CN113533952A - CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer - Google Patents

CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer Download PDF

Info

Publication number
CN113533952A
CN113533952A CN202110779302.3A CN202110779302A CN113533952A CN 113533952 A CN113533952 A CN 113533952A CN 202110779302 A CN202110779302 A CN 202110779302A CN 113533952 A CN113533952 A CN 113533952A
Authority
CN
China
Prior art keywords
phase space
space reconstruction
load tap
ceemd
fault diagnosis
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
CN202110779302.3A
Other languages
Chinese (zh)
Inventor
陈志华
柯强
胡经伟
李建坤
刘洋
饶燕
万姗
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.)
State Grid Corp of China SGCC
Huanggang Power Supply Co of State Grid Hubei Electric Power Co Ltd
Original Assignee
State Grid Corp of China SGCC
Huanggang Power Supply Co of State Grid Hubei Electric Power Co Ltd
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 State Grid Corp of China SGCC, Huanggang Power Supply Co of State Grid Hubei Electric Power Co Ltd filed Critical State Grid Corp of China SGCC
Priority to CN202110779302.3A priority Critical patent/CN113533952A/en
Publication of CN113533952A publication Critical patent/CN113533952A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/327—Testing of circuit interrupters, switches or circuit-breakers

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)

Abstract

The invention discloses a CEEMD and phase space reconstruction based mechanical fault diagnosis and analysis method for a tap changer of an on-load tap-changing transformer, which comprises the following steps: establishing a complementary set empirical mode decomposition and phase space reconstruction model; acquiring a vibration signal of a tap switch of the on-load tap changing transformer; decomposing the signal into natural mode functions with different frequency characteristics; determining the delay time and the embedding dimension of the IMFs through a C-C method; extracting characteristic quantities (Lyapunov exponent and correlation dimension) of the chaotic characteristics of the vibration signals; and establishing and learning a vector quantization neural network (LVQ) to carry out state classification and identification system model to obtain the mechanical fault diagnosis result of the on-load tap changer of the voltage regulating transformer. The invention decomposes the vibration signals with different frequencies through complementary set empirical mode, utilizes the chaotic characteristic of the vibration signals, and adopts the combination of phase space reconstruction and LVQ to carry out classification pretreatment, thereby effectively realizing the identification of the fault state of the tap changer.

Description

CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer
Technical Field
The application belongs to the field of mechanical fault diagnosis of on-load tap-changer tap-changers, and particularly relates to a CEEMD and phase space reconstruction based mechanical fault diagnosis method for the on-load tap-changers.
Background
With the rapid development of power grids and the improvement of the requirements of the nation on the quality of electric energy, the on-load voltage regulation power transformer is more and more widely applied. The on-load tap changer plays an important role in a power system and plays roles in regulating power flow and stabilizing conforming voltage. The on-load tap changer (OLTC) performs the regulation function by its only movable part, the OLTC failure probability increases with the number of times of regulation, and mechanical failure is the main failure type of the on-load tap changer.
The OLTC on-line monitoring and fault diagnosis technology is used for judging the mechanical state of the OLTC by recording mechanical vibration signals and driving motor current signals acquired by a sensing device and extracting time-frequency domain characteristics of the signals through signal processing. The OLTC vibration structure is a very complex nonlinear dynamic system, the vibration signal is represented with chaotic characteristics, and only a small amount of effective information can be extracted by analyzing the OLTC vibration signal by using a time-frequency domain method.
Disclosure of Invention
The application provides a CEEMD and phase space reconstruction-based on-load tap-changer mechanical fault diagnosis and analysis method, which aims to solve the problem of low accuracy of the existing on-load tap-changer mechanical fault diagnosis method.
In order to solve the technical problem, the application discloses the following technical scheme:
a CEEMD and phase space reconstruction based mechanical fault diagnosis method for a tap changer of an on-load tap-changer comprises the following steps:
establishing a complementary set empirical mode decomposition and phase space reconstruction model;
acquiring a vibration signal of a tap switch of the on-load tap changing transformer;
decomposing the signal into natural mode functions with different frequency characteristics;
determining the delay time and the embedding dimension of the IMFs through a C-C method;
extracting characteristic quantities (Lyapunov exponent and correlation dimension) of the chaotic characteristics of the vibration signals;
and establishing and learning a vector quantization neural network (LVQ) to carry out state classification and identification system model, and obtaining a mechanical fault diagnosis result.
Further, establishing a complementary set empirical mode decomposition and phase space reconstruction model, including:
acquiring historical data of vibration data of the on-load tap changer of the voltage regulating transformer;
extracting different frequency components of the historical data of the vibration signal;
performing phase space reconstruction on the different frequency components, and extracting chaotic characteristic quantities (Lyapunov exponent and correlation dimension) reflecting the phase space;
further, the configuration of the frequency components, wherein,
a pair of white noises is added to the original three vibration signals (normal switching, switch sliding gear and motor mechanism fault) respectively. To extract the features of the OLTC vibration signal, the original vibration signal x is filtered using CEEMDi(t) adding a pair of white noises ni(t) two different new signals P are obtainedi(t) and Ni(t)。
Pi＝x(t)+ni(t) (1)
Ni＝x(t)-ni(t) (2)
EMD decomposition is respectively carried out on the two different new signals to obtain two groups of m integrated IMF components
Figure BDA0003155600890000021
And
Figure BDA0003155600890000022
Figure BDA0003155600890000023
repeating the steps, adding a new normal distribution white noise sequence every time, and obtaining the IMF component c every timej(t) as final result.
Selecting any IMF component time sequence { Yi(t) } constructing the state vector via different delays, the reconstruction phase space can be represented by equation (6).
Ym(n)＝[x(n),x(n+τ),…,x(n+(m-1)τ)] (4)
x(k)＝x(t+kΔt),k＝1,2,…,N (5)
Figure BDA0003155600890000024
Wherein x (k) is a discretized system value at the time k; τ is the delay time; m is the embedding dimension; t is the sampling start time; Δ t is the sampling interval; n is the sample length.
Optionally, a C-C algorithm is used to perform phase space reconstruction on the delay time and the embedding dimension.
According to different delay time tau, each order of natural modal component sequence { Yi(t) is divided into τ disjoint time sequences S (m, N, r, τ).
Figure BDA0003155600890000031
Figure BDA0003155600890000032
M＝N-(m-1)τ (9)
dij＝||yi-yj||∞ (10)
Figure BDA0003155600890000033
Wherein d isijIs an infinite function; r is the search radius, and is taken to be less than max (d)ij) Any value of (a); θ (x) is the Heaviside function; c (m, N, r, tau) is the correlation integral of the embedding time series;
selecting the maximum radius rmaxAnd a minimum radius rminThen, the corresponding radius difference Δ S (m, N, r, τ), average inspection statistic S, average difference Δ S (t), and index S are calculatedcor(t) and comparing the zero position of S (m, N, r, tau) or the minimum point of Delta S (m, N, r, tau), and recording the smaller value of the two as the time delay tau.
Figure BDA0003155600890000034
Figure BDA0003155600890000035
Figure BDA0003155600890000036
Figure BDA0003155600890000037
Wherein n ismM possible values; n iskAre k possible values.
Drawing
Figure BDA0003155600890000038
And Scor(τ) curve of change, delay time window τwIs at Scor(τ) delay time corresponding to when global minimum is obtainedThen, the embedding dimension m is calculated according to equation (16).
Figure BDA0003155600890000039
Because the redundancy of the reconstruction phase space and the crowdedness degree of the singular attractor track can be changed due to the change of the two important parameters, the delay time and the embedding dimension obtained by the C-C algorithm are optimal values.
Optionally, calculating a feature quantity (lyapunov exponent and correlation dimension) for extracting the chaotic feature of the vibration signal:
and (3) constructing an m-dimensional phase space new sequence for the sample sequence with the N sampling points by using a phase space reconstruction technology, wherein the calculation formula is (16).
Finding and initiating phase point P0(t0) The closest point is tracked and the distance L between the two points is measured0If greater than epsilon.
L0＝|P(t1)-P0(t0)|>ε (17)
Finding a point and P0(t0) Minimum included angle and nearest point P (t)1)。
L'1＝|P(t1)-P0(t0)|<ε (18)
And repeating the steps until all the sequence points are iterated, wherein the maximum Lyapunov lambda calculation formula is as follows.
Figure BDA0003155600890000041
Wherein T is the number of iterations.
Usually, the calculation is performed by a saturated correlation dimension method (G-P method), when the neighborhood radius r tends to zero, the limit is the correlation dimension, the calculation formula is as follows,
Figure BDA0003155600890000042
Figure BDA0003155600890000043
wherein C (m, r) is the correlation integral; d (m, r) is a cumulative distribution function.
Optionally, the establishing and learning of the vector quantization neural network (LVQ) for performing state classification and identification system model includes:
initializing the network weight wijAnd a learning rate η.
The samples are input. Transmitting the energy characteristic entropy vector of the vibration signal to an input layer of the network, and calculating a neuron of a competition layer and an input vector HijDistance D ofiThe small one is selected as neuron.
Figure BDA0003155600890000044
And correcting the connection weight value. And (4) correcting the weight of the neuron according to whether the network identifies correctly, inputting forward correction if the network identifies correctly, and otherwise, inputting reverse correction, wherein the correction formula is as follows.
Figure BDA0003155600890000051
It is determined whether the loop is terminated. When the iteration times are larger than the set maximum iteration times, the training is terminated, otherwise, the training is continued.
And (5) fault diagnosis of test data. According to the steps, the training model is generated, and then the test data can be input into the model to finish the state diagnosis.
Compared with the prior art, the invention has the following beneficial effects:
(1) the invention researches the chaotic characteristic of an intrinsic mode sequence of an on-load tap changer vibration sequence, provides a characteristic extraction method based on an intrinsic mode function and phase space reconstruction analysis from the aspect of chaotic dynamics analysis, and quantitatively describes the state of a tap changer by extracting phase space characteristics to prove that the chaotic dynamics theory is reasonable for mechanical state analysis;
(2) IMF components obtained through CEEMD analysis are utilized to determine delay time and an embedded dimension reconstruction phase space by utilizing a C-C algorithm, so that the vibration characteristics of the chaotic system can be reserved, and the state information of the tap changer can be well contained;
(3) the LVQ neural network learning efficiency can be self-adjusted, the convergence rate is high, the learning classification effect is good, but the diagnosis time is long, the boundary between weight vectors of different classes can be researched at the gradual convergence rate in the later research, and the diagnosis time and the accuracy rate can possibly rise;
(4) the experimental data of the on-load tap changer researched by the invention come from laboratory simulation and are slightly different from the data acquired by actually-operated equipment, but the analysis method disclosed by the invention has reference significance, so that a new thought is provided for an equipment on-line monitoring and mechanical fault identification method.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
In order to more clearly explain the technical solution of the present application, the following brief description will be given using the accompanying drawings, and it is obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.
Fig. 1 is a flowchart of a mechanical fault diagnosis and analysis method for an on-load tap changer of a voltage regulator transformer based on CEEMD and phase space reconstruction provided by the present application.
Detailed Description
In order to make the technical solutions of the present application better understood, the technical solutions will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that all other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.
Aiming at different propagation speeds of vibration signals with different frequencies and distortion of most of monitored signal waveforms, a novel method for extracting characteristic quantity based on Complementary Ensemble Empirical Mode Decomposition (CEEMD) combined with phase space reconstruction and LVQ neural network diagnosis is adopted in consideration of chaotic characteristics of the vibration signals, and the method has great advantages in identifying mechanical fault types of the on-load tap changer of the on-load tap-changer.
As shown in fig. 1, the method for diagnosing mechanical fault of on-load tap changer of voltage-regulating transformer based on CEEMD and phase space reconstruction provided by the present application includes:
s100: and establishing a complementary modal decomposition and phase space reconstruction model.
S200: and obtaining a vibration signal of the tapping switch of the on-load tap changing transformer. Firstly, collected OLTC vibration signals are sorted, normal switching, switch sliding gears and motor mechanism fault states are selected for classification, 45 groups of data are collected in each running state, 90 groups of data are used for network training, and 45 groups of data are used for testing and analyzing algorithm accuracy.
S300: the signal is decomposed into natural mode functions of different frequency characteristics. The selection of the test data is IMF components obtained by CEEMD decomposition of the vibration signals, the intrinsic mode components of each order are arranged according to frequency after the vibration signals are decomposed, the components contain vibration information with different time characteristic scales, chaotic system information in a mechanical system can be contained in the intrinsic mode components, and then the IMF is used as a chaotic time sequence to carry out phase space reconstruction. Using CEEMD to the original vibration signal xi(t) adding a pair of white noises ni(t) two different new signals P are obtainedi(t) and Ni(t)。
Pi＝x(t)+ni(t) (1)
Ni＝x(t)-ni(t) (2)
EMD decomposition is respectively carried out on the two different new signals to obtain two groups of m integrated IMF components
Figure BDA0003155600890000061
And
Figure BDA0003155600890000062
Figure BDA0003155600890000063
repeating the steps, adding a new normal distribution white noise sequence every time, and obtaining the IMF component c every timej(t) as final result.
Selecting any IMF component time sequence { Yi(t) } constructing the state vector via different delays, the reconstruction phase space can be represented by equation (6).
Ym(n)＝[x(n),x(n+τ),…,x(n+(m-1)τ)] (4)
x(k)＝x(t+kΔt),k＝1,2,…,N (5)
Figure BDA0003155600890000071
Wherein x (k) is a discretized system value at the time k; τ is the delay time; m is the embedding dimension; t is the sampling start time; Δ t is the sampling interval; n is the sample length.
And S400, determining delay time and embedding dimension through a C-C algorithm.
Performing phase space reconstruction on the delay time and the embedding dimension by adopting a C-C algorithm, wherein each order of inherent modal component sequence { Y is subjected to difference according to the delay time taui(t) is divided into τ disjoint time sequences S (m, N, r, τ).
Figure BDA0003155600890000072
Figure BDA0003155600890000073
M＝N-(m-1)τ (9)
dij＝||yi-yj||∞ (10)
Figure BDA0003155600890000074
Wherein d isijIs an infinite function; r is the search radius, and is taken to be less than max (d)ij) Any value of (a); θ (x) is the Heaviside function; c (m, N, r, tau) is the correlation integral of the embedding time series;
selecting the maximum radius rmaxAnd a minimum radius rminThen, the corresponding radius difference Δ S (m, N, r, τ) and the mean test statistic are calculated
Figure BDA0003155600890000075
Average difference Δ S (t) and index Scor(t) and comparing the zero position of S (m, N, r, tau) or the minimum point of Delta S (m, N, r, tau), and recording the smaller value of the two as the time delay tau.
Figure BDA0003155600890000076
Figure BDA0003155600890000077
Figure BDA0003155600890000081
Figure BDA0003155600890000082
Wherein n ismM possible values; n iskAre k possible values.
Drawing
Figure BDA0003155600890000083
And Scor(τ) curve of change, delay time window τwIs at Scor(τ) the delay time corresponding to the global minimum is obtained, and then the embedding dimension m is calculated according to equation (16).
Figure BDA0003155600890000084
Because the redundancy of the reconstruction phase space and the crowdedness degree of the singular attractor track can be changed due to the change of the two important parameters, the delay time and the embedding dimension obtained by the C-C algorithm are optimal values.
S500: and extracting the characteristic quantity of the chaotic characteristic of the vibration signal.
And (3) constructing an m-dimensional phase space new sequence for the sample sequence with the sampling point of N, wherein the calculation formula is (16).
Finding and initiating phase point P0(t0) The closest point is tracked and the distance L between the two points is measured0If greater than epsilon.
L0＝|P(t1)-P0(t0)|>ε (17)
Finding a point and P0(t0) The angle between the phases is smallest and is closest to a point P (t 1).
L'1＝|P(t1)-P0(t0)|<ε (18)
And repeating the steps until all the sequence points are iterated, wherein the maximum Lyapunov lambda calculation formula is as follows.
Figure BDA0003155600890000085
Wherein T is the number of iterations.
Usually, the calculation is performed by a saturated correlation dimension method (G-P method), when the neighborhood radius r tends to zero, the limit is the correlation dimension D, and the calculation formula is as follows,
Figure BDA0003155600890000086
Figure BDA0003155600890000087
wherein C (m, r) is the correlation integral; d (m, r) is a cumulative distribution function.
S600: and establishing and learning a vector quantization neural network (LVQ) to carry out state classification and identification system model.
Initializing the network weight wijAnd a learning rate η.
The samples are input. Transmitting the energy characteristic entropy vector of the vibration signal to an input layer of the network, and calculating a neuron of a competition layer and an input vector HijDistance D ofiThe small one is selected as neuron.
Figure BDA0003155600890000091
And correcting the connection weight value. And (4) correcting the weight of the neuron according to whether the network identifies correctly, inputting forward correction if the network identifies correctly, and otherwise, inputting reverse correction, wherein the correction formula is as follows.
Figure BDA0003155600890000092
It is determined whether the loop is terminated. When the iteration times are larger than the set maximum iteration times, the training is terminated, otherwise, the training is continued.
And (5) fault diagnosis of test data. According to the steps, the training model is generated, and then the test data can be input into the model to finish the state diagnosis.
The application provides an OLTC vibration signal dynamic characteristic extraction method based on CEEMD and phase space reconstruction. The method is characterized in that three common mechanical fault types of OLTC (on-load tap changer) including normal switching, switch sliding gear and motor mechanism faults are simulated through experiments, CEEMD (computer aided empirical mode decomposition) decomposition is carried out on vibration signals, six-order IMF (inertial measurement function) components are generated, embedding dimension and delay time are calculated respectively according to the component signals, then phase space reconstruction is carried out, grid dimension and point distribution factors are calculated to form characteristic vectors, and a learning vector LVQ (linear variable differential Q) neural network is input to carry out state recognition.
Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims. The above-described embodiments of the present application do not limit the scope of the present application.

Claims (6)

1. A CEEMD and phase space reconstruction based mechanical fault diagnosis method for a tap changer of an on-load tap-changer is characterized by comprising the following steps:
establishing a complementary set empirical mode decomposition and phase space reconstruction model;
acquiring a vibration signal of a tap switch of the on-load tap changing transformer;
decomposing the signal into natural mode functions with different frequency characteristics;
determining the delay time and the embedding dimension of the IMFs through a C-C method;
extracting the characteristic quantity of the chaotic characteristic of the vibration signal; and establishing and learning a vector quantization neural network for state classification and identification system model.
2. The CEEMD and phase space reconstruction based on-load tap changer mechanical fault diagnosis method of claim 1, wherein a complementary set empirical mode decomposition and phase space reconstruction model is established, comprising the steps of:
acquiring historical data of vibration data of the on-load tap changer of the voltage regulating transformer;
extracting different frequency components of the historical data of the vibration signal;
and performing phase space reconstruction on the different frequency components, and extracting chaotic characteristic quantity reflecting the phase space.
3. The CEEMD and phase space reconstruction based on-load tap changer mechanical fault diagnosis method of claim 2, characterized in that in the phase space reconstruction of the frequency components,
adding a pair of white noises to original three vibration signals respectively, wherein the original three vibration signals comprise: normal switching, switch sliding gear and motor mechanism failure; to extract the features of the OLTC vibration signal, the original vibration signal x is filtered using CEEMDi(t) adding a pair of white noises ni(t) two different new signals P are obtainedi(t) and Ni(t)，
Pi＝x(t)+ni(t) (1)
Ni＝x(t)-ni(t) (2)
EMD decomposition is respectively carried out on the two different new signals to obtain two groups of m integrated IMF components
Figure FDA0003155600880000011
And
Figure FDA0003155600880000012
Figure FDA0003155600880000013
repeating the steps, adding a new normal distribution white noise sequence every time, and obtaining the IMF component c every timej(t) as a final result;
selecting any IMF component time sequence { Yi(t) } constructing a state vector through different delays, and reconstructing a phase space can be represented by equation (6);
Ym(n)＝[x(n),x(n+τ),…,x(n+(m-1)τ)] (4)
x(k)＝x(t+kΔt),k＝1,2,…,N (5)
Figure FDA0003155600880000021
wherein x (k) is a discretized system value at the time k; τ is the delay time; m is the embedding dimension; t is the sampling start time; Δ t is the sampling interval; n is the sample length.
4. The CEEMD and phase space reconstruction based on-load tap changer mechanical fault diagnosis method of claim 2, characterized in that a C-C algorithm is used for the phase space reconstruction of the delay time and the embedding dimension, wherein,
according to different delay time tau, each order of natural modal component sequence { Yi(t) is divided into tau disjoint time sequences S (m, N, r, tau),
Figure FDA0003155600880000022
Figure FDA0003155600880000023
M＝N-(m-1)τ (9)
dij＝||yi-yj||∞ (10)
Figure FDA0003155600880000024
wherein d isijIs an infinite function; r is the search radius, and is taken to be less than max (d)ij) Any value of (a); θ (x) is the Heaviside function; c (m, N, r, τ))、
Figure FDA0003155600880000025
Integrating the correlation of the embedded time series;
selecting the maximum radius rmaxAnd a minimum radius rminThen, the corresponding radius difference Δ S (m, N, r, τ) and the mean test statistic are calculated
Figure FDA0003155600880000026
Average difference Δ S (t) and index Scor(t) and comparing the zero position of S (m, N, r, tau) or the minimum point of Delta S (m, N, r, tau), the smaller of the two is recorded as the time delay tau,
Figure FDA0003155600880000031
Figure FDA0003155600880000032
Figure FDA0003155600880000033
Figure FDA0003155600880000034
wherein n ismM possible values; n iskK possible values;
drawing
Figure FDA0003155600880000035
And Scor(τ) curve of change, delay time window τwIs at Scor(τ) obtaining a delay time corresponding to the global minimum, calculating an embedding dimension m according to equation (16),
Figure FDA0003155600880000036
because the redundancy of the reconstruction phase space and the crowdedness degree of the singular attractor track can be changed due to the change of the two parameters, the delay time and the embedding dimension obtained by the C-C algorithm are optimal values.
5. The CEEMD and phase space reconstruction based on-load tap changer mechanical fault diagnosis method of claim 4, characterized in that the calculation of the characteristic quantities for extracting chaotic characteristics of the vibration signals comprises the steps of,
constructing a new m-dimensional phase space sequence for the sample sequence with the N sampling points by using a phase space reconstruction technology, wherein a calculation formula is (16);
finding and initiating phase point P0(t0) The closest point is tracked and the distance L between the two points is measured0Whether or not it is greater than epsilon,
L0＝|P(t1)-P0(t0)|>ε (17)
finding a point and P0(t0) Minimum included angle and nearest point P (t)1)，
L'1＝|P(t1)-P0(t0)|<ε (18)
Repeating the steps until all the sequence points are iterated, wherein the maximum Lyapunov lambda calculation formula is as follows,
Figure FDA0003155600880000037
wherein T is the iteration number;
usually, the calculation is performed by a saturated correlation dimension method, when the neighborhood radius r tends to zero, the obtained limit is the correlation dimension D, the calculation formula is as follows,
Figure FDA0003155600880000041
Figure FDA0003155600880000042
wherein C (m, r) is the correlation integral; d (m, r) is a cumulative distribution function.
6. The CEEMD and phase space reconstruction based on-load tap changer mechanical fault diagnosis method of claim 1, characterized in that establishing and learning a vector quantization neural network for state classification recognition system model comprises the following steps,
initializing the network weight wijAnd a learning rate η;
inputting a sample, transmitting the energy characteristic entropy vector of the vibration signal to an input layer of the network, and calculating a neuron of a competition layer and an input vector HijDistance D ofiSelecting the small one as a neuron,
Figure FDA0003155600880000043
correcting the connection weight, correcting the neuron weight according to whether the network identifies correctly, inputting a positive correction if the network identifies correctly, otherwise, inputting a negative correction, wherein the correction formula is as follows,
Figure FDA0003155600880000044
judging whether the loop is terminated, when the iteration times are larger than the set maximum iteration times, terminating the training, otherwise, continuing the training;
and (4) performing fault diagnosis on the test data, generating a training model according to the steps, and inputting the test data into the model to finish state diagnosis.
CN202110779302.3A 2021-07-09 2021-07-09 CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer Pending CN113533952A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110779302.3A CN113533952A (en) 2021-07-09 2021-07-09 CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110779302.3A CN113533952A (en) 2021-07-09 2021-07-09 CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer

Publications (1)

Publication Number Publication Date
CN113533952A true CN113533952A (en) 2021-10-22

Family

ID=78098258

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110779302.3A Pending CN113533952A (en) 2021-07-09 2021-07-09 CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer

Country Status (1)

Country Link
CN (1) CN113533952A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114397521A (en) * 2021-12-24 2022-04-26 中国人民解放军海军航空大学 Fault diagnosis method and system for electronic equipment
CN114447838A (en) * 2022-01-21 2022-05-06 山东电工电气集团有限公司 High-voltage cable partial discharge mode identification method based on small sample
CN116401533A (en) * 2023-06-07 2023-07-07 中国南方电网有限责任公司超高压输电公司广州局 Fault diagnosis method and device for on-load tap-changer
CN117878971A (en) * 2024-03-12 2024-04-12 西安热工研究院有限公司 Novel frequency modulation method and system for fused salt energy storage coupling thermal power generating unit

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
周翔等: "基于混沌理论和K-means聚类的有载分接开关机械状态监测", 《中国电机工程学报》 *
张丽萍等: "低压断路器振动特性分析及其故障诊断研究", 《电机与控制学报》 *
张新广等: "经验模态分解和关联维数在机械故障诊断中的应用研究", 《机床与液压》 *
王斌等: "基于CEEMD-MPE和ELM的齿轮箱故障诊断研究", 《组合机床与自动化加工技术》 *
陆金铭等: "基于经验模式分解与关联维数的柴油机故障诊断", 《船海工程》 *
高树国等: "应用改进Hilbert-Huang变换下的Volterra模型诊断OLTC机械故障", 《高压电器》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114397521A (en) * 2021-12-24 2022-04-26 中国人民解放军海军航空大学 Fault diagnosis method and system for electronic equipment
CN114447838A (en) * 2022-01-21 2022-05-06 山东电工电气集团有限公司 High-voltage cable partial discharge mode identification method based on small sample
CN116401533A (en) * 2023-06-07 2023-07-07 中国南方电网有限责任公司超高压输电公司广州局 Fault diagnosis method and device for on-load tap-changer
CN116401533B (en) * 2023-06-07 2024-04-02 中国南方电网有限责任公司超高压输电公司广州局 Fault diagnosis method and device for on-load tap-changer
CN117878971A (en) * 2024-03-12 2024-04-12 西安热工研究院有限公司 Novel frequency modulation method and system for fused salt energy storage coupling thermal power generating unit

Similar Documents

Publication Publication Date Title
CN113533952A (en) CEEMD and phase space reconstruction-based mechanical fault diagnosis method for tap changer of on-load tap-changing transformer
Ali et al. Machine learning-based fault diagnosis for single-and multi-faults in induction motors using measured stator currents and vibration signals
CN111307453A (en) Transmission system fault diagnosis method based on multi-information fusion
CN108556682B (en) Driving range prediction method, device and equipment
CN112415414A (en) Method for predicting remaining service life of lithium ion battery
CN110705456A (en) Micro motor abnormity detection method based on transfer learning
CN113469470B (en) Energy consumption data and carbon emission correlation analysis method based on electric brain center
CN110726898A (en) Power distribution network fault type identification method
CN111856209A (en) Power transmission line fault classification method and device
CN111060652A (en) Method for predicting concentration of dissolved gas in transformer oil based on long-term and short-term memory network
CN110962828B (en) Method and equipment for predicting brake pressure of electric automobile
CN113889198A (en) Transformer fault diagnosis method and equipment based on oil chromatogram time-frequency domain information and residual error attention network
CN117074961A (en) Variation self-encoder power battery abnormality detection method, system, device and storage medium
CN115758082A (en) Fault diagnosis method for rail transit transformer
CN115130514A (en) Construction method and system for health index of engineering equipment
CN115586444A (en) Lithium battery residual life prediction method based on VMD and BP neural network
CN115166514A (en) Motor fault identification method and system based on self-adaptive spectrum segmentation and denoising
CN113655778B (en) Underwater propeller fault diagnosis system and method based on time-frequency energy
CN115809716A (en) Flight path prediction method and device based on multi-resolution analysis and wavelet packet reconstruction
CN114046903A (en) Real-time temperature prediction method and system for high-voltage cable core
CN114580262A (en) Lithium ion battery health state estimation method
CN117406100A (en) Lithium ion battery remaining life prediction method and system
CN112151067A (en) Passive detection method for digital audio tampering based on convolutional neural network
CN112132328A (en) Photovoltaic output power ultra-short-term local emotion reconstruction neural network prediction method
Li et al. Fault diagnosis of ZD6 turnout system based on wavelet transform and GAPSO-FCM

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: 20211022

RJ01 Rejection of invention patent application after publication