CN115047305A - Inverter open-circuit fault identification method based on signal processing reconstruction - Google Patents
Inverter open-circuit fault identification method based on signal processing reconstruction Download PDFInfo
- Publication number
- CN115047305A CN115047305A CN202210513910.4A CN202210513910A CN115047305A CN 115047305 A CN115047305 A CN 115047305A CN 202210513910 A CN202210513910 A CN 202210513910A CN 115047305 A CN115047305 A CN 115047305A
- Authority
- CN
- China
- Prior art keywords
- fault
- open
- elm
- circuit fault
- woa
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000012545 processing Methods 0.000 title claims description 23
- 241000283153 Cetacea Species 0.000 claims abstract description 22
- 239000013598 vector Substances 0.000 claims abstract description 20
- 238000005457 optimization Methods 0.000 claims abstract description 10
- 238000005070 sampling Methods 0.000 claims description 33
- 210000002569 neuron Anatomy 0.000 claims description 18
- 238000004364 calculation method Methods 0.000 claims description 15
- 238000000354 decomposition reaction Methods 0.000 claims description 15
- 238000000819 phase cycle Methods 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 7
- 230000007547 defect Effects 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000003780 insertion Methods 0.000 claims description 3
- 230000037431 insertion Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 150000001875 compounds Chemical class 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 5
- 238000003745 diagnosis Methods 0.000 description 8
- 238000010276 construction Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 229910052739 hydrogen Inorganic materials 0.000 description 2
- 239000001257 hydrogen Substances 0.000 description 2
- 125000004435 hydrogen atom Chemical class [H]* 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
- G01R31/26—Testing of individual semiconductor devices
- G01R31/2601—Apparatus or methods therefor
-
- 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
- G01R31/26—Testing of individual semiconductor devices
- G01R31/2607—Circuits therefor
- G01R31/2608—Circuits therefor for testing bipolar transistors
-
- 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
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/54—Testing for continuity
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Inverter Devices (AREA)
- Locating Faults (AREA)
Abstract
The invention provides an online identification method for open-circuit faults of a three-phase two-level inverter based on WOA-ELM, and relates to the technical field of fault detection of three-phase inverters. Constructing a multi-parameter fused six-dimensional open-circuit fault feature vector, constructing an Extreme Learning Machine (ELM) model of the open-circuit fault of the power tube by taking the feature vector as input and the fault mode number as output, and optimizing model parameters by applying a Whale Optimization Algorithm (WOA); embedding an optimal WOA-ELM inverter open-circuit fault identification model into a control link of the optimal WOA-ELM inverter open-circuit fault identification model, acquiring three-phase current in real time, extracting fault characteristic quantity to serve as input of the optimal WOA-ELM model, and outputting a fault mode number; the method converts the fault mode number into a binary number to realize real-time alarm of the fault; the total time of fault identification of the method is about 2.97 ms.
Description
Technical Field
The technical scheme of the invention relates to the technical field of open-circuit fault detection of a three-phase two-level inverter power tube, in particular to an online identification method for open-circuit faults of the inverter power tube based on a multi-parameter fusion six-dimensional fault characteristic quantity extraction and WOA-ELM fault identification model.
Background
In recent years, inverters are widely used in the fields of new energy power generation, aviation special power supplies and the like to realize electric energy exchange. However, as the device operates in a high-frequency, high-voltage and high-current operating state for a long time, the performance of the IGBT in the inverter will be reduced, and a fault is very likely to occur. If the diagnosis cannot be carried out in time, the equipment can not work normally, and immeasurable economic loss and potential safety hazard can be caused. Therefore, as an important component of the inverter system, it is necessary to accurately locate the inverter fault in time.
In order to improve the reliability of the inverter, research on related open-circuit fault diagnosis technologies has received extensive attention. The invention patent with the patent number of CN201510367945.1 discloses an inverter fault diagnosis method based on wavelet analysis and SVM, which comprises the steps of collecting three-phase voltage at the output side of an alternating current side, respectively extracting energy from each frequency band signal after wavelet transformation, determining fault characteristic vectors, and finally classifying faults by using the SVM. When the voltage sensor is used for implementation, the voltage sensor is required to be added, and the system cost is increased.
In summary, certain research results have been obtained in the identification of open-circuit faults of inverters, but some problems still need to be solved. First, in practical applications, the system usually does not allow a detection circuit to be added to the inverter, so as to avoid generating electromagnetic interference and affecting the normal operation of the system. Second, if the load changes, the current also changes, and if the influence of the current is not considered, the failure diagnosis result is affected, and the reliability is lowered.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the method for online diagnosing the open-circuit fault of the inverter is provided, and real-time alarm of the open-circuit fault of the power tube of the inverter is realized.
The technical scheme adopted by the invention is as follows:
(1) signal reconstruction method based on symmetry processing reconstructs three-phase current at alternating current side
And reconstructing output signals of the inverter in normal operation and various fault states by adopting an inverter fault signal reconstruction method considering load fluctuation. Firstly, carrying out symmetry analysis and non-zero verification on sampled current data, wherein a symmetry index lambda and a non-zero index sigma are expressed as follows:
in the formula, psi represents phase sequence, namely three phases of a, b and c; n is the total number of sampling points in a sampling window, and the sampling time length is set to be T; τ is the fundamental period (50 ms); n is a radical of hydrogen f The number of sampling points in a fundamental frequency period tau is shown, and T is an integral multiple of tau for convenient calculation; k denotes the kth fundamental frequency period within the sampling duration T, then k ∈ [1, T/τ]。
Setting a threshold value mu, if lambda is smaller than the threshold value, the signals are symmetrical, and adopting a formula (2) to process the signals; if λ is larger than the threshold, the signal is asymmetric, and the signal processing is performed by using equations (3) to (7). If sigma is equal to 0, the phase current is 0, and at the moment, the upper tube and the lower tube of the phase bridge arm are simultaneously opened, and the 0 value is maintained when the signal is reconstructed. Finally outputting a complete reconstruction signal I new . The specific signal processing steps are as follows:
1) inputting a set of sample data I ψ ＝(I 1 ,I 2 ,…,I N ) And measuring the symmetry of the signal according to the formula (1), if the lambda is smaller than the threshold value rho, the signal is symmetrical, and acquiring the reconstructed signal according to the formula (2).
In the formula I k (i) Expressed as the ith data in the kth fundamental frequency period in the sampling window, k e [1, T/τ](ii) a m is the maximum value in the sampling window, i.e. m equals max (I) 1 ,I 2 ,…,I N ) (ii) a s is the minimum value in the sampling window, i.e. s ═ min (I) 1 ,I 2 ,…,I N )；I kmax Is the maximum value in the k-th fundamental frequency period, i.e. I kmax ＝max[I k (i)]，i∈[1,N f ]；I kmin Is the minimum value in the k-th fundamental frequency period, i.e. I kmin ＝min[I k (i)]. And if the lambda is smaller than the threshold value rho, carrying out subsequent step processing.
2) If lambda is smaller than the threshold value rho, the signals are asymmetric, and the maximum absolute value of the sampling data in the kth fundamental frequency period is calculated to be I' kmax L's' kmin ＝-I’ kmax Then the preliminary reconstruction signal I 'can be obtained according to equations (3) - (5)' inew 。
I' kmax ＝max(|I k,i |) (3)
I' kmin ＝-I' kmax (4)
3) Calculating raw data and preliminary reconstruction signal I' inew The coefficient of proportionality c (i) therebetween.
4) Combining the symmetric part and the asymmetric part in the signal, and obtaining the finally processed signal I according to the formula (7) new 。
(2) Constructing six-dimensional open-circuit fault feature vector fused by multiple parameters
Improvement of wavelet packet algorithm
In order to eliminate the frequency aliasing defect caused by the wavelet packet transformation inherent algorithm, two operators C and D capable of deleting redundant information in a frequency band are introduced. Wherein the operator C aims to zero the redundant components after convolution of the wavelet decomposition low-frequency filter H and the reconstruction low-frequency filter H. Let x be represented as a wavelet packet low frequency coefficient,
for the outputs of operators C and D, the expression of operator C is:
in the formula (8), N j The data length of the current wavelet packet decomposition layer j is obtained; n is 0,1, N j-1 ；k＝0,1,···,N j-1 (ii) a Operator
The operator D aims at setting the redundant components after the convolution of the wavelet decomposition high-frequency filter G and the reconstruction high-frequency filter G to be zero, and the expression is as follows:
in the improved wavelet packet calculation process, an C, D operator eliminates the frequency aliasing defect through Fourier transform and inverse transform thereof, and combines alternate zero insertion and alternate sampling.
II, open-circuit fault characteristic based on low-frequency coefficient mean value
Initial input signal is converted by using improved wavelet packet based on Mallat algorithm
Decomposing and reconstructing three layers of wavelet packets, wherein the reconstruction nodes are [3,1 ] in sequence]、[3,2]、[3,3]、[3,4]、[3,5]、[3,6]、[3,7]、[3,8]The reconstructed sub-signals are in turn
The reconstructed node signals and the initial input signals satisfy the following relations:
selecting [3,1]And the node (low-frequency node) is used as a main node for extracting fault characteristics. Selecting the mean value of wavelet packet reconstruction coefficients of the main node as one of fault characteristic values, and recording as M φ And phi denotes a phase sequence. And to M φ Normalization processing is carried out, and the calculation formula is as follows:
in the formula, M i * Is the normalized low frequency coefficient mean, | max (M) i ) And | is the maximum value of the mean absolute value of the three-phase low-frequency coefficient.
Open circuit fault characteristics based on wavelet entropy
According to the basic theory of information entropy, the wavelet packet information entropy H is defined as:
selecting another fault characteristic value of each phase of current wavelet packet information entropy, and recording as H φ And phi denotes a phase sequence.
To H φ Normalization processing is carried out, and the calculation formula is as follows:
IV, multi-parameter fused six-dimensional open-circuit fault feature vector
Constructing a six-dimensional open-circuit fault feature vector with multi-parameter fusion as follows:
x＝[M a ,M b ,M c ,H a ,H b ,H c ] (14)
the characteristic quantity is used for carrying out comparison on three-phase current i according to an improved wavelet packet algorithm a 、i b 、i c The inverter is obtained by decomposition and reconstruction, when the inverter is in 1 non-fault state and 21 single-double tube open-circuit fault states, the inverter is different in pairs and has clear identification.
(3) Inverter power tube open-circuit fault ELM identification model construction
For three-phase current i according to improved wavelet packet algorithm a 、i b 、i c Six-dimensional fault characteristic quantity M obtained by decomposition and reconstruction a 、M b 、M c 、H a 、H b 、H c Taking six-dimensional fault characteristic quantity as input layer neuron x of ELM identification model i (i-1, …,6) and the input column vector is x-x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6] T 。
Constructing a hidden layer with l neurons, where b 1 ,…,b l A threshold value for a corresponding neuron within the hidden layer; numbering the failure modes y j J (j-1, …,21) as output layer neuron of ELM recognition model, and for hidden layer neuron g l (L ═ 1, …, L) weight β jl (j-1, …, 21; L-1, …, L) form output layer neuron y j 。
The ELM recognition model of the open-circuit fault of the inverter power tube can be expressed as:
(4) parameter optimization training for inverter power tube open-circuit fault ELM identification model based on WOA algorithm
For the optimal power tube open-circuit fault ELM identification model parameter matrix (w, b, beta), a WOA algorithm is implanted into the ELM model.
The Whale Optimization Algorithm (WOA) is a novel meta-heuristic optimization algorithm proposed by Mirjalli and Lewis of the university of Grifys, Australia in 2016, and has the advantages of few parameters, simplicity in calculation and the like. WOA simulates hunting behavior of whales in the ocean, where the location of each whale represents a basic optimal solution.
The WOA-ELM identification model parameter optimization specific flow of the power tube open-circuit fault is as follows:
1) and processing the sampled data. And performing signal reconstruction on the sampled data, and extracting to obtain an open-circuit fault characteristic vector.
2) And initializing parameters. Initializing whale population number S in WOA algorithm N And maximum number of iterations T of the algorithm max Initializing whale colony positions X (w, b) in the WOA algorithm, namely initial weight w and threshold b in the ELM algorithm.
3) And calculating the fitness. And calculating the fitness of each whale in each iteration by taking the algorithm classification error rate as the fitness value, and selecting the individual with the minimum fitness as the optimal individual. The fitness function is:
F(X(t))＝(1-k/N)×100 (3.29)
in the formula, N is the total amount of the samples; k represents the number of correct algorithm classifications. As can be seen from the function, the closer the individual fitness value is to 0, the more the individual is qualified.
4) And (4) performing iterative updating. And updating the position of the next generation according to rules of surrounding prey, hunting and searching prey by the algorithm.
5) And determining the optimal. And when the algorithm reaches the iteration times or the positions of the optimal whales in the two iterations are the same, finishing the iteration and outputting the optimal individuals, namely the optimal initial w and b.
6) And training the ELM identification model according to the obtained optimal initial w and b to obtain an ELM optimal identification model parameter matrix (w, b, beta).
(5) WOA-ELM-based open-circuit fault online identification and fault alarm
Embedding an optimal WOA-ELM inverter power tube open-circuit fault identification model into a control link of the optimal WOA-ELM inverter power tube open-circuit fault identification model, and detecting a three-phase current i by using a built-in current sensor a 、i b And i c And calculating six-dimensional fault characteristic quantity, and using the six-dimensional fault characteristic quantity as the optimal inverter power tubeThe open fault WOA-ELM identifies the inputs to the model.
And converting the fault mode number output by the 2-bit decimal system of the WOA-ELM identification model of the open-circuit fault of the inverter power tube into a 6-bit binary fault mode number to realize real-time alarm of the open-circuit fault of the inverter power tube.
Compared with the prior art, the invention has the following prominent substantive characteristics and remarkable progress:
(1) the fault identification method provided by the invention can accurately position the open-circuit faults of the single and double open tubes of the 21 inverters, has high diagnosis speed and high reliability, and can provide a basis for fault tolerance of the system in the follow-up process.
(2) No extra detection equipment is needed, and the cost is low. Detection of three-phase currents i by means of built-in current sensors a 、i b And i c The fault diagnosis is carried out, and the cost is low.
(3) A WOA-ELM identification model of the open-circuit fault of the power tube is constructed, and the WOA is used for optimizing ELM parameters. Embedding an optimal WOA-ELM inverter power tube open-circuit fault identification model into a control link of the inverter power tube open-circuit fault identification model, and acquiring three-phase current i from a built-in current sensor in real time a 、i b And i c The measured value realizes online fault identification, and the identification accuracy can reach 98.32%.
(4) According to the method, the fault mode number output by the 2-bit decimal system of the WOA-ELM inverter power tube open-circuit fault identification model is converted into the 6-bit binary fault mode number, and real-time alarm of the inverter power tube open-circuit fault can be realized.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts. The invention is further illustrated by the following examples in conjunction with the drawings.
Fig. 1 shows the ac-side three-phase current after signal reconstruction.
FIG. 2 is a WOA-ELM identification model of power tube open circuit fault.
FIG. 3 is a fitness curve for a WOA algorithm optimized ELM.
FIG. 4 is a WOA-ELM algorithm identification time curve.
Fig. 5 is a normalized low frequency coefficient mean of 22 open circuit failure modes.
FIG. 6 is normalized wavelet packet entropy values for 22 open circuit failure modes.
FIG. 7 is a WOA-ELM fault diagnosis based on measured data.
The specific implementation mode is as follows:
the invention is further described below with reference to the accompanying drawings and the detailed description.
The inverter fault is first classified. The probability of the simultaneous failure of a plurality of power tubes of the inverter is extremely low, and at most 2 power tubes can be assumed to have simultaneous failure, so that four failure modes exist: i. single tube open circuit (e.g. T) 1 Or T 2 Open circuit, etc.); in-phase two-tube open circuit (e.g. T) 1 、T 4 Open circuit, etc.); III, two pipes on the same side are open (such as T) 1 、T 3 Open circuit, etc.); IV, two tubes out of phase at different sides are open (such as T) 1 、T 6 Open circuit, etc.).
TABLE 1 inverter Power tube open-Circuit Fault numbering
The first step is as follows: inverter fault signal reconstruction taking into account load fluctuations
And reconstructing output signals of the inverter in normal operation and various fault states by adopting an inverter fault signal reconstruction method considering load fluctuation. Firstly, carrying out symmetry analysis and non-zero verification on sampled current data, wherein a symmetry index lambda and a non-zero index sigma are expressed as follows:
in the formula, psi tableShowing phase sequence, namely three phases of a, b and c; n is the total number of sampling points in a sampling window, and the sampling time length is set to be T; τ is the fundamental period (50 ms); n is a radical of hydrogen f The number of sampling points in a fundamental frequency period tau is shown, and T is an integral multiple of tau for convenient calculation; k represents the kth fundamental frequency period in the sampling time length T, and k belongs to [1, T/tau ]]。
Setting a threshold value mu, if lambda is smaller than the threshold value, the signals are symmetrical, and adopting a formula (2) to process the signals; if λ is larger than the threshold, the signal is asymmetric, and the signal processing is performed by using equations (3) to (7). If sigma is equal to 0, the phase current is 0, and at the moment, the upper tube and the lower tube of the phase bridge arm are simultaneously opened, and the 0 value is maintained when the signal is reconstructed. Finally outputting a complete reconstruction signal I new . The specific signal processing steps are as follows:
(1) inputting a set of sampling data I ψ ＝(I 1 ,I 2 ,…,I N ) And measuring the symmetry of the signal according to the formula (1), if the lambda is smaller than the threshold value rho, the signal is symmetrical, and acquiring the reconstructed signal according to the formula (2).
In the formula I k (i) Expressed as the ith data in the kth fundamental frequency period in the sampling window, k e [1, T/τ](ii) a m is the maximum value within the sampling window, i.e. m is max (I) 1 ,I 2 ,…,I N ) (ii) a s is the minimum value within the sampling window, i.e. s ═ min (I) 1 ,I 2 ,…,I N )；I kmax Is the maximum value in the k-th fundamental frequency period, i.e. I kmax ＝max[I k (i)]，i∈[1,N f ]；I kmin Is the minimum value in the k-th fundamental frequency period, i.e. I kmin ＝min[I k (i)]. And if the lambda is smaller than the threshold value rho, carrying out subsequent step processing.
(2) If lambda is smaller than the threshold value rho, the signal is asymmetric, and the maximum value of the absolute value of the sampling data in the k-th fundamental frequency period is calculated to be I' kmax L of' kmin ＝-I’ kmax Then the preliminary reconstruction signal I 'can be obtained according to equations (3) - (5)' inew 。
I' kmax ＝max(|I k,i |) (3)
I' kmin ＝-I' kmax (4)
(3) Calculating raw data and preliminary reconstruction signal I' inew C (i) is a proportionality coefficient.
(4) Combining the symmetric part and the asymmetric part in the signal, and obtaining the finally processed signal I according to the formula (3.7) new 。
The second step is that: extraction of six-dimensional open-circuit fault characteristic quantity of inverter power tube
Three-phase AC side current i after single-tube and double-tube open circuit fault of inverter a 、i b And i c The method has the advantages that the method contains characteristic information of the fault of the switching tube, the mutation speed is high, time domain characteristic parameters such as kurtosis and pulse indexes of the switching tube contain fault information, and fault characteristic quantities which are beneficial to improving the rapidity and accuracy of fault identification can be extracted from the time domain characteristic parameters.
(1) Improvement of wavelet packet algorithm
In order to eliminate the frequency aliasing defect caused by the wavelet packet transformation inherent algorithm, two operators C and D capable of deleting redundant information in a frequency band are introduced. Wherein the operator C aims at zeroing the redundant components after convolution of the wavelet decomposition low-frequency filter H and the reconstruction low-frequency filter H. Let x be represented as the wavelet packet low frequency coefficient,
for the outputs of operators C and D, the expression of operator C is:
in the formula (8), N j The data length of the current wavelet packet decomposition layer j is obtained; n is 0,1, N j-1 ；k＝0,1,···,N j-1 (ii) a Operator
The operator D aims at setting the redundant components after the convolution of the wavelet decomposition high-frequency filter G and the reconstruction high-frequency filter G to be zero, and the expression is as follows:
in the improved wavelet packet calculation process, the C, D operator eliminates the frequency aliasing defect through Fourier transform and inverse transform thereof, and combines alternate zero insertion and alternate sampling.
(2) Open-circuit fault characteristics based on low-frequency coefficient mean value
Initial input signal is converted by using improved wavelet packet based on Mallat algorithm
Decomposing and reconstructing three layers of wavelet packets, wherein the reconstruction nodes are [3,1 ] in sequence]、[3,2]、[3,3]、[3,4]、[3,5]、[3,6]、[3,7]、[3,8]The reconstructed sub-signals are in turn
The reconstructed node signals and the initial input signals satisfy the following relations:
selecting [3,1]The node (low-frequency node) is used as a main node for extracting fault characteristics. Selecting the mean value of wavelet packet reconstruction coefficients of the main node as one of fault characteristic values, and recording as M φ And phi denotes a phase sequence. And to M φ Normalization processing is carried out, and the calculation formula is as follows:
in the formula, M i * Is the normalized low frequency coefficient mean, | max (M) i ) And | is the maximum value of the mean absolute value of the three-phase low-frequency coefficient.
(3) Open-circuit fault characteristics based on wavelet entropy
According to the basic theory of information entropy, the wavelet packet information entropy H is defined as:
selecting another fault characteristic value of each phase of current wavelet packet information entropy, and recording as H φ And phi denotes a phase sequence.
To H φ Normalization processing is carried out, and the calculation formula is as follows:
(4) multi-parameter fused six-dimensional open-circuit fault feature vector
Constructing a multi-parameter fused six-dimensional open-circuit fault feature vector as follows:
x＝[M a ,M b ,M c ,H a ,H b ,H c ] (14)
the characteristic quantity is used for carrying out comparison on three-phase current i according to an improved wavelet packet algorithm a 、i b 、i c The inverter is obtained by decomposition and reconstruction, when the inverter is in 1 non-fault state and 21 single-double tube open-circuit fault states, the inverter is different in pairs and has clear identification.
The third step: inverter power tube open-circuit fault ELM identification model construction
For three-phase current i according to improved wavelet packet algorithm a 、i b 、i c Six-dimensional fault characteristic quantity M obtained by decomposition and reconstruction a 、M b 、M c 、H a 、H b 、H c Taking six-dimensional fault characteristic quantity as input layer neuron x of ELM identification model i (i-1, …,6) and the input column vector is x-x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6] T 。
Constructing a hidden layer with l neurons, where b 1 ,…,b l A threshold value for a corresponding neuron within the hidden layer; numbering the failure modes y j J (j ═ 1, …,21) as output layer neurons of the ELM recognition model, for hidden layer neurons g l (L ═ 1, …, L) weight β jl (j-1, …, 21; L-1, …, L) form output layer neuron y j 。
The ELM recognition model of the open-circuit fault of the inverter power tube can be expressed as:
the fourth step: WOA-ELM-based open-circuit fault online identification and fault alarm
And (4) implanting a WOA algorithm into the ELM model to preferably select an ELM identification model parameter matrix (w, b, beta) of the open-circuit fault of the power tube.
The Whale Optimization Algorithm (WOA) is a novel meta-heuristic optimization algorithm proposed by Mirjalili and Lewis of the university of Gregorphis Australia in 2016, and has the advantages of few parameters, simplicity in calculation and the like. WOA simulates hunting behavior of whales in the ocean, where the location of each whale represents a basic optimal solution.
The specific process for optimizing the WOA-ELM identification model parameters of the power tube open-circuit fault comprises the following steps:
(1) and processing the sampling data. And performing signal reconstruction on the sampled data, and extracting to obtain an open-circuit fault characteristic vector.
(2) And initializing parameters. Initializing whale population number S in WOA algorithm N And the maximum number of iterations T of the algorithm max Initializing whale colony positions X (w, b) in the WOA algorithm, namely initial weight w and threshold b in the ELM algorithm.
(3) And calculating the fitness. And calculating the fitness of each whale in each iteration by taking the algorithm classification error rate as the fitness value, and selecting the individual with the minimum fitness as the optimal individual. The fitness function is:
F(X(t))＝(1-k/N)×100 (3.29)
in the formula, N is the total amount of the sample; k represents the number of correct algorithm classifications. From this function, it can be seen that the closer the individual fitness value is to 0, the more the individual is in line with the requirement.
(4) And (4) performing iterative updating. And updating the position of the next generation according to rules of surrounding prey, hunting and searching prey by the algorithm.
(5) And determining the optimal. And when the algorithm reaches the iteration times or the positions of the optimal whales in the two iterations are the same, finishing the iteration and outputting the optimal individuals, namely the optimal initial w and b.
(6) And training the ELM identification model according to the obtained optimal initial w and b to obtain an ELM optimal identification model parameter matrix (w, b, beta).
Setting WOA algorithm parameters: initial population number S N 20; maximum number of iterations T max 100; the upper limit and the lower limit of ELM weight w to be optimized are [ -1,1](ii) a The upper and lower limits of the threshold b are [0,1 ]]. And (4) selecting 660(22 x 30) groups of training samples and 110(22 x 5) groups of test samples to input the ELM fault diagnosis model. To minimize the influence of random factors, the algorithm was executed 100 times, and the average of the accuracy of 100 fault diagnoses was calculated.
When six-dimensional fault feature vector training is adopted, the fitness curve of the optimized ELM by the WOA algorithm and the identification time curve of a single fault sample after the algorithm is executed 100 times are respectively shown in fig. 3 and 4. As can be seen from fig. 3, the fitness function value decreases with increasing number of iterations, and the best fitness value stabilizes after approximately 20 iterations. As can be seen from FIG. 4, the average recognition time of the trained WOA-ELM model for a single fault sample is 2.97 × 10 -5 s。
The fifth step: WOA-ELM-based open-circuit fault online identification and fault alarm
In order to verify the effectiveness of the WOA-ELM open-circuit fault identification model, an inversion experiment platform based on DSP28335 is built. The experiment is based on the measured data of the three-phase current after the fault (the fault time is random), the low-frequency coefficient mean value and the wavelet entropy value are extracted, and the fault identification is carried out by using the trained WOA-ELM model. The six-dimensional eigenvector distributions and partial test results for the 21 open-circuit faults are shown in fig. 5, fig. 6, and table 2, respectively. The result of the fault recognition is shown in fig. 7.
And finally, designing an open-circuit fault alarm device for the inverter power tube. The invention provides a method for converting a 2-bit decimal output fault mode number of a WOA-ELM inverter power tube open-circuit fault identification model into a 6-bit binary fault mode number, and realizing real-time alarm of the inverter power tube open-circuit fault.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
TABLE 2 comparison of partial input data and output results based on experimental data
Claims (8)
1. An inverter power tube open-circuit fault identification method based on WOA-ELM is characterized in that: the signal reconstruction method based on symmetry processing reconstructs three-phase current, an improved wavelet packet algorithm is used for extracting a wavelet entropy and a low-frequency coefficient mean value, a multi-parameter fused six-dimensional open-circuit fault feature vector is constructed, the constructed six-dimensional open-circuit fault feature vector is used as input, a fault mode number is used as output, an open-circuit fault WOA-ELM identification model of an inverter power tube is designed, a fault alarm module is designed, and the total fault identification time is about 2.97 ms.
2. The method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein: a signal reconstruction method for symmetry processing.
Reconstructing output signals of the inverter in normal operation and various fault states by adopting an inverter fault signal reconstruction method considering load fluctuation; firstly, carrying out symmetry analysis and non-zero verification on sampled current data, wherein a symmetry index lambda and a non-zero index sigma are expressed as follows:
in the formula, psi represents phase sequence, namely three phases of a, b and c; n is the total number of sampling points in a sampling window, and the sampling time length is set to be T; τ is the fundamental period (50 ms); n is a radical of f The number of sampling points in a fundamental frequency period tau is shown, and T is an integral multiple of tau for convenient calculation; k denotes the kth fundamental frequency period within the sampling duration T, then k ∈ [1, T/τ]；
Setting a threshold value mu, if lambda is smaller than the threshold value, the signals are symmetrical, and adopting a formula (2) to process the signals; if lambda is larger than the threshold value, the signal is asymmetric, and the formulas (3) to (7) are adopted for signal processing; if sigma is equal to 0, the phase current is 0, the upper tube and the lower tube of the bridge arm of the phase are simultaneously opened, and the value of 0 is maintained unchanged when the signal is reconstructed; finally outputting a complete reconstruction signal I new (ii) a The specific signal processing steps are as follows:
(1) inputting a set of sample data I ψ ＝(I 1 ,I 2 ,…,I N ) Measuring the signal symmetry according to a formula (1), if lambda is smaller than a threshold value rho, the signal is symmetrical, and acquiring a reconstructed signal according to a formula (2);
in the formula I k (i) Expressed as the ith data in the kth fundamental frequency period in the sampling window, k e 1,T/τ](ii) a m is the maximum value within the sampling window, i.e. m is max (I) 1 ,I 2 ,…,I N ) (ii) a s is the minimum value within the sampling window, i.e. s ═ min (I) 1 ,I 2 ,…,I N )；I kmax Is the maximum value in the k-th fundamental frequency period, i.e. I kmax ＝max[I k (i)]，i∈[1,N f ]；I kmin Is the minimum value in the k-th fundamental frequency period, i.e. I kmin ＝min[I k (i)](ii) a If the lambda is smaller than the threshold rho, carrying out subsequent step processing;
(2) if lambda is smaller than the threshold value rho, the signal is asymmetric, and the maximum value of the absolute value of the sampling data in the k-th fundamental frequency period is calculated to be I' kmax L of' kmin ＝-I’ kmax Then the preliminary reconstruction signal I 'can be obtained according to equations (3) - (5)' inew ；
I' kmax ＝max(|I k,i |) (3)
I' kmin ＝-I' kmax (4)
(3) Calculating raw data and preliminary reconstruction signal I' inew A proportionality coefficient of C (i);
(4) combining the symmetric part and the asymmetric part in the signal, and obtaining the finally processed signal I according to the formula (7) new ；
3. The method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein the method comprises the following steps: and constructing a multi-parameter fused six-dimensional fault feature vector.
(1) Improvement of wavelet packet algorithm
In order to eliminate the frequency aliasing defect caused by the wavelet packet transformation inherent algorithm, two operators C and D capable of deleting redundant information in a frequency band are introduced; the operator C aims to zero the redundant components formed by convolution of the wavelet decomposition low-frequency filter H and the reconstruction low-frequency filter H; let x be represented as a wavelet packet low frequency coefficient,
for the outputs of operators C and D, the expression of operator C is:
in the formula (8), N j The data length of the current wavelet packet decomposition layer j is obtained; n is 0,1, N j-1 ；k＝0,1,···,N j-1 (ii) a Operator V ═ e -j2π/Nj ；
The operator D aims at setting the redundant components after the convolution of the wavelet decomposition high-frequency filter G and the reconstruction high-frequency filter G to be zero, and the expression is as follows:
in the improved wavelet packet calculation process, an C, D operator eliminates the frequency aliasing defect through Fourier transform and inverse transform thereof, and combining alternate point zero insertion and alternate point sampling;
(2) open-circuit fault characteristics based on low-frequency coefficient mean value
Initial input signal is converted by using improved wavelet packet based on Mallat algorithm
Decomposing and reconstructing three layers of wavelet packets, wherein the reconstruction nodes are [3,1 ] in sequence]、[3,2]、[3,3]、[3,4]、[3,5]、[3,6]、[3,7]、[3,8]The reconstructed sub-signals are in turn
The reconstructed node signals and the initial input signals satisfy the following relations:
selecting [3,1 ]]Taking a node (low-frequency node) as a main node to extract fault characteristics; selecting the mean value of wavelet packet reconstruction coefficients of the main node as one of fault characteristic values, and recording as M φ Phi denotes a phase sequence; and to M φ Normalization processing is carried out, and the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
is the normalized low frequency coefficient mean, | max (M) i ) I is the maximum value of the mean absolute value of the three-phase low-frequency coefficient;
(3) open-circuit fault characteristics based on wavelet entropy
According to the basic theory of information entropy, the wavelet packet information entropy H is defined as:
selecting another fault characteristic value of each phase of current wavelet packet information entropy, and recording as H φ Phi denotes a phase sequence;
to H φ Normalization processing is carried out, and the calculation formula is as follows:
(4) multi-parameter fused six-dimensional open-circuit fault feature vector
Constructing a multi-parameter fused six-dimensional open-circuit fault feature vector as follows:
x＝[M a ,M b ,M c ,H a ,H b ,H c ] (14)
the characteristic quantity is used for carrying out comparison on three-phase current i according to an improved wavelet packet algorithm a 、i b 、i c The inverter is obtained by decomposition and reconstruction, when the inverter is in 1 non-fault state and 21 single-double tube open-circuit fault states, the inverter is different in pairs and has clear identification.
4. The method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein the method comprises the following steps: and (3) constructing an ELM identification model of the open-circuit fault of the inverter power tube.
For three-phase current i according to improved wavelet packet algorithm a 、i b 、i c Six-dimensional fault characteristic quantity M obtained by decomposition and reconstruction a 、M b 、M c 、H a 、H b 、H c Taking six-dimensional fault characteristic quantity as input layer neuron x of ELM identification model i (i-1, …,6) and the input column vector is x-x 1 ,x 2 ,x 3 ,x 4 ,x 5 ,x 6] T ；
Constructing a hidden layer with l neurons, where b 1 ,…,b l A threshold value for a corresponding neuron within the hidden layer; numbering the failure modes y j J (j ═ 1, …,21) as output layer neurons of the ELM recognition model, for hidden layer neurons g l (L ═ 1, …, L) weight β jl (j-1, …, 21; L-1, …, L) form output layer neuron y j ；
The ELM recognition model of the open-circuit fault of the inverter power tube can be expressed as:
5. the method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein: utilizing a WOA algorithm to carry out parameter optimization training on an ELM identification model of the open-circuit fault of the inverter power tube:
implanting a WOA algorithm into the ELM for the ELM parameter matrix (w, b, beta) of the optimal power tube open-circuit fault ELM identification model; the Whale Optimization Algorithm (WOA) is a novel meta-heuristic optimization algorithm proposed by Mirjalli and Lewis of the university of Gregorphis Australia in 2016, and has the advantages of few parameters, simple calculation and the like; WOA simulates hunting behavior of whales in the ocean, wherein the position of each whale represents a basic optimal solution;
the specific process for optimizing the WOA-ELM identification model parameters of the power tube open-circuit fault comprises the following steps:
(1) and (3) processing the sampling data: performing signal reconstruction on the sampled data, and extracting to obtain an open-circuit fault characteristic vector;
(2) initializing parameters: initializing whale population number S in WOA algorithm N And the maximum number of iterations T of the algorithm max Initializing a whale colony position X (w, b) in the WOA algorithm, namely an initial weight w and a threshold b in the ELM algorithm;
(3) calculating the fitness: and calculating the fitness of each whale in each iteration by taking the algorithm classification error rate as the fitness value, and selecting the individual with the minimum fitness as the optimal individual. The fitness function is:
F(X(t))＝(1-k/N)×100(3.29)
in the formula, N is the total amount of the samples; k represents the number of correct algorithm classification; the closer the individual fitness value is to 0, the more the individual meets the requirement according to the function;
(4) and (3) iterative updating: updating the position of the next generation according to the rules of surrounding, hunting and searching the prey by the algorithm;
(5) determining the optimal: when the algorithm reaches the iteration times or the optimal whale positions in the two iterations are the same, the iteration is completed, and the optimal individuals, namely the optimal initial w and b, are output;
(6) and training the ELM identification model according to the obtained optimal initial w and b to obtain an ELM optimal identification model parameter matrix (w, b, beta).
6. The method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein: according to the method, an optimal WOA-ELM inverter power tube open-circuit fault identification model is embedded into a control link of the optimal WOA-ELM inverter power tube open-circuit fault identification model, and a built-in current sensor is used for detecting three-phase current i a 、i b And i c And calculating six-dimensional fault characteristic quantity, and taking the six-dimensional fault characteristic quantity as the input of the WOA-ELM identification model of the optimal inverter power tube open-circuit fault.
7. The method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein the method comprises the following steps: according to the method, the fault mode number output in a 2-digit decimal mode of the WOA-ELM identification model of the open-circuit fault of the power tube of the inverter is converted into a 6-digit binary fault mode number, and real-time alarm of the open-circuit fault of the power tube of the inverter is achieved.
8. The method for identifying the open-circuit fault of the inverter power tube according to claim 1, wherein: the average identification time of the WOA-ELM model proposed by the method for a single fault sample is 2.97 multiplied by 10 -5 s。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210513910.4A CN115047305B (en) | 2022-05-12 | 2022-05-12 | Inverter open-circuit fault identification method based on signal processing reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210513910.4A CN115047305B (en) | 2022-05-12 | 2022-05-12 | Inverter open-circuit fault identification method based on signal processing reconstruction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115047305A true CN115047305A (en) | 2022-09-13 |
| CN115047305B CN115047305B (en) | 2024-07-05 |
Family
ID=83156824
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210513910.4A Active CN115047305B (en) | 2022-05-12 | 2022-05-12 | Inverter open-circuit fault identification method based on signal processing reconstruction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115047305B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108254688A (en) * | 2018-04-02 | 2018-07-06 | 中国矿业大学 | Power converter of switch reluctance motor small echo no-load voltage ratio method for diagnosing faults |
| CN109672356A (en) * | 2019-01-15 | 2019-04-23 | 同济大学 | The Active Fault-tolerant Control Method of ANPC three-level inverter single switch tube open circuit failure |
| CN112068033A (en) * | 2020-09-02 | 2020-12-11 | 河北工业大学 | 1/6 periodic current-based inverter power tube open-circuit fault online identification method |
| CN112578312A (en) * | 2020-12-04 | 2021-03-30 | 燕山大学 | Micro-grid fault diagnosis method based on whale algorithm optimization extreme learning machine |
| CN112613493A (en) * | 2021-01-11 | 2021-04-06 | 桂林电子科技大学 | Rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM |
| CN114236374A (en) * | 2021-12-13 | 2022-03-25 | 中国矿业大学 | Real-time diagnosis method for open circuit fault of rectifier |
-
2022
- 2022-05-12 CN CN202210513910.4A patent/CN115047305B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108254688A (en) * | 2018-04-02 | 2018-07-06 | 中国矿业大学 | Power converter of switch reluctance motor small echo no-load voltage ratio method for diagnosing faults |
| CN109672356A (en) * | 2019-01-15 | 2019-04-23 | 同济大学 | The Active Fault-tolerant Control Method of ANPC three-level inverter single switch tube open circuit failure |
| CN112068033A (en) * | 2020-09-02 | 2020-12-11 | 河北工业大学 | 1/6 periodic current-based inverter power tube open-circuit fault online identification method |
| CN112578312A (en) * | 2020-12-04 | 2021-03-30 | 燕山大学 | Micro-grid fault diagnosis method based on whale algorithm optimization extreme learning machine |
| CN112613493A (en) * | 2021-01-11 | 2021-04-06 | 桂林电子科技大学 | Rolling bearing fault diagnosis method based on multi-feature extraction and WOA-ELM |
| CN114236374A (en) * | 2021-12-13 | 2022-03-25 | 中国矿业大学 | Real-time diagnosis method for open circuit fault of rectifier |
Non-Patent Citations (14)
| Title |
|---|
| ZHANSHAN WANG等: "Multiscale Adaptive Fault Diagnosis Based on Signal Symmetry Reconstitution Preprocessing for Microgrid Inverter Under Changing Load Condition", IEEE TRANSACTIONS ON SMART GRID, 31 March 2018 (2018-03-31) * |
| 万晓凤;刘琦;杜利平;胡伟;罗旋;: "光伏并网***二电平逆变器的故障诊断", 电测与仪表, no. 10, 25 May 2016 (2016-05-25) * |
| 万红;任晓红;范晋瑜;俞啸;丁恩杰;: "三电平逆变器开路故障诊断研究", 工矿自动化, no. 04, 24 March 2020 (2020-03-24) * |
| 余运俊;裴石磊;万晓凤;辛建波;: "NPC三电平光伏逆变器开路故障诊断", 电子测量与仪器学报, no. 05, 15 May 2018 (2018-05-15) * |
| 吴锋: "功率变换器功率器件开路故障诊断方法的研究", 中国博士学位论文全文数据库工程科技Ⅱ辑, 15 March 2019 (2019-03-15) * |
| 唐圣学等: "三电平逆变器IGBT 器件接线故障在线诊断方法研究", 仪器仪表学报, 30 November 2020 (2020-11-30) * |
| 姚芳等: "基于GWO-ELM 的逆变器开路故障诊断", 电源学报, 31 January 2020 (2020-01-31) * |
| 孟庆学;王云亮;: "基于MATLAB GUIDE的逆变装置故障诊断软件***设计", 天津理工大学学报, no. 03, 15 June 2009 (2009-06-15) * |
| 季颖;李小波;冯鹏飞;王睿轶;王泉;: "城轨车辆辅助逆变器IGBT状态识别研究", 计算机测量与控制, no. 06, 25 June 2018 (2018-06-25) * |
| 张洪涛;安庆;: "基于小波变换和神经网络的逆变电路故障诊断方法", 计算机测量与控制, no. 06, 25 June 2010 (2010-06-25) * |
| 张淑清;苑世钰;姚玉永;穆勇;王丽丽;: "基于ICEEMD及AWOA优化ELM的机械故障诊断方法", 仪器仪表学报, no. 11, 10 January 2020 (2020-01-10) * |
| 徐继亚;王艳;纪志成;: "基于鲸鱼算法优化WKELM的滚动轴承故障诊断", ***仿真学报, no. 09, 8 September 2017 (2017-09-08) * |
| 董超群: "微电网并网逆变器功率管故障容错及动态冗余控制", 中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑, 15 February 2023 (2023-02-15) * |
| 黄湛钧: "微电网中三相逆变器开路故障诊断算法的研究", 中国博士学位论文全文数据库 工程科技Ⅱ辑, 15 April 2022 (2022-04-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115047305B (en) | 2024-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109635928B (en) | Voltage sag reason identification method based on deep learning model fusion | |
| CN112710969B (en) | Open-circuit fault diagnosis method for switching tube of single-phase half-bridge five-level inverter | |
| CN112070128B (en) | Transformer fault diagnosis method based on deep learning | |
| CN110726898B (en) | Power distribution network fault type identification method | |
| CN110954761A (en) | NPC three-level inverter fault diagnosis method based on signal sparse representation | |
| CN113125992B (en) | NPC three-level inverter fault diagnosis method and system based on DBN | |
| CN114355240A (en) | Power distribution network ground fault diagnosis method and device | |
| CN113159226B (en) | Inverter fault diagnosis method with integration of depth features and statistical features | |
| CN111060304A (en) | Circuit breaker fault diagnosis method based on neural network | |
| CN114065307A (en) | Hardware Trojan horse detection method and system based on bipartite graph convolutional neural network | |
| CN115329816A (en) | Multi-level inverter fault diagnosis method based on simulation data migration learning | |
| CN116484299A (en) | Charging pile fault diagnosis method based on integration of gradient lifting tree and multi-layer perceptron | |
| CN109901064B (en) | ICA-LVQ-based high-voltage circuit breaker fault diagnosis method | |
| CN117272102A (en) | Transformer fault diagnosis method based on double-attention mechanism | |
| CN114118149A (en) | Induction motor fault diagnosis system based on finite element simulation and symmetric feature migration | |
| CN112068033B (en) | On-line identification method for open-circuit faults of inverter power tube based on 1/6 period current | |
| Xu et al. | Loop optimization noise-reduced LSTM based classifier for PD detection | |
| CN113610119A (en) | Method for identifying power transmission line developmental fault based on convolutional neural network | |
| CN116702060A (en) | Multi-level inverter power device fault diagnosis method | |
| CN115047305A (en) | Inverter open-circuit fault identification method based on signal processing reconstruction | |
| CN116401610A (en) | Inverter parameter fault diagnosis method based on depth residual error network and joint distribution | |
| CN116559728A (en) | Power transmission line fault diagnosis method based on wavelet transformation-multilayer perceptron | |
| CN114019310B (en) | Open-circuit fault pre-judging method, equipment and medium for energy storage converter | |
| CN114296005A (en) | Modular multilevel converter submodule composite fault diagnosis method | |
| CN115310604A (en) | UPS system online fault diagnosis method based on neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |