CN113988136A - Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization - Google Patents

Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization Download PDF

Info

Publication number
CN113988136A
CN113988136A CN202111276801.7A CN202111276801A CN113988136A CN 113988136 A CN113988136 A CN 113988136A CN 202111276801 A CN202111276801 A CN 202111276801A CN 113988136 A CN113988136 A CN 113988136A
Authority
CN
China
Prior art keywords
circuit breaker
cnn
training
algorithm
voltage circuit
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
CN202111276801.7A
Other languages
Chinese (zh)
Inventor
张丹
黄钟汀
陈永毅
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.)
Zhejiang University of Technology ZJUT
Original Assignee
Zhejiang University of Technology ZJUT
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 Zhejiang University of Technology ZJUT filed Critical Zhejiang University of Technology ZJUT
Priority to CN202111276801.7A priority Critical patent/CN113988136A/en
Publication of CN113988136A publication Critical patent/CN113988136A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Signal Processing (AREA)
  • Medical Informatics (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention discloses a medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization, which comprises the following steps of: 1) collecting vibration signals of the medium-voltage circuit breaker in a normal state, a tripping closed electromagnet blockage, a main shaft blockage and a half shaft blockage as an original data set; 2) carrying out normalization processing on the training set data and the test set data; 3) constructing a CNN deep neural network model; 4) performing optimization training on the SVM classifier by combining the trained CNN deep neural network model with a quantum particle swarm algorithm; 5) and inputting the test sample data into the trained fault diagnosis model to diagnose the fault of the circuit breaker. The method effectively extracts the data features by utilizing the advantage of strong capability of extracting the features by the convolutional neural network; and the quantum particle group algorithm is further utilized to effectively eliminate the advantage of local optimal phenomenon, so that the accuracy of data classification is improved.

Description

Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization
Technical Field
The invention relates to the technical field of real-time online diagnosis of main mechanical faults of medium-voltage circuit breakers, in particular to an intelligent fault diagnosis method of a medium-voltage circuit breaker based on a convolutional neural network and quantum particle swarm optimization, which is applied to the medium-voltage circuit breaker to solve the problem of fault diagnosis.
Background
The circuit breaker plays an important role in the power system as an important power device, and the operating state of the circuit breaker directly affects the safety and reliability of the power system. However, the maintenance cost of the circuit breaker is huge, and the traditional manual inspection mode not only wastes time and labor, but also brings unnecessary shutdown. The non-invasive fault diagnosis method is gradually becoming an important research direction of the circuit breaker maintenance technology due to the convenience of data acquisition.
The fault diagnosis technology mainly comprises two aspects of fault signal feature extraction and fault diagnosis algorithm design. Abundant vibration signals are contained in the operation process of the circuit breaker, the vibration signals in live operation are used as input signals, characteristic factors representing states can be effectively extracted, the position of the circuit breaker with a fault is analyzed, and state monitoring and diagnosis of the circuit breaker are achieved. In the selection of the fault diagnosis algorithm, the traditional artificial intelligence method comprises the following steps: performing breaker fault diagnosis by adopting a Back Propagation (BP) algorithm and a Radial Basis Function (RBF); evaluating the state of the circuit breaker by adopting a fuzzy theory; and constructing a mechanical fault diagnosis network of the circuit breaker by adopting a Particle Swarm Optimization (PSO) algorithm and a radial basis function. Although the diagnosis precision of the traditional artificial intelligence algorithm is obviously improved and the traditional artificial intelligence algorithm is also applied to engineering practice to a certain extent, the intelligent algorithms generally have the problems of low diagnosis speed, poor generalization capability, easy generation of local optimal solution and the like, the unnecessary shutdown and maintenance time of the circuit breaker is increased, and the development of the fault diagnosis technology of the circuit breaker is limited.
Aiming at various defects of the traditional artificial intelligence algorithm, the intelligent fault diagnosis method for the medium-voltage circuit breaker based on the Convolutional Neural Network (CNN) and the Quantum Particle Swarm Optimization (QPSO) algorithm is provided. The method is characterized in that a circuit breaker fault diagnosis classification network is established on the basis of CNN, the CNN is trained by using circuit breaker vertical vibration signal data as training samples, CNN parameters are kept unchanged in order to prevent the CNN model from having a local optimal problem, the trained CNN model is combined with a quantum particle swarm algorithm to optimize parameters of an SVM classification model, and optimal SVM parameters are output. An SVM fault classification model is constructed based on the optimal parameters, the test samples are input into a CNN-QPSO-SVM fault diagnosis model, the final fault diagnosis result is output, and the fault type of the medium-voltage circuit breaker can be effectively identified.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization, which is reasonable in design.
The purpose of the invention is realized by the following technical scheme:
a medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization comprises the following steps:
(1) and collecting vibration signals of the medium-voltage circuit breaker in a normal state, when the electromagnet is closed by tripping, when the main shaft is blocked and when the half shaft is blocked, and performing model training as original signals.
(2) Carrying out normalization processing on the original signal, wherein the normalization formula is as follows:
Figure BDA0003329697860000021
xifor the original sample data, xmaxIs xiMaximum value of (a), xminIs xiIs measured. In order to find the optimal parameter combination of the SVM, the normalized samples are classified, a part of data is selected as a training data set, and the rest of data is used as a testing data set.
(3) And constructing a CNN deep neural network model, inputting a training sample into the CNN for training, wherein the CNN deep neural network comprises 6 convolutional layers, 4 pooling layers, 1 Softmax layer and 1 dropout layer, and the back of each 2 convolutional layers is connected with 1 pooling layer. And after the training is completed according to the sequence, the training accuracy and the training loss rate of the CNN model are obtained.
(4) And optimally training the SVM classifier by combining the trained CNN deep neural network model with a quantum particle swarm algorithm.
(5) Constructing a fault classification model of the SVM medium-voltage circuit breaker based on the optimal parameters output by the quantum particle swarm optimization, inputting test sample data into the trained CNN-QPSO-SVM fault diagnosis model, and outputting a fault diagnosis result.
Further, in the step (3), the Softmax layer in the CNN network classifies different types of data; in order to avoid the over-fitting phenomenon, the network adopts a dropout regularization method, and a dropout layer is added behind a Softmax layer. The CNN training method adopts an Adam gradient descent method, the loss function is a cross entropy loss function based on Softmax, the batch size is 2, and the iteration number is 200. And after the training is finished, keeping the parameters of the CNN model unchanged.
Further, the specific process of the step (4) is as follows:
the central point introduced by quantum particle swarm optimization is as follows:
Figure BDA0003329697860000031
i.e. the average particle history best position;
in the process of carrying out SVM parameter optimization by using a quantum particle swarm algorithm, the deviation of the adaptive value of each generation of particles is tracked in real time, if the deviation of a certain generation is smaller than a threshold value, the average optimal position is updated, and the updating formula of the average optimal position is as follows:
Pi(t+1)＝φPi(t)+(1-φ)Pg(t)
Xi(t+1)＝Pi(t+1)+/-λ|mbest-Xi(t)|ln(1/u)
phi and u are random variables and are uniformly distributed on (0, 1); λ is the only control parameter, called "innovation coefficient", generally takes a value less than 1;
the quantum-behaved particle swarm algorithm also introduces the evolution speed and the aggregation of particles to avoid premature convergence: the evolutionary speed of the particle group is defined as e, the aggregation degree of the particles is defined as c, the convergence speed of the algorithm is defined as beta, c, and the relational expression of e is as follows:
β＝β0+cβc-eβe
in the formula: beta is a0Is an initial value of beta, usually beta0＝1，βc,βeIs [0,1 ]]A random number in between;
in the iteration process, the deviation of the adaptive value of each generation of particles is tracked in real time, if the adaptive value of a certain generation is smaller than a threshold value, interference operation is carried out on the average optimal position to enhance the global optimization capability of the algorithm, and the algorithm outputs the optimal SVM parameter until the algorithm meets the termination condition.
The invention has the following beneficial effects: the method adopts the one-dimensional convolutional neural network to realize the characteristic extraction of the original vibration signal of the medium-voltage circuit breaker, and compared with the two-dimensional convolutional neural network, the one-dimensional convolutional neural network has more excellent characterization capability on the vibration signal of the medium-voltage circuit breaker and can effectively extract data characteristics; meanwhile, a QPSO algorithm optimization module is introduced, the advantage of local optimal phenomenon is effectively eliminated by using a quantum particle group algorithm, and the accuracy of fault classification is improved.
Drawings
FIG. 1 is a flow chart of the intelligent fault diagnosis method of the medium-voltage circuit breaker based on the CNN-QPSO-SVM of the invention;
FIG. 2 is a schematic diagram of the basic structure of the CNN deep neural network of the present invention;
FIG. 3 is a waveform diagram of a vibration signal in a normal state according to an embodiment of the present invention;
fig. 4 is a waveform of a vibration signal showing a blocked state of a trip closing electromagnet according to an embodiment of the present invention;
FIG. 5 is a waveform of a vibration signal in a spindle jam state according to an embodiment of the present invention;
FIG. 6 is a waveform of a vibration signal in a half-shaft blocked state according to an embodiment of the present invention;
FIG. 7 is a waveform illustrating diagnostic accuracy of a training data set and a test data set in accordance with an embodiment of the present invention;
FIG. 8 is a diagnostic loss rate waveform for a training data set and a test data set in accordance with an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, 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 some, but not all, embodiments of the present invention.
The technical conception of the invention is as follows: firstly, inputting an original vibration signal in the vertical direction of the medium-voltage circuit breaker into a constructed CNN deep neural network model for feature extraction, then identifying the fault type by using an SVM classifier, improving the SVM classification precision by using a QPSO algorithm, and finally diagnosing the fault type of the medium-voltage circuit breaker by using a trained CNN-QPSO-SVM fault diagnosis method.
Example (b): for several common fault types in the medium-voltage circuit breaker, referring to fig. 3-6, an intelligent diagnosis method for the medium-voltage circuit breaker based on CNN-QPSO-SVM comprises the following steps:
s1: and recording vibration acceleration parameters in the opening/closing process of the circuit breaker by using an acceleration sensor, wherein the sampling frequency is 10 kHz. Acceleration sensor lays on the casing of circuit breaker for gather vertical vibration signal, gathered 7 groups vibration signal data altogether, contained: a normal state; tripping and closing electromagnet blockage; the main shaft is blocked; half shaft plugging 4 major failure types. The types of failures and their numbers are shown in table 1.
TABLE 1 common class 4 mechanical Fault numbering for Circuit breakers
Figure BDA0003329697860000051
S2: the training data and the test data are standardized, and the normalization formula is as follows:
Figure BDA0003329697860000052
xifor the original sample data, xmaxIs xiMaximum value of (a), xminIs xiIs measured. Of these 7 sets of data, 3 were selected as training data sets and the remaining 4 were selected as test data sets for each state, with 7 experiments being performed in each case. As is apparent from fig. 3 to 6, the vibration signals have no significant difference or change in the time domain, and the feature analysis of the weak vibration signals is particularly important for state recognition.
S3: a CNN deep neural network model shown in FIG. 2 is constructed, training samples are input into the CNN for training, the CNN deep neural network has 6 convolutional layers, 4 pooling layers, 1 Softmax layer and 1 dropout layer, and the back of each 2 convolutional layers is connected with 1 pooling layer. The structural parameters of each layer of the constructed one-dimensional convolution depth neural network model are shown in table 2.
TABLE 2 one-dimensional convolutional neural network structural parameters
Figure BDA0003329697860000061
The pooling layer is divided into an average pooling type and a maximum pooling type; and the Softmax layer classifies data of different classes, a dropout regularization method is adopted to avoid the occurrence of an over-fitting phenomenon, and a dropout layer is added behind the Softmax layer. The CNN training method adopts an Adam gradient descent method, the loss function is a cross entropy loss function based on Softmax, the batch size is 2, and the iteration number is 200.
S4: keeping the CNN parameters unchanged, and optimally training the SVM classifier by combining the trained CNN deep neural network model with a quantum particle swarm algorithm. Central point introduced by QPSO algorithm:
Figure BDA0003329697860000062
i.e. the average particle history best position. The updating process adopts the following formula:
Pi(t+1)＝φPi(t)+(1-φ)Pg(t)
Xi(t+1)＝Pi(t+1)+/-λ|mbest-Xi(t)|ln(1/u)
phi and u are random variables and are uniformly distributed on (0, 1); λ is the only control parameter, called "innovation coefficient", generally with a value less than 1. PiUpdate for ith particle position, Pg represents current globally optimal particle, XiIndicating the location update of the nth seek, mbestRepresenting the average particle history optimal position.
The QPSO algorithm also introduces the speed of evolution and the degree of aggregation of the particles to avoid premature convergence: the evolutionary speed of the particle group is defined as e, the aggregation degree of the particles is defined as c, the convergence speed of the algorithm is defined as beta, c, and the relational expression of e is as follows:
β＝β0+cβc-eβe
in the formula: beta is a0Is an initial value of beta, usually beta0＝1，βc,βeIs [0,1 ]]A random number in between. In the iteration process, the deviation of the adaptive value of each generation of particles is tracked in real time, if the adaptive value of a certain generation is smaller than a threshold value, interference operation is carried out on the average optimal position to enhance the global optimization capability of the algorithm, and the algorithm outputs the optimal SVM parameter until the algorithm meets the termination condition.
S5: an SVM medium-voltage circuit breaker fault diagnosis model is constructed based on optimal parameters output by a QPSO algorithm, test sample data are input into the trained CNN-QPSO-SVM fault diagnosis model, a fault diagnosis result is output, and the diagnosis accuracy and loss rate of the training data set and the test data set are shown in figures 7-8.
In order to further prove the superiority of the method, experimental data are combined with the intelligent fault diagnosis method of the medium-voltage circuit breaker based on the CNN-LSTM, the CNN-SVM and the SVM for comparison. The same training data set was used for training using different fault diagnosis methods, and the same test data set was diagnosed at the same time, with the resulting diagnosis accuracy as shown in table 3.
TABLE 3 comparison of diagnostic accuracy for different algorithms
Figure BDA0003329697860000071
From table 3, it can be seen that the prediction accuracy of the CNN-QPSO-SVM algorithm is 100%, and when the CNN-SVM or SVM algorithm is used alone for prediction, the prediction accuracy is 75% and 50%, respectively, and the fault type cannot be identified effectively, which indicates that QPSO parameter optimization has a great influence on the improvement of the prediction accuracy of the CNN and SVM algorithms, so that parameter optimization is performed when the CNN and SVM algorithms are used for prediction, and the algorithm prediction accuracy can be improved to a great extent. In addition, although the prediction accuracy of the CNN-QPSO-SVM algorithm and the CNN-LSTM algorithm is close to 100%, compared with a two-dimensional convolutional neural network, the one-dimensional convolutional neural network has more excellent characterization capability on the vibration signal of the circuit breaker, so that the CNN-QPSO-SVM algorithm has higher identification accuracy than the CNN-LSTM algorithm. From the above comparison it can be derived that: the CNN-QPSO-SVM algorithm is the best performing of the several algorithms.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Although terms like fault diagnosis, deep convolutional neural networks, etc. are used more herein, the possibility of using other terms is not excluded. These terms are used merely to more conveniently describe and explain the nature of the present invention and they are to be interpreted as any additional limitation which is not in accordance with the spirit of the present invention.

Claims (3)

1. A medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization is characterized by comprising the following steps:
1) collecting vibration signals of the medium-voltage circuit breaker in a normal state, when the electromagnet is closed by tripping, when the main shaft is blocked and when the half shaft is blocked, as original signals, carrying out model training;
2) carrying out normalization processing on the original signal, wherein the normalization formula is as follows:
Figure FDA0003329697850000011
wherein xiFor the original sample data, xmaxIs xiMaximum value of (a), xminIs xiMinimum value of (d); in order to find the optimal parameter combination of the SVM, the normalized samples are classified, one part of data is selected as a training data set, and the other part of data is selected as a testing data set;
3) constructing a CNN deep neural network model, inputting a training sample into the CNN for training, wherein the CNN deep neural network comprises 6 convolutional layers, 4 pooling layers, 1 Softmax layer and 1 dropout layer, and the back of each 2 convolutional layers is connected with 1 pooling layer; after training is completed according to the sequence, the training accuracy and the training loss rate of the CNN model are obtained;
4) performing optimization training on the SVM classifier by combining the trained CNN deep neural network model with a quantum particle swarm algorithm;
5) constructing a fault classification model of the SVM medium-voltage circuit breaker based on the optimal parameters output by the quantum particle swarm optimization, inputting test sample data into the trained CNN-QPSO-SVM fault diagnosis model, and outputting a fault diagnosis result.
2. The medium voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization of claim 1, characterized in that in the step 3), Softmax layer in CNN network classifies different kinds of data; in order to avoid the over-fitting phenomenon, the network adopts a dropout regularization method, and a dropout layer is added behind a Softmax layer; the CNN training method adopts an Adam gradient descent method, the loss function is a cross entropy loss function based on Softmax, the batch size is 2, and the iteration number is 200; and after the training is finished, keeping the parameters of the CNN model unchanged.
3. The method for diagnosing the fault of the medium-voltage circuit breaker based on deep learning and intelligent optimization as claimed in claim 1, wherein the specific process of the step 4) is as follows:
the central point introduced by quantum particle swarm optimization is as follows:
Figure FDA0003329697850000021
i.e. the average particle history best position;
in the process of carrying out SVM parameter optimization by using a quantum particle swarm algorithm, the deviation of the adaptive value of each generation of particles is tracked in real time, if the deviation of a certain generation is smaller than a threshold value, the average optimal position is updated, and the updating formula of the average optimal position is as follows:
Pi(t+1)＝φPi(t)+(1-φ)Pg(t)
Xi(t+1)＝Pi(t+1)+/-λ|mbest-Xi(t)|ln(1/u)
phi and u are random variables and are uniformly distributed on (0, 1); λ is the only control parameter, called innovation coefficient, generally less than 1; piUpdate for ith particle position, Pg represents current globally optimal particle, XiIndicating the location update of the nth seek, mbestRepresenting the historical optimal position of the average particle;
the quantum-behaved particle swarm algorithm also introduces the evolution speed and the aggregation of particles to avoid premature convergence: the evolutionary speed of the particle group is defined as e, the aggregation degree of the particles is defined as c, the convergence speed of the algorithm is defined as beta, c, and the relational expression of e is as follows:
β＝β0+cβc-eβe
in the formula: beta is a0Is an initial value of beta, usually beta0＝1，βc,βeIs [0,1 ]]A random number in between;
in the iteration process, the deviation of the adaptive value of each generation of particles is tracked in real time, if the adaptive value of a certain generation is smaller than a threshold value, interference operation is carried out on the average optimal position to enhance the global optimization capability of the algorithm, and the algorithm outputs the optimal SVM parameter until the algorithm meets the termination condition.
CN202111276801.7A 2021-10-29 2021-10-29 Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization Pending CN113988136A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111276801.7A CN113988136A (en) 2021-10-29 2021-10-29 Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111276801.7A CN113988136A (en) 2021-10-29 2021-10-29 Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization

Publications (1)

Publication Number Publication Date
CN113988136A true CN113988136A (en) 2022-01-28

Family

ID=79744861

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111276801.7A Pending CN113988136A (en) 2021-10-29 2021-10-29 Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization

Country Status (1)

Country Link
CN (1) CN113988136A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114781507A (en) * 2022-04-18 2022-07-22 杭州电子科技大学 1 DCNN-DS-based water chilling unit fault diagnosis method
CN115902615A (en) * 2023-01-09 2023-04-04 佰聆数据股份有限公司 Method and device for analyzing defects of power circuit breaker
CN116070151A (en) * 2023-03-17 2023-05-05 国网安徽省电力有限公司超高压分公司 Ultra-high voltage direct current transmission line fault detection method based on generalized regression neural network

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114781507A (en) * 2022-04-18 2022-07-22 杭州电子科技大学 1 DCNN-DS-based water chilling unit fault diagnosis method
CN114781507B (en) * 2022-04-18 2024-04-05 杭州电子科技大学 1 DCNN-DS-based water chilling unit fault diagnosis method
CN115902615A (en) * 2023-01-09 2023-04-04 佰聆数据股份有限公司 Method and device for analyzing defects of power circuit breaker
CN116070151A (en) * 2023-03-17 2023-05-05 国网安徽省电力有限公司超高压分公司 Ultra-high voltage direct current transmission line fault detection method based on generalized regression neural network

Similar Documents

Publication Publication Date Title
CN113988136A (en) Medium-voltage circuit breaker fault diagnosis method based on deep learning and intelligent optimization
CN113935406B (en) Mechanical equipment unsupervised fault diagnosis method based on countermeasure flow model
CN110849626B (en) Self-adaptive sparse compression self-coding rolling bearing fault diagnosis system
CN110502991B (en) Internal combustion engine health monitoring method and system based on random convolutional neural network structure
CN111914883B (en) Spindle bearing state evaluation method and device based on deep fusion network
CN111504676A (en) Equipment fault diagnosis method, device and system based on multi-source monitoring data fusion
CN111753891B (en) Rolling bearing fault diagnosis method based on unsupervised feature learning
CN112257530B (en) Rolling bearing fault diagnosis method based on blind signal separation and support vector machine
CN113505655A (en) Bearing fault intelligent diagnosis method for digital twin system
CN115791174B (en) Rolling bearing abnormality diagnosis method, system, electronic equipment and storage medium
CN113158722A (en) Rotary machine fault diagnosis method based on multi-scale deep neural network
CN116244617A (en) Intelligent high-voltage switch cabinet fault diagnosis method and system based on heterogeneous graph structure learning
CN114034486A (en) Unsupervised transfer learning-based bearing fault diagnosis method for pump mechanical equipment
Sá et al. A multilayer perceptron-based approach for stator fault detection in permanent magnet wind generators
CN116205265A (en) Power grid fault diagnosis method and device based on deep neural network
CN114152442B (en) Rolling bearing cross-working condition fault detection method based on migration convolutional neural network
CN114217256A (en) Fault diagnosis method for current transformer of motor driving system based on adaptive sparse filtering
CN117332353A (en) Method for diagnosing abnormal vibration faults of rotor of steam turbine generator
CN116400168A (en) Power grid fault diagnosis method and system based on depth feature clustering
CN114330413A (en) Fault type identification and positioning method for traction motor bearing
CN113822337A (en) Industrial control abnormity detection method based on multi-dimensional sequence
Zhao et al. Fault Diagnosis of Rolling Bearings based on GA-SVM model
Zhang et al. Sequential mechanical fault diagnosis in high voltage circuit breaker using attention mechanism
Xiang et al. An Intelligent Fault Diagnosis Method of Rolling Bearing with Wide Convolution Kernel Network
Liu et al. Research on fault diagnosis of nuclear gate valve based on analysis of energy variation characteristics of vibration signal

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