CN110084106A - Microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network - Google Patents
Microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network Download PDFInfo
- Publication number
- CN110084106A CN110084106A CN201910206025.XA CN201910206025A CN110084106A CN 110084106 A CN110084106 A CN 110084106A CN 201910206025 A CN201910206025 A CN 201910206025A CN 110084106 A CN110084106 A CN 110084106A
- Authority
- CN
- China
- Prior art keywords
- fault
- neural network
- probabilistic neural
- phase
- network model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 37
- 230000009466 transformation Effects 0.000 title claims abstract description 20
- 239000013598 vector Substances 0.000 claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 37
- 238000003745 diagnosis Methods 0.000 claims abstract description 32
- 238000004088 simulation Methods 0.000 claims abstract description 24
- 238000009499 grossing Methods 0.000 claims abstract description 15
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 14
- 230000002068 genetic effect Effects 0.000 claims abstract description 12
- 238000004458 analytical method Methods 0.000 claims abstract description 8
- 238000003062 neural network model Methods 0.000 claims description 38
- 210000000349 chromosome Anatomy 0.000 claims description 15
- 210000002569 neuron Anatomy 0.000 claims description 15
- 238000005259 measurement Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 13
- 230000035772 mutation Effects 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 108090000623 proteins and genes Proteins 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 7
- 238000011217 control strategy Methods 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 4
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 125000004432 carbon atom Chemical group C* 0.000 claims description 3
- 238000000354 decomposition reaction Methods 0.000 claims description 3
- 238000000819 phase cycle Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 abstract description 5
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000005457 optimization Methods 0.000 abstract description 3
- 238000010276 construction Methods 0.000 abstract 1
- 230000000153 supplemental effect Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 6
- 238000010248 power generation Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
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
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
- G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Probability & Statistics with Applications (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The present invention provides a kind of microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network, it include: the fault simulation system for building microgrid inverter in simulation software first, then the threephase load electric current under different faults type is acquired as measuring signal, its fault characteristic signals is extracted using wavelet multi_resolution analysis method;In turn by the fault characteristic signals normalized construction feature vector of extraction;The fault feature vector of extraction is finally chosen a part as training data to be trained PNN model, the optimal smoothing of probabilistic neural network is found because of period of the day from 11 p.m. to 1 a.m σ using genetic algorithm simultaneously, when inverter breaks down, current signal is subjected to fault location according to the above method, to realize microgrid inverter fault diagnosis.The beneficial effects of the present invention are: structural model is simple, be easy determination, fast convergence rate and converges on Bayes optimization solution, sample supplemental capabilities are strong, do not need re -training, and precision is high, practicality is strong, and are easy to engineering combination.
Description
Technical Field
The invention relates to the field of power electronic fault diagnosis, in particular to a micro-grid inverter fault diagnosis method based on wavelet transformation and a probabilistic neural network.
Background
Today, global economy, rapid technological development and increasingly large industrial scale. The demand of people for energy is increasing day by day. Along with the continuous exhaustion of traditional energy sources such as coal, petroleum and the like and the serious pollution to the environment, countries in the world are gradually aware of the importance of the energy sources and the environment. Therefore, clean energy sources such as solar energy, wind energy and the like are vigorously developed. Distributed power generation is used as an effective utilization way of clean energy, and a microgrid is a low-voltage system consisting of a distributed power supply, energy storage, loads and the like. The micro-grid is used as an important bridge and a link of a large power grid and distributed power generation, so that the impact of the distributed power generation on the power grid can be effectively reduced, and the power supply reliability is improved. And the power conversion and power control in the micro-grid are mainly carried by the inverter. If the inverter fails and cannot be diagnosed and repaired, irretrievable economic loss and safety risks are caused. Therefore, safe, stable and reliable operation of the inverter is very important for the power system.
The fault diagnosis is actually one mode identification, and the main research content of the fault diagnosis is to realize the reason mining, information acquisition, feature extraction and system state analysis and identification of the fault. The information acquisition module in the Matlab/Simulink module is adopted in the invention, and the information characteristic extraction mainly comprises Fourier transform, wavelet transform, principal component analysis method and the like at present. The fault diagnosis and location mainly comprises methods such as an expert system method, an artificial neural network, a support vector machine and the like. The fault types of the switching devices of the microgrid inverter are complex, the traditional manual search cannot meet the rapid popularization of a large-scale microgrid, and with the rapid development of artificial intelligence, computer technology and signal processing technology, the microgrid inverter is suitable for the large-scale microgrid. Based on the idea of data driving, data such as current signals reflecting the continuous change of the operation state in the operation process of the inverter are utilized to carry out data analysis and feature extraction, and the fault position of the inverter can be rapidly positioned in real time.
The invention adopts wavelet transformation to extract the characteristic vector of the PNN network model, and then adopts a genetic algorithm to determine the optimal smoothing factor sigma to be brought into the PNN network model for fault location. The wavelet can extract important information of time domain and frequency domain of the signal at the same time, and can amplify local information and add certain noise to simulate signal interference existing in real objects, and is called as a mathematical microscope. And the probabilistic neural network completes the work of the nonlinear learning algorithm by the linear learning algorithm due to simple structure determination and high diagnosis speed of the network model. The stability is strong, and the Bayesian optimal solution is converged to meet the requirement of real-time processing.
Disclosure of Invention
In order to solve the problems that the fault type of a switching device of a microgrid inverter is complex and the traditional manual search cannot meet the rapid popularization of a large-scale microgrid, the invention provides a microgrid inverter fault diagnosis method based on wavelet transformation and a probabilistic neural network, which mainly comprises the following steps:
s101: according to the running condition of the microgrid inverter to be diagnosed, a fault simulation system of the microgrid inverter is built, and three-phase load current signals of the fault simulation system under different fault types are collected to be used as measuring signals; the fault type is a plurality of fault numbers preset according to the possible faults;
s102: extracting fault characteristic signals of the measurement signals by adopting a wavelet multi-resolution analysis method, and performing normalization and noise addition processing on the extracted fault characteristic signals to obtain fault sample data;
s103: building a probabilistic neural network model, and training the probabilistic neural network model by using the fault sample data to obtain a trained probabilistic neural network model; in the training process of the probabilistic neural network model, searching an optimal smooth factor of the probabilistic neural network model by adopting a genetic algorithm;
s104: and diagnosing the actual fault of the microgrid inverter to be diagnosed by using the trained probabilistic neural network model, thereby realizing the microgrid inverter fault diagnosis based on the wavelet transformation and the probabilistic neural network.
Further, in step S101, a microgrid inverter PQ control strategy is adopted to build a fault simulation system on the Matlab/Simulink platform, wherein a three-phase grid-connected inverter in the fault simulation system adopts an SVPWM modulation method, a fault simulation module is added behind a pulse trigger module in the fault simulation system, and then an open-circuit fault of the microgrid inverter is simulated by controlling the trigger pulse; the method comprises the following specific steps:
s201: acquiring three-phase voltage and three-phase current at the side of a power grid, and converting the three-phase voltage and the three-phase current into a voltage signal under a dq axis and a current signal under the dq axis through Clarke and Park conversion;
s202: taking a voltage signal and a current signal under the dq axis as the input of a phase-locked loop, and acquiring the phase theta of the voltage and the current by using a phase-locked loop technology;
s203: the obtained voltage and current phases are transmitted to the microgrid central controller; the microgrid central controller sends a power instruction P according to the received voltage and current phasesrefAnd QrefAnd generating a current command IdrefAnd IqrefAnd obtaining U through a double-ring control structure and Park inverse transformationαβ;
S204: the obtained UαβDirect current U of microgrid controllerdcAs the input of SVPWM, generating 6 pulses as the driving pulses of the microgrid inverter through SVPWM modulation; finally, a fault trigger module is added, the driving pulse is used as the input of the fault trigger module, and a microgrid inverter fault simulation system is generated; the fault triggering module simulates the turn-off of the switching tube by controlling the pulse of the switching tube.
Further, in step S102, a wavelet multi-resolution analysis method is used to extract a fault feature signal of the measurement signal, and the extracted fault feature signal is normalized and processed by adding noise, which specifically includes:
s301: decomposing the measuring signal by using a db3 wavelet 5-layer decomposition method to obtain all frequency components S of the measuring signalph|j(ii) a It is composed ofIn the formula, ph is a phase sequence number and takes the value of A, B or C; j is the frequency band serial number corresponding to each phase, 0 is 0-125 Hz, and 1 is 126-250 Hz;
s302: selecting a frequency component Sph|jThe low-frequency partial coefficient is subjected to energy reconstruction to obtain a frequency component Sph|jCorresponding energy Eph|jThe reconstruction formula is shown in formula (1):
in the above formula, xph|jkRepresenting the frequency component Sph|jN is the frequency component Sph|jThe total number of discrete points of (a);
s303: taking a vector formed by the low-frequency band energy of the three phases as a fault feature vector T, as shown in formula (2):
T=[EA|0,EA|1,EB|0,EB|1,EC|0,EC|1](2)
in the above formula, EA|0、EA|1、EB|0、EB|1、EC|0And EC|1Respectively is the energy of the 0-125 Hz frequency section of the A phase, the energy of the 126-250 Hz frequency section of the A phase, the energy of the 0-125 Hz frequency section of the B phase, the energy of the 126-250 Hz frequency section of the B phase, the energy of the 0-125 Hz frequency section of the C phase and the energy of the 126-250 Hz frequency section of the C phase;
s304: the fault feature vector T is subjected to normalization processing as shown in formula (3):
Ej=EA|j+EB|j+EC|j(3)
obtaining a normalized fault feature vector T1, where the normalized fault feature vector T1 is shown in formula (4):
s305: adding 5% of random noise into the normalized fault feature vector T1 to simulate the interference of other signals in a real object, and obtaining a fault feature vector T2 added with the random noise;
under different fault types, a plurality of measurement signals are respectively collected, and after the measurement signals are processed in the steps S301 to S305, corresponding fault characteristic vectors T2 are obtained to form fault sample data.
Further, in step S103, the probabilistic neural network model is composed of 4 layers of an input layer, a mode layer, a summation layer and an output layer;
the input layer is used for receiving the value of a fault feature vector T2 in the fault sample data so as to transfer the fault feature vector T2 to a probabilistic neural network; the number of neurons of the input layer and the dimension of the fault feature vector T2 are equal;
the mode layer is used for calculating the matching relation between the input fault feature vector T2 and each fault type in the fault sample data, and the calculation formula is shown as formula (5):
in the above formula, WiThe weight value from the input layer to the mode layer is connected; x is the value corresponding to the input fault characteristic vector T2; δ is a smoothing factor; the number of the pattern layer neurons is equal to the sum of the sample numbers of the corresponding fault types, i is an integer larger than 0 and represents the serial numbers of the corresponding neurons;
the summation layer is used for accumulating the probabilities belonging to a certain fault type so as to obtain an estimated probability density function of each fault type, and the probability density function is obtained by a Parzen method, as shown in the formula (6):
in the above formula, XaiAs a fault type thetaAThe fault feature vector T2 corresponding to the ith training sample of (a); m is fault type thetaATotal number of training samples; p is probability and is a preset value; the value of δ determines the width of the bell curve centered at the sample point; the number of the neurons of the summation layer is equal to the total number of the fault types;
and the output layer consists of a threshold discriminator and is used for selecting a neuron with the maximum estimated probability density from the estimated probability densities of all fault types as output and outputting the fault types.
Further, in step S103, when the probabilistic neural network model is trained using the fault sample data: dividing the fault sample data into two parts of training data and testing data; training the probabilistic neural network model by using training data to modify the expansion speed of a radial basis function in the probabilistic neural network model, and searching an optimal smooth factor of the probabilistic neural network by using a genetic algorithm in the training process to obtain an optimal probabilistic neural network model; the test set is used for inputting the probabilistic neural network model, testing the probabilistic neural network model and judging whether the predicted output result is consistent with the actual output result or not to obtain indexes such as diagnosis speed, accuracy and the like; and continuously and circularly training and testing until the precision reaches a preset value or the iteration frequency reaches a maximum set value, stopping iteration, and taking the probabilistic neural network model at the moment as a trained probabilistic neural network model.
Further, in step S103, finding an optimal smoothing factor of the probabilistic neural network by using a genetic algorithm, specifically including:
coding the smoothing factor into a chromosome to form an initial population, and performing selection, crossing, variation and updating iteration by calculating fitness to obtain an optimal chromosome, namely an optimal smoothing factor;
calculating the fitness through a fitness function, and then performing selection, intersection and variation on the fitness function; wherein the fitness function is shown in equation (7);
in the above formula, M is the number of samples corresponding to training data, I is the number of classes of training classes (fault types), and I is the total number of classes; y isimA and yimRespectively, the predicted value and the actual value of the mth sample, and a is a positive minimum value, so as to avoid the denominator being 0.
The selection operation adopts a roulette method, namely a selection strategy based on fitness proportion and the selection probability p of each individual iiAs in equation (8):
in the above formula, the first and second carbon atoms are,Fithe fitness value of the individual i is obtained, and k is a coefficient and is a preset value; n is the number of population individuals, and the value ranges of i and j are the same and are all [1, N];
The crossing operation adopts a real number crossing method, and the kth individual chromosome akAnd the l-th chromosome a1The interleaving operation method at j bit is shown in equation (9):
in the above formula, akjFor chromosomes produced after crossover operations, b is [0,1 ]]A random number in between;
mutation operation for selecting jth gene a of ith individualijPerforming mutation, wherein the mutation operation is shown as a formula (10):
in the above formula, amaxIs gene aijUpper bound of aminIs gene aijThe lower bound of (c); f (g) r2(1-g/Gmax)2;r2Is a random number; g is the current iteration number; gmaxThe maximum number of evolutions; r is [0,1 ]]Random number in between.
The technical scheme provided by the invention has the beneficial effects that: the classification effect of the commonly used BP neural network is greatly influenced by the initial weight and the model structure, the classification result lacks transparency, and the learning speed is slow and the method is easy to fall into the local optimal solution. The technical scheme provided by the invention has the advantages of simple structure model, easy determination, high convergence speed, convergence on Bayes optimization solution, strong sample addition capability, no need of retraining, high precision, strong practicality and easy engineering combination.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
fig. 1 is a flowchart of a microgrid inverter fault diagnosis method based on wavelet transformation and a probabilistic neural network in an embodiment of the present invention;
fig. 2 is a schematic diagram of a microgrid inverter rotating coordinate system PQ control in the embodiment of the present invention;
FIG. 3 is a waveform diagram of a fault of three-phase current on the grid side when some switching tubes are in fault in the embodiment of the invention;
fig. 4 is an exploded view of the switching tube 1 in the embodiment of the invention, showing the wavelet transformation of the fault a current;
FIG. 5 is a diagram of a probabilistic neural network base model in an embodiment of the present invention;
FIG. 6 is a flow chart of the genetic algorithm optimizing the probabilistic neural network smoothing factor σ in an embodiment of the present invention;
fig. 7(a) to 7(c) are partial failure diagnosis result diagrams in the embodiment of the present invention.
Detailed Description
For a more clear understanding of the technical features, objects and effects of the present invention, embodiments of the present invention will now be described in detail with reference to the accompanying drawings.
The embodiment of the invention provides a microgrid inverter fault diagnosis method based on wavelet transformation and a probabilistic neural network.
Referring to fig. 1, fig. 1 is a flowchart of a microgrid inverter fault diagnosis method based on wavelet transformation and a probabilistic neural network in an embodiment of the present invention, which specifically includes the following steps:
s101: according to the running condition of the microgrid inverter to be diagnosed, a fault simulation system of the microgrid inverter is built, and three-phase load current signals of the fault simulation system under different fault types are collected to be used as measuring signals; the fault type is a plurality of fault numbers preset according to the possible faults;
s102: extracting fault characteristic signals of the measurement signals by adopting a wavelet multi-resolution analysis method, and performing normalization and noise addition processing on the extracted fault characteristic signals to obtain fault sample data;
s103: building a probabilistic neural network model, and training the probabilistic neural network model by using the fault sample data to obtain a trained probabilistic neural network model; in the training process of the probabilistic neural network model, searching an optimal smooth factor of the probabilistic neural network model by adopting a genetic algorithm;
s104: and diagnosing the actual fault of the microgrid inverter to be diagnosed by using the trained probabilistic neural network model, thereby realizing the microgrid inverter fault diagnosis based on the wavelet transformation and the probabilistic neural network.
In the step S101, a microgrid inverter PQ control strategy is adopted to build a fault simulation system on a Matlab/Simulink platform, wherein a three-phase grid-connected inverter in the fault simulation system adopts an SVPWM modulation method, a fault simulation module is added behind a pulse trigger module in the fault simulation system, and then the open-circuit fault of the microgrid inverter is simulated by controlling the trigger pulse; simulating a fault of a switching tube by using a fault simulation system to generate a fault signal, and storing the fault signal in a working space for storage; fig. 3 is a waveform diagram of a three-phase current fault on the grid side when a part of the switching tubes are in fault.
The microgrid inverter PQ control strategy is an inverter control mode commonly used for new energy grid connection such as photovoltaic power generation and wind power generation, fig. 2 is a microgrid inverter rotating coordinate system PQ control (microgrid inverter PQ control strategy) schematic diagram, and a fault simulation system is built on a Matlab/Simulink platform according to the topological structure, and the method specifically comprises the following steps:
s201: collecting three-phase voltage (U) at the side of the grida、UbAnd Uc) And three-phase current (I)a、IbAnd Ic) And further converting the three-phase voltages and the three-phase currents to voltage signals (U) under the dq axis by Clarke and Park conversiondAnd Uq) And current signal (I) in the dq axisd、Iq);
S202: taking a voltage signal and a current signal under the dq axis as the input of a phase-locked loop, and acquiring the phase theta of the voltage and the current by using a phase-locked loop technology;
s203: the obtained voltage and current phases are transmitted to the microgrid central controller; the microgrid central controller sends a power instruction P according to the received voltage and current phasesrefAnd QrefAnd generating a current command IdrefAnd IqrefAnd through double-ring control structure and Park inverse transformationTo obtain Uαβ;
S204: the obtained UαβDirect current U of microgrid controllerdcAs the input of SVPWM, generating 6 pulses as the driving pulses of the microgrid inverter through SVPWM modulation; finally, a fault trigger module is added, the driving pulse is used as the input of the fault trigger module, and a microgrid inverter fault simulation system is generated; the fault triggering module simulates the turn-off of the switching tube by controlling the pulse of the switching tube.
Since the short-circuit fault of the switching tube can also be converted into an open-circuit fault (line protection) finally, only the open-circuit fault of the switching tube is considered in the invention, and the fault types are as follows (5 major type, 22 minor type):
1) all the switching tubes run normally, and the switching tubes have no fault;
2) the fault of a single switching tube is in 6 subclasses;
3) the faults of the switching tubes of the same bridge arm are in 3 subclasses;
4) the switching tube faults of the same upper bridge arm or the same lower bridge arm are totally 6 types;
5) and 6 types of faults of the crossed bridge arms of the switching tubes.
In step S102, a wavelet multi-resolution analysis method is used to extract a fault feature signal of the measurement signal, and the extracted fault feature signal is normalized and subjected to noise addition processing, which specifically includes the following steps:
s301: decomposing the measuring signal by using a db3 wavelet 5-layer decomposition method (as shown in fig. 4, which is an exploded view of the switching tube 1 fault a current wavelet transformation in the embodiment of the present invention), so as to obtain all frequency components S of the measuring signalph|j(ii) a Wherein, ph is a phase sequence number and takes the value of A, B or C; j is the frequency band serial number corresponding to each phase, 0 is 0-125 Hz, and 1 is 126-250 Hz;
s302: selecting frequency components since the high frequency part of the different measurement signals is not significantly changedSph|jThe low-frequency partial coefficient is subjected to energy reconstruction to obtain a frequency component Sph|jCorresponding energy Eph|jThe reconstruction formula is shown as formula (1):
in the above formula, xph|jkRepresenting the frequency component Sph|jN is the frequency component Sph|jThe total number of discrete points of (a);
s303: taking a vector formed by the low-frequency band energy of the three phases as a fault feature vector T, as shown in formula (2):
T=[EA|0,EA|1,EB|0,EB|1,EC|0,EC|1](2)
in the above formula, EA|0、EA|1、EB|0、EB|1、EC|0And EC|1Respectively is the energy of the 0-125 Hz frequency section of the A phase, the energy of the 126-250 Hz frequency section of the A phase, the energy of the 0-125 Hz frequency section of the B phase, the energy of the 126-250 Hz frequency section of the B phase, the energy of the 0-125 Hz frequency section of the C phase and the energy of the 126-250 Hz frequency section of the C phase;
s304: the fault feature vector T is normalized, as shown in formula (3):
Ej=EA|j+EB|j+EC|j(3)
obtaining a normalized fault feature vector T1, where the normalized fault feature vector T1 is shown in formula (4):
s305: adding 5% of random noise into the normalized fault feature vector T1 to simulate the interference of other signals in a real object, and obtaining a fault feature vector T2 added with the random noise;
under different fault types, a plurality of measurement signals are respectively collected, and after the measurement signals are processed in the steps S301 to S305, corresponding fault characteristic vectors T2 are obtained to form fault sample data.
In step S103, the probabilistic neural network model is composed of 4 layers of an input layer, a mode layer, a summation layer, and an output layer; FIG. 5 shows a basic model of a probabilistic neural network;
the input layer is used for receiving the value of a characteristic vector T2 in the fault sample data so as to transfer the fault characteristic vector T2 to a probabilistic neural network; the number of neurons of the input layer and the dimension of the fault feature vector T2 are equal;
the mode layer is used for calculating the matching relation between the input fault feature vector T2 and each fault type in the fault sample data, and the calculation formula is shown as formula (5):
in the above formula, WiThe weight value from the input layer to the mode layer is connected; x is a value corresponding to the input fault characteristic vector; δ is a smoothing factor; the number of the pattern layer neurons is equal to the sum of the sample numbers of the corresponding fault types, i is an integer larger than 0 and represents the serial numbers of the corresponding neurons;
the summation layer is used for accumulating the probabilities belonging to a certain fault type so as to obtain an estimated probability density function of each fault type, and the probability density function is obtained by a Parzen method, as shown in the formula (6):
in the above formula, XaiAs a fault type thetaAThe fault feature vector (T2) corresponding to the ith training sample of (a); m is fault type thetaATotal number of training samples; p is probability and is a preset value; the value of δ determines the width of the bell curve centered at the sample point; the number of the neurons of the summation layer is equal to the total number of the fault types;
and the output layer consists of a threshold discriminator and is used for selecting a neuron with the maximum estimated probability density from the estimated probability densities of all fault types as output and outputting the fault types.
In step S103, when training a Probabilistic Neural Network (PNN) model using the fault sample data: dividing the fault sample data into two parts, namely training data and testing data (in the embodiment of the invention, 175 groups of data are selected for each fault type, wherein 150 groups of data are used as training data, and 25 groups of data are used as testing data); training the PNN network model by using the training data to modify the expansion speed of a radial basis function in the PNN network model, and searching an optimal smooth factor of the probabilistic neural network by using a genetic algorithm in the training process to obtain an optimal PNN classification model; the test set is used for inputting the PNN network model, testing the PNN network model and judging whether the predicted output result is consistent with the actual output result or not to obtain indexes such as diagnosis speed, accuracy and the like; and (4) continuously and circularly training and testing until the precision (the error percentage between the model output and the actual fault type label) reaches a preset value or the iteration frequency reaches a maximum set value, stopping iteration, and taking the PNN network model at the moment as a trained probabilistic neural network model.
In step S103, finding an optimal smoothing factor of the probabilistic neural network by using a genetic algorithm, specifically including:
coding the smoothing factor into a chromosome to form an initial population, and performing selection, crossing, variation and updating iteration by calculating fitness to obtain an optimal chromosome, namely an optimal smoothing factor; the flow is shown in fig. 5.
Calculating the fitness through a fitness function, and then performing selection, intersection and variation on the fitness function; wherein the fitness function is shown in equation (7);
in the above formula, M is the number of samples corresponding to training data, I is the number of classes of training classes (fault types), and I is the total number of classes; y isimA and yimRespectively, the predicted value and the actual value of the mth sample, and a is a positive minimum value, so as to avoid the denominator being 0.
The selection operation adopts a roulette method, namely a selection strategy based on fitness proportion and the selection probability p of each individual iiAs in equation (8):
in the above formula, the first and second carbon atoms are,Fithe fitness value of the individual i is obtained, and k is a coefficient and is a preset value; n is the number of population individuals, and the value ranges of i and j are the same and are all [1, N];
The crossing operation adopts a real number crossing method, and the kth individual chromosome akAnd the l-th chromosome a1The interleaving operation method at j bit is shown in equation (9):
in the above formula, akjFor chromosomes produced after crossover operations, b is [0,1 ]]A random number in between;
mutation operation for selecting jth gene a of ith individualijPerforming mutation, wherein the mutation operation is shown as a formula (10):
in the above formula, amaxIs gene aijUpper bound of aminIs gene aijThe lower bound of (c); f (g) r2(1-g/Gmax)2;r2Is a random number; g is the current iteration number; gmaxThe maximum number of evolutions; r is [0,1 ]]A random number in between; the specific flow chart is shown in fig. 6.
Table 1 shows a list of fault type tags used in the embodiment of the present invention:
TABLE 1 Fault type tag
Type of failure | Category label | Type of failure | Category label |
Switching tube 1 fault | 1 | Switching tubes 3 and 5 fail simultaneously | 35 |
Switching tube 2 failure | 2 | Switching tubes 2 and 4 fail simultaneously | 24 |
Switch with a switch bodyPipe 3 failure | 3 | Switching tubes 2 and 6 fail simultaneously | 26 |
Switching tube 4 fault | 4 | Switching tubes 4 and 6 fail simultaneously | 46 |
Switching tube 5 fault | 5 | Switching tubes 1 and 4 fail simultaneously | 14 |
Switching tube 6 failure | 6 | Switching tubes 1 and 6 fail simultaneously | 16 |
Switching tubes 1 and 2 fail simultaneously | 12 | Switching tubes 3 and 2 fail simultaneously | 32 |
Switching tubes 3 and 4 fail simultaneously | 34 | Switching tubes 3 and 6 fail simultaneously | 36 |
Switching tubes 5 and 6 fail simultaneously | 56 | Switching tubes 5 and 2 fail simultaneously | 52 |
Switching tubes 1 and 3 fail simultaneously | 13 | Switching tubes 5 and 6 fail simultaneously | 56 |
Switching tubes 1 and 5 fail simultaneously | 15 | No fault of switch tube | 0 |
As shown in table 2, a diagnosis result table for performing fault diagnosis on the microgrid inverter by using different methods in the embodiment of the present invention is shown:
TABLE 2 Fault diagnosis result Table
Fault diagnosis method | BP neural network | Support Vector Machine (SVM) | Probabilistic Neural Network (PNN) |
Fault diagnosis accuracy | 91.78% | 96.00% | 97.73% |
Time to failure diagnosis | 15-30 seconds | 150-420 seconds | 5-10 seconds |
Fig. 7(a) to 7(c) show partial failure diagnosis results in the embodiment of the present invention.
The invention has the beneficial effects that: the classification effect of the commonly used BP neural network is greatly influenced by the initial weight and the model structure, the classification result lacks transparency, and the learning speed is slow and the method is easy to fall into the local optimal solution. The technical scheme provided by the invention has the advantages of simple structure model, easy determination, high convergence speed, convergence on Bayes optimization solution, strong sample addition capability, no need of retraining, high precision, strong practicality and easy engineering combination.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (6)
1. A microgrid inverter fault diagnosis method based on wavelet transformation and a probabilistic neural network is characterized in that: the method comprises the following steps:
s101: according to the running condition of the microgrid inverter to be diagnosed, a fault simulation system of the microgrid inverter is built, and three-phase load current signals of the fault simulation system under different fault types are collected to be used as measuring signals; the fault type is a plurality of fault numbers preset according to the possible faults;
s102: extracting fault characteristic signals of the measurement signals by adopting a wavelet multi-resolution analysis method, and performing normalization and noise addition processing on the extracted fault characteristic signals to obtain fault sample data;
s103: building a probabilistic neural network model, and training the probabilistic neural network model by using the fault sample data to obtain a trained probabilistic neural network model; in the training process of the probabilistic neural network model, searching an optimal smooth factor of the probabilistic neural network model by adopting a genetic algorithm;
s104: and diagnosing the actual fault of the microgrid inverter to be diagnosed by using the trained probabilistic neural network model, thereby realizing the microgrid inverter fault diagnosis based on the wavelet transformation and the probabilistic neural network.
2. The microgrid inverter fault diagnosis method based on wavelet transformation and probabilistic neural network as claimed in claim 1, characterized in that: in the step S101, a microgrid inverter PQ control strategy is adopted to build a fault simulation system on a Matlab/Simulink platform, wherein a three-phase grid-connected inverter in the fault simulation system adopts an SVPWM modulation method, a fault simulation module is added behind a pulse trigger module in the fault simulation system, and then the open-circuit fault of the microgrid inverter is simulated by controlling the trigger pulse; the method comprises the following specific steps:
s201: acquiring three-phase voltage and three-phase current at the side of a power grid, and converting the three-phase voltage and the three-phase current into a voltage signal under a dq axis and a current signal under the dq axis through Clarke and Park conversion;
s202: taking a voltage signal and a current signal under the dq axis as the input of a phase-locked loop, and acquiring the phase theta of the voltage and the current by using a phase-locked loop technology;
s203: the obtained voltage and current phases are transmitted to the microgrid central controller; the microgrid central controller sends a power instruction P according to the received voltage and current phasesrefAnd QrefAnd generating a current command IdrefAnd IqrefAnd obtaining U through a double-ring control structure and Park inverse transformationαβ;
S204: will be provided withObtained UαβDirect current U of microgrid controllerdcAs the input of SVPWM, generating 6 pulses as the driving pulses of the microgrid inverter through SVPWM modulation; finally, a fault trigger module is added, the driving pulse is used as the input of the fault trigger module, and a microgrid inverter fault simulation system is generated; the fault triggering module simulates the turn-off of the switching tube by controlling the pulse of the switching tube.
3. The microgrid inverter fault diagnosis method based on wavelet transformation and probabilistic neural network as claimed in claim 1, characterized in that: in step S102, a wavelet multi-resolution analysis method is used to extract a fault feature signal of the measurement signal, and the extracted fault feature signal is normalized and subjected to noise addition processing, which specifically includes:
s301: decomposing the measuring signal by using a db3 wavelet 5-layer decomposition method to obtain all frequency components S of the measuring signalph|j(ii) a Wherein,phthe value is A, B or C for the phase sequence number; j is the serial number of the frequency segment corresponding to each phase, 0 is 0-125 Hz, and 1 is 126-250 Hz;
s302: selecting a frequency component Sph|jThe low-frequency partial coefficient is subjected to energy reconstruction to obtain a frequency component Sph|jCorresponding energy Eph|jThe reconstruction formula is shown in formula (1):
in the above formula, xph|jkRepresenting the frequency component Sph|jN is the frequency component Sph|jThe total number of discrete points of (a);
s303: taking a vector formed by the low-frequency band energy of the three phases as a fault feature vector T, as shown in formula (2):
T=[EA|0,EA|1,EB|0,EB|1,EC|0,EC|1](2)
in the above formula, EA|0、EA|1、EB|0、EB|1、EC|0And EC|1Respectively is the energy of the 0-125 Hz frequency section of the A phase, the energy of the 126-250 Hz frequency section of the A phase, the energy of the 0-125 Hz frequency section of the B phase, the energy of the 126-250 Hz frequency section of the B phase, the energy of the 0-125 Hz frequency section of the C phase and the energy of the 126-250 Hz frequency section of the C phase;
s304: the fault feature vector T is subjected to normalization processing as shown in formula (3):
Ej=EA|j+EB|j+EC|j(3)
obtaining a normalized fault feature vector T1, where the normalized fault feature vector T1 is shown in formula (4):
s305: adding 5% of random noise into the normalized fault feature vector T1 to simulate the interference of other signals in a real object, and obtaining a fault feature vector T2 added with the random noise;
under different fault types, a plurality of measurement signals are respectively collected, and after the measurement signals are processed in the steps S301 to S305, corresponding fault characteristic vectors T2 are obtained to form fault sample data.
4. The microgrid inverter fault diagnosis method based on wavelet transformation and probabilistic neural network as claimed in claim 1, characterized in that: in step S103, the probabilistic neural network model is composed of 4 layers of an input layer, a mode layer, a summation layer, and an output layer;
the input layer is used for receiving the value of a fault feature vector T2 in the fault sample data so as to transfer the fault feature vector T2 to a probabilistic neural network; the number of neurons of the input layer and the dimension of the fault feature vector T2 are equal;
the mode layer is used for calculating the matching relation between the input fault feature vector T2 and each fault type in the fault sample data, and the calculation formula is shown as formula (5):
in the above formula, WiThe weight value from the input layer to the mode layer is connected; x is the value corresponding to the input fault characteristic vector T2; δ is a smoothing factor; the number of the pattern layer neurons is equal to the sum of the sample numbers of the corresponding fault types, i is an integer larger than 0 and represents the serial numbers of the corresponding neurons;
the summation layer is used for accumulating the probabilities belonging to a certain fault type so as to obtain an estimated probability density function of each fault type, and the probability density function is obtained by a Parzen method, as shown in the formula (6):
in the above formula, XaiAs a fault type thetaAThe fault feature vector T2 corresponding to the ith training sample of (a); m is fault type thetaATotal number of training samples; p is probability and is a preset value; the value of δ determines the width of the bell curve centered at the sample point; the number of the neurons of the summation layer is equal to the total number of the fault types;
and the output layer consists of a threshold discriminator and is used for selecting a neuron with the maximum estimated probability density from the estimated probability densities of all fault types as output and outputting the fault types.
5. The microgrid inverter fault diagnosis method based on wavelet transformation and probabilistic neural network as claimed in claim 1, characterized in that: in step S103, when the probabilistic neural network model is trained using the fault sample data: dividing the fault sample data into two parts of training data and testing data; training the probabilistic neural network model by using training data to modify the expansion speed of a radial basis function in the probabilistic neural network model, and searching an optimal smooth factor of the probabilistic neural network by using a genetic algorithm in the training process to obtain an optimal probabilistic neural network model; the test set is used for inputting the probabilistic neural network model, testing the probabilistic neural network model and judging whether the predicted output result is consistent with the actual output result or not to obtain indexes such as diagnosis speed, accuracy and the like; and continuously and circularly training and testing until the precision reaches a preset value or the iteration frequency reaches a maximum set value, stopping iteration, and taking the probabilistic neural network model at the moment as a trained probabilistic neural network model.
6. The microgrid inverter fault diagnosis method based on wavelet transformation and probabilistic neural network as claimed in claim 1, characterized in that: in step S103, finding an optimal smoothing factor of the probabilistic neural network by using a genetic algorithm, specifically including:
coding the smoothing factor into a chromosome to form an initial population, and performing selection, crossing, variation and updating iteration by calculating fitness to obtain an optimal chromosome, namely an optimal smoothing factor;
calculating the fitness through a fitness function, and then performing selection, intersection and variation on the fitness function; wherein the fitness function is shown in equation (7);
in the above formula, M is the number of samples corresponding to training data, I is the number of classes of training classes (fault types), and I is the total number of classes; y isimA and yimRespectively a predicted value and an actual value of the mth sample, wherein a is a positive minimum value, and in order to avoid the denominator being 0;
the selection operation adopts a roulette method, namely a selection strategy based on fitness proportion and the selection probability p of each individual iiAs in equation (8):
in the above formula, the first and second carbon atoms are,Fithe fitness value of the individual i is obtained, and k is a coefficient and is a preset value; n is the number of population individuals, and the value ranges of i and j are the same and are all [1, N];
The crossing operation adopts a real number crossing method, and the kth individual chromosome akAnd the l-th chromosome a1The interleaving operation method at j bit is shown in equation (9):
in the above formula, akjFor chromosomes produced after crossover operations, b is [0,1 ]]A random number in between;
mutation operation for selecting jth gene a of ith individualijPerforming mutation, wherein the mutation operation is shown as a formula (10):
in the above formula, amaxIs gene aijUpper bound of aminIs gene aijThe lower bound of (c); f (g) r2(1-g/Gmax)2;r2Is a random number; g is the current iteration number; gmaxThe maximum number of evolutions; r is [0,1 ]]Random number in between.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910206025.XA CN110084106A (en) | 2019-03-19 | 2019-03-19 | Microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910206025.XA CN110084106A (en) | 2019-03-19 | 2019-03-19 | Microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110084106A true CN110084106A (en) | 2019-08-02 |
Family
ID=67413258
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910206025.XA Pending CN110084106A (en) | 2019-03-19 | 2019-03-19 | Microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110084106A (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111239549A (en) * | 2020-02-18 | 2020-06-05 | 国网信通亿力科技有限责任公司 | Power distribution fault rapid positioning method based on discrete wavelet transform |
CN111366814A (en) * | 2020-03-31 | 2020-07-03 | 上海电力大学 | Power grid fault diagnosis method based on multi-source data and multi-dimensional fault coding space |
CN111679158A (en) * | 2020-08-04 | 2020-09-18 | 国网内蒙古东部电力有限公司呼伦贝尔供电公司 | Power distribution network fault identification method based on synchronous measurement data similarity |
CN111812435A (en) * | 2020-06-28 | 2020-10-23 | 国网新源控股有限公司回龙分公司 | Method for diagnosing fault of static frequency converter based on BP neural network |
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 |
CN112257335A (en) * | 2020-10-10 | 2021-01-22 | 西南交通大学 | Oil-immersed transformer fault diagnosis method combining PNN and SVM |
CN112255495A (en) * | 2020-09-10 | 2021-01-22 | 西安理工大学 | Micro-grid high-resistance fault detection method |
CN112414446A (en) * | 2020-11-02 | 2021-02-26 | 南昌智能新能源汽车研究院 | Data-driven transmission sensor fault diagnosis method |
CN113075469A (en) * | 2020-01-06 | 2021-07-06 | 株洲中车时代电气股份有限公司 | Inversion overcurrent fault diagnosis method, device and system |
CN113516066A (en) * | 2021-07-05 | 2021-10-19 | 内蒙古工业大学 | Power quality disturbance signal classification method and device, storage medium and electronic equipment |
CN113702862A (en) * | 2021-08-27 | 2021-11-26 | 广东省科学院电子电器研究所 | Power supply state panoramic monitoring method and device based on cloud deployment |
CN116520990A (en) * | 2023-04-28 | 2023-08-01 | 暨南大学 | Sign language identification method and system based on lightweight neural network and glove |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102570392A (en) * | 2012-01-17 | 2012-07-11 | 上海电力学院 | Method for identifying exciting inrush current of transformer based on improved probability neural network |
CN105914736A (en) * | 2016-05-05 | 2016-08-31 | 河海大学 | Inverter power supply modeling method in power distribution network short circuit |
US20170052060A1 (en) * | 2014-04-24 | 2017-02-23 | Alstom Transport Technologies | Method and system for automatically detecting faults in a rotating shaft |
-
2019
- 2019-03-19 CN CN201910206025.XA patent/CN110084106A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102570392A (en) * | 2012-01-17 | 2012-07-11 | 上海电力学院 | Method for identifying exciting inrush current of transformer based on improved probability neural network |
US20170052060A1 (en) * | 2014-04-24 | 2017-02-23 | Alstom Transport Technologies | Method and system for automatically detecting faults in a rotating shaft |
CN105914736A (en) * | 2016-05-05 | 2016-08-31 | 河海大学 | Inverter power supply modeling method in power distribution network short circuit |
Non-Patent Citations (4)
Title |
---|
Z HUANG, Z WANG AND H ZHANG: "A diagnosis algorithm for multiple open-circuited faults of microgrid inverters based on main fault component analysis", 《EEE TRANSACTIONS ON ENERGY CONVERSION》 * |
刘琦: "微网多逆变器并联的故障诊断研究", 《中国优秀硕士学位全文数据库工程科技Ⅱ辑》 * |
周沙: "基于概率神经网络的变压器局部放电模式识别研究", 《中国优秀硕士学位全文数据库工程科技Ⅱ辑》 * |
高杰: "基于优化PNN网络的变压器故障诊断研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113075469A (en) * | 2020-01-06 | 2021-07-06 | 株洲中车时代电气股份有限公司 | Inversion overcurrent fault diagnosis method, device and system |
CN111239549A (en) * | 2020-02-18 | 2020-06-05 | 国网信通亿力科技有限责任公司 | Power distribution fault rapid positioning method based on discrete wavelet transform |
CN111366814A (en) * | 2020-03-31 | 2020-07-03 | 上海电力大学 | Power grid fault diagnosis method based on multi-source data and multi-dimensional fault coding space |
CN111812435A (en) * | 2020-06-28 | 2020-10-23 | 国网新源控股有限公司回龙分公司 | Method for diagnosing fault of static frequency converter based on BP neural network |
CN111679158A (en) * | 2020-08-04 | 2020-09-18 | 国网内蒙古东部电力有限公司呼伦贝尔供电公司 | Power distribution network fault identification method based on synchronous measurement data similarity |
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 |
CN112068033B (en) * | 2020-09-02 | 2024-03-26 | 河北工业大学 | On-line identification method for open-circuit faults of inverter power tube based on 1/6 period current |
CN112255495A (en) * | 2020-09-10 | 2021-01-22 | 西安理工大学 | Micro-grid high-resistance fault detection method |
CN112255495B (en) * | 2020-09-10 | 2023-10-24 | 西安理工大学 | Micro-grid high-resistance fault detection method |
CN112257335A (en) * | 2020-10-10 | 2021-01-22 | 西南交通大学 | Oil-immersed transformer fault diagnosis method combining PNN and SVM |
CN112414446A (en) * | 2020-11-02 | 2021-02-26 | 南昌智能新能源汽车研究院 | Data-driven transmission sensor fault diagnosis method |
CN113516066A (en) * | 2021-07-05 | 2021-10-19 | 内蒙古工业大学 | Power quality disturbance signal classification method and device, storage medium and electronic equipment |
CN113516066B (en) * | 2021-07-05 | 2023-08-08 | 内蒙古工业大学 | Power quality disturbance signal classification method and device, storage medium and electronic equipment |
CN113702862A (en) * | 2021-08-27 | 2021-11-26 | 广东省科学院电子电器研究所 | Power supply state panoramic monitoring method and device based on cloud deployment |
CN116520990A (en) * | 2023-04-28 | 2023-08-01 | 暨南大学 | Sign language identification method and system based on lightweight neural network and glove |
CN116520990B (en) * | 2023-04-28 | 2023-11-24 | 暨南大学 | Sign language identification method and system based on lightweight neural network and glove |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110084106A (en) | Microgrid inverter method for diagnosing faults based on wavelet transformation and probabilistic neural network | |
Vaish et al. | Machine learning applications in power system fault diagnosis: Research advancements and perspectives | |
CN108120903B (en) | A kind of low-current single-phase earth fault line selection method based on pulse nerve membranous system | |
WO2018045689A1 (en) | Decision tree support vector machine (svm) fault diagnosis method for photovoltaic diode clamped three-level inverter | |
CN104616061B (en) | Island detection method based on wavelet packet logarithmic energy entropy and genetic algorithm optimization | |
CN110542819A (en) | transformer fault type diagnosis method based on semi-supervised DBNC | |
CN112485587B (en) | Layered positioning method for fault section of distribution network containing distributed photovoltaic | |
CN105203869A (en) | Microgrid island detection method based on extreme learning machine | |
CN116316637A (en) | Dynamic topology identification method, system, equipment and storage medium for power distribution network | |
Wang et al. | Data-driven fault detection and isolation in DC microgrids without prior fault data: A transfer learning approach | |
CN114330571B (en) | Power system dominant instability mode identification method and system based on transfer learning | |
CN115688574A (en) | Transient stability online evaluation method and system for power system | |
CN116131313A (en) | Explanatory analysis method for association relation between characteristic quantity and transient power angle stability | |
Samal et al. | Data-mining model-based enhanced differential relaying scheme for microgrids | |
Aazami et al. | Deep neural networks based method to islanding detection for multi-sources microgrid | |
CN111814394B (en) | Power system safety assessment method based on correlation and redundancy detection | |
Gong et al. | Transient stability assessment of electric power system based on voltage phasor and cnn-lstm | |
Biswal et al. | A combined approach of empirical wavelet transform and artificial neural network based high impedance fault classification in micro-grid system | |
Jiang et al. | Fault Diagnosis of Power Electronic Circuits Based on Adaptive Simulated Annealing Particle Swarm Optimization. | |
Ren et al. | Fault Diagnosis of UHVDC Transmission System Based on Gated Recurrent Unit | |
Fan et al. | Ultra-Short-Term Wind Power Forecasting Based on Fluctuation Pattern Clustering and Prediction | |
Patel et al. | Wavelet and Machine learning based approach for Fault classification in AC Micro-grid system | |
Li et al. | A Sensitive Fault Detection Method for AC-DC Hybrid Distribution Network Based on Intelligent Learning | |
Zhang et al. | A Critical Review of Data-Driven Transient Stability Assessment of Power Systems: Principles, Prospects and Challenges. Energies 2021, 14, 7238 | |
Bhuyan et al. | Fault Classification in a DG Connected Power System using Artificial 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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190802 |
|
RJ01 | Rejection of invention patent application after publication |