kinetic energy difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum angular speed of the generator in the system at the fault occurrence moment and the characteristic data of the maximum acceleration power of the generator in the system at the fault occurrence moment;
the characteristic data of the maximum rotation angle difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum angular speed difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum kinetic energy difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum angular speed of the generator in the system at the moment of fault removal and the characteristic data of the maximum acceleration power of the generator in the system at the moment of fault removal.
Optionally, the specific calculation formula for performing normalization processing on the feature data is as follows:
wherein x' represents the data after normalization processing; x represents data before normalization processing; x is the number ofmaxMaximum value, x, representing characteristic dataminRepresenting the minimum value of the characteristic data.
Optionally, the training the deep belief network to be trained includes:
and obtaining the connection weight and the bias coefficient of each layer in the deep belief network after unsupervised pre-training and supervised fine tuning.
Optionally, the deep confidence network includes a plurality of layers of unsupervised limited boltzmann machines and a layer of supervised feedforward neural network.
Optionally, the multilayer unsupervised restricted boltzmann machine is used for extracting feature information in data and pre-training parameters;
the layer of supervised feedforward neural network is used for global parameter optimization and sample classification and parameter fine adjustment.
Optionally, the collecting characteristic quantities of the power system includes: the characteristic quantity at the initial time of the power system, the characteristic quantity at the fault occurrence time, and the characteristic quantity after the fault is removed.
In addition, the embodiment of the invention also provides a power system fault identification device based on the deep confidence network, and the device comprises:
a dataset acquisition and labeling module: the system comprises a training data set and a fault information marking module, wherein the training data set is used for obtaining a training data set in power system time domain simulation software and marking fault information on the training data set;
the characteristic data extraction and normalization processing module: the characteristic data is used for extracting characteristic data representing the state of the power system based on the data in the obtained training data set and carrying out normalization processing on the characteristic data;
the deep confidence network training module: the deep belief network training system is used for inputting the extracted feature data and the sample fault labeling information into a deep belief network to be trained, and training the deep belief network to be trained to obtain a trained deep belief network;
a fault area identification module: and after the fault occurs, acquiring the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault.
Optionally, the deep belief network training module further includes: the method is used for obtaining the connection weight and the bias coefficient of each layer in the deep belief network after unsupervised pre-training and supervised fine tuning.
In the implementation of the invention, the characteristic values of the state of the power system can be represented by collecting the characteristic values before the fault, at the moment of the fault occurrence and at the moment of the fault removal of the power system, the characteristic values and the fault marking information are input into the deep confidence network so as to train the deep confidence network, the new characteristic values collected by the fault are input into the trained deep confidence network, the identification of the fault area and the fault position of the power system can be realized, and the accuracy of fault identification is improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic flow chart of a method for identifying faults of a power system based on a deep belief network in the implementation of the present invention;
FIG. 2 is a schematic structural diagram of a deep belief network based power system fault identification apparatus in an implementation of the present invention;
FIG. 3 is a schematic diagram of a deep belief network model architecture in the practice of the present invention;
FIG. 4 is a schematic diagram of a restricted Boltzmann machine in a deep belief network model in the practice of the present invention;
fig. 5 is a schematic diagram of an IEEE39 system in the practice of the invention.
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.
Examples
Referring to fig. 1, fig. 1 is a schematic flow chart illustrating a method for identifying a fault of a power system based on a deep belief network according to an embodiment of the present invention.
As shown in fig. 1, a method for identifying a fault of a power system based on a deep belief network includes:
s11: acquiring a training data set in power system time domain simulation software, and carrying out fault information labeling on the training data set;
in a specific implementation process of the present invention, the obtaining a training data set in the power system time domain simulation software includes: in the time domain simulation software of the power system, fault data are obtained by setting different fault positions, fault types and fault duration; acquiring historical fault data of the power system in power system time domain simulation software; and combining the fault data with the historical fault data of the power system to obtain a training data set.
S12: extracting characteristic data representing the state of the power system based on the obtained data in the training data set, and performing normalization processing on the characteristic data;
in a specific implementation process of the present invention, the extracting feature data representing a state of the power system includes: the method comprises the following steps of (1) characteristic data of a system active load level at the initial moment of a system, characteristic data of a system reactive load level at the initial moment of the system, characteristic data of a system active output level at the initial moment of the system and characteristic data of a system reactive output level at the initial moment of the system; the characteristic data of the maximum rotation angle difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum angular speed difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum kinetic energy difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum angular speed of the generator in the system at the fault occurrence moment and the characteristic data of the maximum acceleration power of the generator in the system at the fault occurrence moment; the characteristic data of the maximum rotation angle difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum angular speed difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum kinetic energy difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum angular speed of the generator in the system at the moment of fault removal and the characteristic data of the maximum acceleration power of the generator in the system at the moment of fault removal.
Specifically, the feature data is normalized to be concentrated between [0,1], and a calculation formula is specifically as follows:
wherein x' representsNormalizing the processed data; x represents data before normalization processing; x is the number ofmaxMaximum value, x, representing characteristic dataminRepresenting the minimum value of the characteristic data.
S13: inputting the extracted feature data and the sample fault labeling information into a deep confidence network to be trained, and training the deep confidence network to be trained to obtain a trained deep confidence network;
in a specific implementation process of the present invention, the training of the deep belief network to be trained includes: and obtaining the connection weight and the bias coefficient of each layer in the deep belief network after unsupervised pre-training and supervised fine tuning.
Specifically, as shown in fig. 3, fig. 3 is a schematic diagram of a deep confidence network model in the implementation of the present invention, where the deep confidence network includes a plurality of layers of unsupervised limited boltzmann machines (RBMs) and a layer of supervised feedforward neural network (BP); it should be noted that the multilayer unsupervised restricted boltzmann machine is used for extracting feature information in data and pre-training parameters; the layer of supervised feedforward neural network is used for global parameter optimization and sample classification and parameter fine adjustment.
The RBM is a two-layer bidirectional recurrent neural network, which consists of a layer of visible units h representing input and a layer of hidden units v representing hidden variables, the basic structure of the RBM is shown in figure 4, and figure 4 is a structural schematic diagram of a restricted Boltzmann machine in a deep confidence network model in the implementation of the invention.
Meanwhile, the RBM is an energy-based model, and when the visible unit h and the hidden unit v are binary variables with states of 0 and 1, the energy function of the unit is:
in the formula, wijConnection weights for the ith visible cell and the jth hidden cell, vi、hjStates of visible cells i and hidden cells j, respectively, ci、bjAre respectively a visible sheetThe bias of element i and hidden element j;
meanwhile, the state probability of the RBM is determined by its energy function, which is:
wherein Z is a normalization factor, also called a partition function;
according to the structure and state probability of the RBM, when the state of each visible layer unit is given, the states of each unit of the hidden layer are independent; similarly, when the state of each hidden layer unit is given, the states of each unit of the visible layer are independent; the activation probabilities of the visible unit h and the implicit unit v are respectively:
wherein σ is an activation function;
the characteristic information learned by one RBM structure is limited, so that the deep belief network adopts a multilayer RBM structure, after RBMs of each layer are trained, the output of a hidden layer is used as the input of a visible layer of the RBMs of the next layer, and the training of all the RBM structures is completed in sequence; finally, in the BP network at the top layer, according to the labeled fault information stated in S11, adjusting an output error, and feeding back the error to each layer of RBM from right to left; and obtaining final parameters of the connection weight w, the bias c and the bias b after a plurality of iterations, and finishing the training of the whole deep confidence network.
S14: and when a fault occurs, collecting the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault.
In a specific implementation process of the present invention, the acquiring characteristic quantities of the power system includes: the characteristic quantity at the initial time of the power system, the characteristic quantity at the fault occurrence time, and the characteristic quantity after the fault is removed.
In specific implementation, as shown in fig. 5, fig. 5 is a schematic diagram of an IEEE39 system in the implementation of the present invention, and a large amount of simulation data is obtained by setting different fault conditions for an IEEE39 node system through PSD-BPA power system time domain simulation software; wherein, the load level is set to nine conditions which are respectively 80%, 85%, 90%, 95%, 100%, 105%, 110%, 115% and 120% of rated load level; the fault type is a three-phase short circuit fault; the fault duration time is 0.1s, 0.15s and 0.2s respectively; setting fault positions to be 0%, 20%, 50% and 80% of a line respectively, generating 4968 fault area identification samples through simulation, and marking fault information according to the fault area division of the figure 5; under the same load level and fault type, taking a line 1-2 as an example, setting fault duration time to be 0.05s, 0.1s,. multidot.9 s, 0.35s and 0.4s, respectively setting fault positions to be 0%, 1%, 2%,.. multidot.98% and 99%, obtaining 7200 fault position identification samples in total, and dividing 0% -99% of the fault positions into four sections in equal proportion for fault information marking; randomly extracting sample data according to a ratio of 8:2 for a fault area identification sample and a fault position identification sample to respectively form a training set and a test set; extracting characteristic quantity data of the system at the initial moment, the fault occurrence moment and the fault removal moment according to the method of the embodiment, and performing normalization processing; training the deep belief network by using the randomly extracted training set samples and the fault marking information according to the method of the embodiment; after training is finished, inputting randomly extracted test set samples into a trained deep confidence network, and comparing the obtained result with test set fault marking information to obtain the accuracy of the fault area identification method based on the deep confidence network, wherein the table 1 is a result table of fault area identification and fault position identification; compared with the existing common artificial intelligence algorithm, the method has the advantage of identifying the fault of the power system based on the deep confidence network.
Table 1 result table of fault area identification and fault location identification
In the implementation of the invention, the characteristic values of the state of the power system can be represented by collecting the characteristic values before the fault, at the moment of the fault occurrence and at the moment of the fault removal of the power system, the characteristic values and the fault marking information are input into the deep confidence network so as to train the deep confidence network, the new characteristic values collected by the fault are input into the trained deep confidence network, the identification of the fault area and the fault position of the power system can be realized, and the accuracy of fault identification is improved.
Examples
Referring to fig. 2, fig. 2 is a schematic structural diagram of a power system fault identification device based on a deep belief network in an implementation of the present invention.
As shown in fig. 2, a power system fault identification device based on a deep belief network, the device includes:
data set acquisition and annotation module 11: the system comprises a training data set and a fault information marking module, wherein the training data set is used for obtaining a training data set in power system time domain simulation software and marking fault information on the training data set;
the feature data extraction and normalization processing module 12: the characteristic data is used for extracting characteristic data representing the state of the power system based on the data in the obtained training data set and carrying out normalization processing on the characteristic data;
the deep confidence network training module 13: the deep belief network training system is used for inputting the extracted feature data and the sample fault labeling information into a deep belief network to be trained, and training the deep belief network to be trained to obtain a trained deep belief network;
in a specific implementation process of the present invention, the deep belief network training module further includes: the method is used for obtaining the connection weight and the bias coefficient of each layer in the deep belief network after unsupervised pre-training and supervised fine tuning.
The defective area identifying module 14: and after the fault occurs, acquiring the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault.
Specifically, the working principle of the device related function module according to the embodiment of the present invention may refer to the related description of the method embodiment, and is not described herein again.
In the implementation of the invention, the characteristic values of the state of the power system can be represented by collecting the characteristic values before the fault, at the moment of the fault occurrence and at the moment of the fault removal of the power system, the characteristic values and the fault marking information are input into the deep confidence network so as to train the deep confidence network, the new characteristic values collected by the fault are input into the trained deep confidence network, the identification of the fault area and the fault position of the power system can be realized, and the accuracy of fault identification is improved.
Those skilled in the art will appreciate that all or part of the steps in the methods of the above embodiments may be implemented by associated hardware instructed by a program, which may be stored in a computer-readable storage medium, and the storage medium may include: a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic or optical disk, or the like.
In addition, the method and the device for identifying the fault of the power system based on the deep confidence network provided by the embodiment of the invention are described in detail, a specific embodiment is adopted to explain the principle and the implementation mode of the invention, and the description of the embodiment is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A power system fault identification method based on a deep belief network is characterized by comprising the following steps:
acquiring a training data set in power system time domain simulation software, and carrying out fault information labeling on the training data set;
extracting characteristic data representing the state of the power system based on the obtained data in the training data set, and performing normalization processing on the characteristic data;
inputting the extracted feature data and the sample fault labeling information into a deep confidence network to be trained, and training the deep confidence network to be trained to obtain a trained deep confidence network;
and when a fault occurs, collecting the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault.
2. The method for power system fault identification based on the deep belief network as claimed in claim 1, wherein the obtaining of the training data set in the power system time domain simulation software comprises:
in the time domain simulation software of the power system, fault data are obtained by setting different fault positions, fault types and fault duration;
acquiring historical fault data of the power system in power system time domain simulation software;
and combining the fault data with the historical fault data of the power system to obtain a training data set.
3. The method for power system fault identification based on the deep belief network as claimed in claim 1, wherein the extracting the feature data characterizing the power system state comprises:
the method comprises the following steps of (1) characteristic data of a system active load level at the initial moment of a system, characteristic data of a system reactive load level at the initial moment of the system, characteristic data of a system active output level at the initial moment of the system and characteristic data of a system reactive output level at the initial moment of the system;
the characteristic data of the maximum rotation angle difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum angular speed difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum kinetic energy difference of the generator in the system at the fault occurrence moment, the characteristic data of the maximum angular speed of the generator in the system at the fault occurrence moment and the characteristic data of the maximum acceleration power of the generator in the system at the fault occurrence moment;
the characteristic data of the maximum rotation angle difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum angular speed difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum kinetic energy difference of the generator in the system at the moment of fault removal, the characteristic data of the maximum angular speed of the generator in the system at the moment of fault removal and the characteristic data of the maximum acceleration power of the generator in the system at the moment of fault removal.
4. The method for identifying the fault of the power system based on the deep belief network as claimed in claim 1, wherein the specific calculation formula for normalizing the characteristic data is as follows:
wherein x' represents the data after normalization processing; x represents data before normalization processing; x is the number ofmaxMaximum value, x, representing characteristic dataminRepresenting the minimum value of the characteristic data.
5. The method for power system fault identification based on the deep belief network as claimed in claim 1, wherein the training of the deep belief network to be trained comprises:
and obtaining the connection weight and the bias coefficient of each layer in the deep belief network after unsupervised pre-training and supervised fine tuning.
6. The method for power system fault identification based on the deep confidence network is characterized in that the deep confidence network comprises a plurality of layers of unsupervised limited boltzmann machines and a layer of supervised feedforward neural network.
7. The method for power system fault identification based on the deep belief network of claim 6, wherein the multilayer unsupervised restricted boltzmann machine is used for extracting feature information in data, pre-training parameters;
the layer of supervised feedforward neural network is used for global parameter optimization and sample classification and parameter fine adjustment.
8. The method for identifying the fault of the power system based on the deep belief network as claimed in claim 1, wherein the collecting the characteristic quantity of the power system comprises: the characteristic quantity at the initial time of the power system, the characteristic quantity at the fault occurrence time, and the characteristic quantity after the fault is removed.
9. An apparatus for power system fault identification based on a deep belief network, the apparatus comprising:
a dataset acquisition and labeling module: the system comprises a training data set and a fault information marking module, wherein the training data set is used for obtaining a training data set in power system time domain simulation software and marking fault information on the training data set;
the characteristic data extraction and normalization processing module: the characteristic data is used for extracting characteristic data representing the state of the power system based on the data in the obtained training data set and carrying out normalization processing on the characteristic data;
the deep confidence network training module: the deep belief network training system is used for inputting the extracted feature data and the sample fault labeling information into a deep belief network to be trained, and training the deep belief network to be trained to obtain a trained deep belief network;
a fault area identification module: and after the fault occurs, acquiring the characteristic quantity of the power system and inputting the characteristic quantity into the trained deep confidence network to obtain a fault area identification result of the new fault.
10. The device for power system fault recognition based on the deep belief network as claimed in claim 9, wherein the deep belief network training module further comprises: the method is used for obtaining the connection weight and the bias coefficient of each layer in the deep belief network after unsupervised pre-training and supervised fine tuning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010183621.3A CN111553112A (en) | 2020-03-16 | 2020-03-16 | Power system fault identification method and device based on deep belief network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010183621.3A CN111553112A (en) | 2020-03-16 | 2020-03-16 | Power system fault identification method and device based on deep belief network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111553112A true CN111553112A (en) | 2020-08-18 |
Family
ID=72005531
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010183621.3A Pending CN111553112A (en) | 2020-03-16 | 2020-03-16 | Power system fault identification method and device based on deep belief network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111553112A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113159345A (en) * | 2021-05-08 | 2021-07-23 | 中国电力科学研究院有限公司 | Power grid fault identification method and system based on fusion neural network model |
| CN113191219A (en) * | 2021-04-15 | 2021-07-30 | 华能威宁风力发电有限公司 | Fan bearing fault self-adaptive identification method |
| CN113723268A (en) * | 2021-08-25 | 2021-11-30 | 国网北京市电力公司 | Method and device for identifying power grid fault, computer storage medium and processor |
| WO2022183698A1 (en) * | 2021-03-05 | 2022-09-09 | 国网电力科学研究院武汉南瑞有限责任公司 | Artificial intelligence fault identification system and method based on power transmission line transient waveform |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932499A (en) * | 2018-07-09 | 2018-12-04 | 武汉科技大学 | A kind of rolling bearing prediction technique and system based on local binary patterns and depth confidence network |
| CN109239527A (en) * | 2018-09-07 | 2019-01-18 | 福州大学 | Distribution network failure recognition methods based on depth confidence network |
-
2020
- 2020-03-16 CN CN202010183621.3A patent/CN111553112A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932499A (en) * | 2018-07-09 | 2018-12-04 | 武汉科技大学 | A kind of rolling bearing prediction technique and system based on local binary patterns and depth confidence network |
| CN109239527A (en) * | 2018-09-07 | 2019-01-18 | 福州大学 | Distribution network failure recognition methods based on depth confidence network |
Non-Patent Citations (3)
| Title |
|---|
| 刘文泽 等: "基于深度置信网络和多维信息融合的变压器故障诊断方法", 《电力工程技术》 * |
| 朱乔木 等: "基于深度置信网络的电力系统暂态稳定评估方法", 《中国电机工程学报》 * |
| 王哲 等: "人工智能技术在电力系统故障诊断中的应用研究", 《电子设计工程》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022183698A1 (en) * | 2021-03-05 | 2022-09-09 | 国网电力科学研究院武汉南瑞有限责任公司 | Artificial intelligence fault identification system and method based on power transmission line transient waveform |
| CN113191219A (en) * | 2021-04-15 | 2021-07-30 | 华能威宁风力发电有限公司 | Fan bearing fault self-adaptive identification method |
| CN113191219B (en) * | 2021-04-15 | 2022-11-01 | 华能威宁风力发电有限公司 | Fan bearing fault self-adaptive identification method |
| CN113159345A (en) * | 2021-05-08 | 2021-07-23 | 中国电力科学研究院有限公司 | Power grid fault identification method and system based on fusion neural network model |
| CN113723268A (en) * | 2021-08-25 | 2021-11-30 | 国网北京市电力公司 | Method and device for identifying power grid fault, computer storage medium and processor |
| CN113723268B (en) * | 2021-08-25 | 2024-01-26 | 国网北京市电力公司 | Method, device, computer storage medium and processor for identifying power grid faults |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111553112A (en) | Power system fault identification method and device based on deep belief network | |
| CN109635928B (en) | Voltage sag reason identification method based on deep learning model fusion | |
| CN110994604B (en) | Power system transient stability assessment method based on LSTM-DNN model | |
| CN111478314B (en) | Transient stability evaluation method for power system | |
| CN110609477B (en) | Electric power system transient stability discrimination system and method based on deep learning | |
| CN108896857B (en) | Deep learning-based complex working condition identification method for transformer | |
| CN102570392A (en) | Method for identifying exciting inrush current of transformer based on improved probability neural network | |
| CN110726898B (en) | Power distribution network fault type identification method | |
| CN112200694A (en) | Dominant instability mode identification model construction and application method based on graph neural network | |
| CN114021433A (en) | Construction method and application of dominant instability mode recognition model of power system | |
| CN112557826A (en) | Ship electric power system fault diagnosis method | |
| CN113093058A (en) | NPC three-level inverter open-circuit fault diagnosis method | |
| CN114330571B (en) | Power system dominant instability mode identification method and system based on transfer learning | |
| CN114169395A (en) | Construction method and application of dominant instability mode recognition model of power system | |
| CN114330486A (en) | Power system bad data identification method based on improved Wasserstein GAN | |
| CN112633315A (en) | Electric power system disturbance classification method | |
| CN117078035A (en) | LSTM-based power system transient stability evaluation method, system and storage medium | |
| CN113488997B (en) | Micro-grid critical energy evaluation method based on disturbance response multi-scale features | |
| CN114384319A (en) | Grid-connected inverter island detection method, system, terminal and medium | |
| CN112882899B (en) | Log abnormality detection method and device | |
| CN117669656A (en) | TCN-Semi PN-based direct-current micro-grid stability real-time monitoring method and device | |
| CN116432023A (en) | Novel power system fault classification method based on sample transfer learning | |
| CN116680532A (en) | Transformer fault online diagnosis method for processing unbalanced small sample based on NNTR-SMOTE oversampling | |
| Yin et al. | Graph attention-based U-net conditional generative adversarial networks for the identification of synchronous generation unit parameters | |
| He et al. | Partial discharge pattern recognition algorithm based on sparse self-coding and extreme learning machine |
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: 20200818 |
|
| RJ01 | Rejection of invention patent application after publication |