CN112597816A - Electric energy quality signal feature extraction method - Google Patents

Electric energy quality signal feature extraction method Download PDF

Info

Publication number
CN112597816A
CN112597816A CN202011418852.4A CN202011418852A CN112597816A CN 112597816 A CN112597816 A CN 112597816A CN 202011418852 A CN202011418852 A CN 202011418852A CN 112597816 A CN112597816 A CN 112597816A
Authority
CN
China
Prior art keywords
atom
disturbance
power quality
quality signal
fundamental wave
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
Application number
CN202011418852.4A
Other languages
Chinese (zh)
Other versions
CN112597816B (en
Inventor
袁莉芬
刘韬
何怡刚
李兵
佐磊
尹柏强
张鹤鸣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hefei University of Technology
Original Assignee
Hefei University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hefei University of Technology filed Critical Hefei University of Technology
Priority to CN202011418852.4A priority Critical patent/CN112597816B/en
Publication of CN112597816A publication Critical patent/CN112597816A/en
Application granted granted Critical
Publication of CN112597816B publication Critical patent/CN112597816B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/02—Preprocessing
    • G06F2218/04—Denoising
    • 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
    • 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

Abstract

The invention relates to a method for extracting characteristics of a power quality signal, which comprises the following steps: constructing a fundamental wave atom library; acquiring a power quality signal of a power grid, carrying out sparse decomposition on the power quality signal on a fundamental wave atomic library, and extracting fundamental wave signal characteristics; constructing five electric energy quality signal atom libraries according to the extracted fundamental wave signal characteristics: namely a similar fundamental wave atom library, a pulse atom library, a harmonic atom library, a flicker atom library and an oscillation atom library; and (3) carrying out sparse decomposition on the electric energy quality signal with the fundamental wave signal characteristics extracted in the step (2) on the atom library constructed in the step (3) to extract the electric energy quality signal characteristics. According to the invention, the sampled power quality signal does not need to be additionally processed, the processing speed is high, and the real-time analysis of the power quality is convenient; various disturbance signal characteristics existing in the power quality can be accurately, quickly and quantitatively extracted; noise signals can be effectively filtered.

Description

Electric energy quality signal feature extraction method
Technical Field
The invention relates to the technical field of power quality analysis and monitoring, in particular to a power quality signal feature extraction method.
Background
In recent years, with the continuous progress of science and technology and the rapid improvement of economic level, the development level of the power industry enters a new stage, and the structure of a power grid and the type of a power load are greatly changed. The progress brings various conveniences, and causes factors of deterioration of electric energy quality to be increased continuously, such as continuous development of extra-high voltage alternating current and direct current transmission, a micro-grid and various renewable energy sources for power generation, and the structure of the grid becomes more complex; the large amount of nonlinear and impact loads in power electronic equipment, electric locomotives, motor train units, arc equipment, charging stations and the like in the power system makes the form of power consumption more complicated. The complexity of the power grid structure and the power consumption form brings a large amount of power quality problems, seriously influences the development level of people's life and economy, and arouses the general attention of power departments, scholars and vast power consumers.
In order to reduce the influence caused by the power quality problem and improve the power quality, the disturbance signal acquired by the power quality monitoring device needs to be analyzed and processed, and the process generally comprises disturbance analysis, feature extraction, data compression, disturbance classification identification, disturbance parameter identification and the like. The disturbance analysis and the feature extraction are the basis of the subsequent processing process, and the advanced functions of classification, positioning and the like can be better completed only by effectively analyzing the power quality signal and extracting the feature parameters. With the fact that an atomic decomposition technology becomes a hotspot in the field of signal processing in recent years, the method decomposes signals on a set of over-complete non-orthogonal bases, achieves adaptive, more flexible and concise representation and processing of the signals, and improves the simplicity and flexibility of signal expression. The most concise analytic expression of the electric energy quality signal is obtained by utilizing an atomic decomposition method, so that the characteristics of the signal can be visually embodied, and various disturbance signal characteristic parameters existing in the signal are effectively extracted. However, the conventional atomic decomposition method is realized based on a matching pursuit algorithm, and the defects of large calculation amount and long operation time limit the application of the atomic decomposition method in the electric energy quality signal analysis and the feature extraction.
Disclosure of Invention
The invention aims to provide a power quality signal feature extraction method which has higher processing speed, is convenient for real-time analysis of power quality, accurately, quickly and quantitatively extracts various disturbance signal features existing in the power quality and effectively filters noise signals.
In order to achieve the purpose, the invention adopts the following technical scheme: a method of extracting characteristics of a power quality signal, the method comprising the sequential steps of:
(1) constructing a fundamental wave atom library;
(2) acquiring a power quality signal of a power grid, carrying out sparse decomposition on the power quality signal on a fundamental wave atomic library, and extracting fundamental wave signal characteristics;
(3) constructing five electric energy quality signal atom libraries according to the extracted fundamental wave signal characteristics: namely a similar fundamental wave atom library, a pulse atom library, a harmonic atom library, a flicker atom library and an oscillation atom library;
(4) and (3) carrying out sparse decomposition on the electric energy quality signal with the fundamental wave signal characteristics extracted in the step (2) on the atom library constructed in the step (3) to extract the electric energy quality signal characteristics.
The step (1) specifically comprises the following steps:
fundamental wave atom gγ1The expression is as follows:
Figure BDA0002821359770000021
wherein i1∈[0，N]，i1The initial value of (a) is any random number from 0 to N; n represents the length of the power quality signal f to be measured; j is a function of1∈[0，N]，j1The initial value of (a) is any random number from 0 to N; f. ofNIs the maximum frequency extracted by the atom library, the value of which is specified by the national power quality standardThe maximum frequency of the fundamental wave allowed by the power system is set to be 50.5 Hz; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal to be measured;
Kγ1is a normalization factor, whose value is:
Figure BDA0002821359770000022
wherein, g1Is the fundamental atom with a normalized coefficient of 1.
And (3) collecting the power quality signals of the power grid in the step (2) at industrial loads, residential loads, transformer substations, photovoltaic power stations, wind power plants, railway traction stations, electric vehicle charging piles and distributed power supply grid-connected positions.
The extracting of the fundamental wave signal features in the step (2) specifically comprises the following steps:
(2a) initialization setting: initial residual signal
Figure BDA0002821359770000023
Equal to the power quality signal f to be measured;
(2b) selecting the best atom g matched with the power quality signal f to be measured in a fundamental wave atom library by utilizing an LSA algorithmγ1(opt)：
Figure BDA0002821359770000024
In the formula, N represents the length of the power quality signal f to be measured;
Figure BDA0002821359770000025
and
Figure BDA0002821359770000026
i are each determined by the LSA algorithm1、j1The optimal solution of (2); t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
initial residual signal
Figure BDA0002821359770000027
The best matching atom satisfies:
Figure BDA0002821359770000031
in the formula, gγ1Representing atoms in a fundamental atom library;
(2c) updating residual signals:
Figure BDA0002821359770000032
(2d) calculating the amplitude of the fundamental wave:
Figure BDA0002821359770000033
extracted fundamental component:
Figure BDA0002821359770000034
(2e) output extracted fundamental frequency:
Figure BDA0002821359770000035
outputting the extracted fundamental wave phase:
Figure BDA0002821359770000036
and outputting a waveform diagram.
The step (3) specifically comprises the following steps:
(3a) class fundamental atom library:
fundamental-like atomic gγ2The expression is as follows:
Figure BDA0002821359770000037
in the formula, u (t) is a step function, and N is the length of the power quality signal f to be measured; f. of1The fundamental frequency extracted by a fundamental atom library is utilized; j is a function of2∈[0，N]，j2The initial value of (a) is any random number from 0 to N; n iss2，ne2∈[0，N]，ns2And ne2The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ2is a normalization factor, whose value is:
Figure BDA0002821359770000038
in the formula, g2The normalized coefficient is a fundamental wave-like atom with 1;
(3b) pulsed atom library
Pulse atom gγ3The expression is as follows:
Figure BDA0002821359770000041
in the formula, u (t) is a step function, and N is the length of the power quality signal f to be measured; n iss3，ne3∈[0，N]，ns3And ne3The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ3is a normalization factor, whose value is:
Figure BDA0002821359770000042
g3is a pulse with a normalized coefficient of 1Punching atoms;
(3c) harmonic atom library
Harmonic atom gγ4The expression is as follows:
Figure BDA0002821359770000043
in the formula, N is the length of the power quality signal f to be measured; i.e. i4∈[0，N]，i4The initial value of (a) is any random number from 0 to N; j is a function of4∈[0，N]，j4The initial value of (a) is any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ4is a normalization factor, whose value is:
Figure BDA0002821359770000044
g4harmonic atoms with a normalized coefficient of 1;
(3d) pool of mutator atoms
Flash atom gγ5The expression is as follows:
Figure BDA0002821359770000045
wherein u (t) is a step function, fsThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N; f. of1And
Figure BDA0002821359770000051
respectively frequency and phase extracted by using fundamental atom libraryN5The value of (a) is set according to the voltage fluctuation and flicker frequency range specified by the electric power system electromagnetic phenomenon parameters and classification standards formulated by IEEE, and the reference value range fN5≥25Hz；i5∈[0，N]，i5The initial value of (a) is any random number from 0 to N; j is a function of5∈[0，N]，j5The initial value of (a) is any random number from 0 to N; n iss5，ne5∈[0，N]，ns5And ne5The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]；
Kγ5Is a normalization factor, whose value is:
Figure BDA0002821359770000052
g5is a flickering atom with a normalized coefficient of 1;
(3e) oscillating atom library
Oscillating atom gγ6The expression is as follows:
Figure BDA0002821359770000053
wherein u (t) is a step function, fsThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N; i.e. i6∈[0，N]，i6The initial value of (a) is any random number from 0 to N; j is a function of6∈[0，N]，j6The initial value of (a) is any random number from 0 to N; k is an element of [0, N ]]The initial value of k is any random number from 0 to N; n iss6，ne6∈[0，N]，ns6And ne6The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]；
Kγ6Is a normalization factor, whose value is:
Figure BDA0002821359770000054
g6is an oscillating atom with a normalized coefficient of 1.
The step (4) of extracting the electric energy quality signal features specifically comprises the following steps:
(4a) sequentially selecting a class fundamental wave atom library, a pulse atom library and a harmonic atom library by using an LSA algorithmSelecting the best matching atom g of class fundamental wave from the flickering atom library and the oscillation atom libraryγ2(opt)Best matching pulse atom gγ3(opt)The harmonic best matching atom gγ4(opt)The flash best matching atom gγ5(opt)And oscillation best matching atom gγ6(opt)；
Figure BDA0002821359770000061
Figure BDA0002821359770000062
Figure BDA0002821359770000063
Figure BDA0002821359770000064
Figure BDA0002821359770000065
In the formula (f)sThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N;
Figure BDA0002821359770000066
and
Figure BDA0002821359770000067
j is respectively calculated by LSA algorithm2、ns2And ne2The current optimal solution of;
Figure BDA0002821359770000068
and
Figure BDA0002821359770000069
n being respectively calculated by LSA algorithms3And ne3The current optimal solution of;
Figure BDA00028213597700000610
and
Figure BDA00028213597700000611
i are each determined by the LSA algorithm4And j4The current optimal solution of;
Figure BDA00028213597700000612
and
Figure BDA00028213597700000613
i are each determined by the LSA algorithm5、j5、ns5And ne5The current optimal solution of;
Figure BDA00028213597700000614
k*、
Figure BDA00028213597700000615
and
Figure BDA00028213597700000616
i are each determined by the LSA algorithm6、j6、k、ns6And ne6The current optimal solution of; t is a time variable, t belongs to [0, N/f ]s]；fN5The value of (a) is set according to the voltage fluctuation and flicker frequency range specified by the electric power system electromagnetic phenomenon parameters and classification standards formulated by IEEE, and the reference value range fN5≥25Hz；；f1And
Figure BDA00028213597700000617
respectively extracting the frequency and the phase by using a fundamental wave atom library;
the best matching atom satisfies:
Figure BDA00028213597700000618
in the formula, gγzRepresenting atoms in a selected atom pool;
Figure BDA00028213597700000619
denotes a residual signal after extracting the disturbance signal, z is 2, 3, 4, 5, 6
(4b) Updating residual signals in sequence:
Figure BDA00028213597700000620
(4c) calculating the amplitude of the disturbance:
Figure BDA00028213597700000621
Figure BDA00028213597700000622
Figure BDA0002821359770000071
Figure BDA0002821359770000072
Figure BDA0002821359770000073
in the formula, A2To A6Sequentially representing a similar fundamental wave disturbance amplitude, a pulse disturbance amplitude, a harmonic disturbance amplitude, a flicker disturbance amplitude and an oscillation disturbance amplitude;
extracted disturbance component:
Figure BDA0002821359770000074
Figure BDA0002821359770000075
Figure BDA0002821359770000076
Figure BDA0002821359770000077
Figure BDA0002821359770000078
in the above formula, V2To V6Sequentially representing a similar fundamental wave disturbance component, a pulse disturbance component, a harmonic disturbance component, a flicker disturbance component and an oscillation disturbance component;
(4d) outputting the extracted disturbance characteristic parameters
Outputting the extracted harmonic disturbance frequency f4Flicker disturbance frequency f5And an oscillation disturbance frequency f6：
Figure BDA0002821359770000079
Outputting extracted similar fundamental wave disturbance initial phase
Figure BDA00028213597700000710
Harmonic disturbance initial phase
Figure BDA00028213597700000711
Initial phase of flicker disturbance
Figure BDA00028213597700000712
And the initial phase of the oscillation disturbance
Figure BDA00028213597700000713
Figure BDA00028213597700000714
Outputting the extracted similar fundamental wave disturbance starting time ts2Start time t of pulse disturbances3Start time t of flicker disturbances5And oscillation disturbance start time ts6：
Figure BDA0002821359770000081
Outputting the extracted similar fundamental wave disturbance termination time te2Pulse disturbance termination time te3End time t of flicker disturbancee5And oscillation disturbance termination time te6：
Figure BDA0002821359770000082
Outputting the extracted oscillation disturbance attenuation coefficient rho6：
Figure BDA0002821359770000083
And outputting the extracted waveform diagram of the disturbance.
According to the technical scheme, the beneficial effects of the invention are as follows: firstly, the sampled power quality signals do not need to be additionally processed, the processing speed is high, and the real-time analysis on the power quality is convenient; secondly, various disturbance signal characteristics existing in the power quality can be accurately, rapidly and quantitatively extracted; thirdly, the invention can effectively filter noise signals.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a diagram showing an original signal, an extracted fundamental wave signal, a voltage ramp, and a total residual signal containing noise in sequence;
FIG. 3 is a diagram showing an original signal, an extracted fundamental wave signal, a voltage sag, and a total residual signal containing noise in sequence;
FIG. 4 is a diagram illustrating an original signal, an extracted fundamental wave signal, a voltage interruption, and a total residual signal with noise in sequence;
FIG. 5 is a diagram illustrating an original signal, an extracted fundamental wave signal, a harmonic wave, and a total residual signal containing noise in sequence;
FIG. 6 is a diagram illustrating an original signal, an extracted fundamental wave signal, inter-harmonics, and a total residual signal with noise in sequence;
FIG. 7 is a diagram illustrating an original signal, an extracted fundamental wave signal, a voltage spike, and a total residual signal with noise in sequence;
FIG. 8 is a diagram illustrating an original signal, an extracted fundamental wave signal, a voltage shear mark, and a total residual signal containing noise in sequence;
FIG. 9 is a diagram illustrating an original signal, an extracted fundamental wave signal, a voltage flicker, and a total residual signal with noise in sequence;
FIG. 10 is a diagram illustrating an original signal, an extracted fundamental wave signal, ringing, and a total residual signal with noise in sequence;
FIG. 11 is a diagram illustrating an original signal, an extracted fundamental wave signal, a divergent oscillation, and a total residual signal with noise in sequence;
FIG. 12 is a diagram illustrating an original signal, an extracted fundamental wave signal, a short-time harmonic, and a total residual signal with noise in sequence;
FIG. 13 is a schematic diagram of an original signal, an extracted fundamental signal, a voltage ramp, harmonics, ringing, and a noisy total residual signal in sequence;
FIG. 14 is a diagram showing an original signal, an extracted fundamental wave signal, a voltage sag, an inter-harmonic, a voltage shear mark, and a total residual signal containing noise in sequence;
fig. 15 is a schematic diagram of the original signal, the extracted fundamental wave signal, the voltage interruption, the voltage flicker, the short-time harmonics and the total residual signal with noise in sequence.
Detailed Description
As shown in fig. 1, a method for extracting characteristics of a power quality signal includes the following steps:
(1) constructing a fundamental wave atom library;
(2) acquiring a power quality signal of a power grid, carrying out sparse decomposition on the power quality signal on a fundamental wave atomic library, and extracting fundamental wave signal characteristics;
(3) constructing five electric energy quality signal atom libraries according to the extracted fundamental wave signal characteristics: namely a similar fundamental wave atom library, a pulse atom library, a harmonic atom library, a flicker atom library and an oscillation atom library;
(4) and (3) carrying out sparse decomposition on the electric energy quality signal with the fundamental wave signal characteristics extracted in the step (2) on the atom library constructed in the step (3) to extract the electric energy quality signal characteristics.
The step (1) specifically comprises the following steps:
fundamental wave atom gγ1The expression is as follows:
Figure BDA0002821359770000091
wherein i1∈[0，N]，i1The initial value of (a) is any random number from 0 to N; n represents the length of the power quality signal f to be measured; j is a function of1∈[0，N]，j1The initial value of (a) is any random number from 0 to N; f. ofNThe maximum frequency extracted by the atom library is set according to the maximum frequency of 50.5Hz of the fundamental wave allowed by the power system specified by the national power quality standard; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal to be measured;
Kγ1is a normalization factor, whose value is:
Figure BDA0002821359770000092
wherein, g1Is the fundamental atom with a normalized coefficient of 1.
And (3) collecting the power quality signals of the power grid in the step (2) at industrial loads, residential loads, transformer substations, photovoltaic power stations, wind power plants, railway traction stations, electric vehicle charging piles and distributed power supply grid-connected positions.
In order to reflect the diversity of the electric energy quality signals as much as possible, MATLAB software is adopted to generate 14 disturbance signal models, wherein 11 single disturbances and 3 composite disturbances are adopted, each disturbance signal is superposed with 30dB of white Gaussian noise, and the signal sampling frequency fs3.2kHz, sample point N1024.
The 11 single perturbations are: voltage sag, voltage interruption, harmonics, inter-harmonics, voltage spikes, voltage shear marks, voltage flicker, ringing, divergent oscillation, transient harmonics, which are respectively recorded as: c1, C2 … … C11. The 'and' connection is used between two composite single disturbances in the composite disturbance, for example, a composite disturbance signal is composed of voltage rising, harmonic waves and voltage spikes and is marked as C1& C4& C6. The 3 generated composite perturbations were C1& C4& C9, C2& C5& C7, and C3& C8& C11.
The extracting of the fundamental wave signal features in the step (2) specifically comprises the following steps:
(2a) initialization setting: initial residual signal
Figure BDA0002821359770000101
Equal to the power quality signal f to be measured;
(2b) selecting the best atom g matched with the power quality signal f to be measured in a fundamental wave atom library by utilizing an LSA algorithmγ1(opt)：
Figure BDA0002821359770000102
In the formula, N represents the length of the power quality signal f to be measured;
Figure BDA0002821359770000103
and
Figure BDA0002821359770000104
i are each determined by the LSA algorithm1、j1The optimal solution of (2); t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
initial residual signal
Figure BDA0002821359770000105
The best matching atom satisfies:
Figure BDA0002821359770000106
in the formula, gγ1Representing atoms in a fundamental atom library;
(2c) updating residual signals:
Figure BDA0002821359770000107
(2d) calculating the amplitude of the fundamental wave:
Figure BDA0002821359770000108
extracted fundamental component:
Figure BDA0002821359770000109
(2e) output extracted fundamental frequency:
Figure BDA00028213597700001010
outputting the extracted fundamental wave phase:
Figure BDA0002821359770000111
and outputting a waveform diagram.
The step (3) specifically comprises the following steps:
(3a) class fundamental atom library:
fundamental-like atomic gγ2The expression is as follows:
Figure BDA0002821359770000112
in the formula, u (t) is a step function, and N is the length of the power quality signal f to be measured; f. of1The fundamental frequency extracted by a fundamental atom library is utilized; j is a function of2∈[0，N]，j2The initial value of (a) is any random number from 0 to N; n iss2，ne2∈[0，N]，ns2And ne2The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ2is a normalization factor, whose value is:
Figure BDA0002821359770000113
in the formula, g2The normalized coefficient is a fundamental wave-like atom with 1;
(3b) pulsed atom library
Pulse atom gγ3The expression is as follows:
Figure BDA0002821359770000114
in the formula, u (t) is a step function, and N is the length of the power quality signal f to be measured; n iss3，ne3∈[0，N]，ns3And ne3The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ3is a normalization factor, whose value is:
Figure BDA0002821359770000115
g3is a pulse atom with a normalized coefficient of 1;
(3c) harmonic atom library
Harmonic atom gγ4The expression is as follows:
Figure BDA0002821359770000121
in the formula, N is the length of the power quality signal f to be measured; i.e. i4∈[0，N]，i4The initial value of (a) is any random number from 0 to N; j is a function of4∈[0，N]，j4The initial value of (a) is any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ4is a normalization factor, whose value is:
Figure BDA0002821359770000122
g4harmonic atoms with a normalized coefficient of 1;
(3d) pool of mutator atoms
Flash atom gγ5The expression is as follows:
Figure BDA0002821359770000123
wherein u (t) is a step function, fsThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N; f. of1And
Figure BDA0002821359770000124
respectively frequency and phase extracted by using fundamental atom libraryN5Power system with value of (d) being set according to IEEESetting the electromagnetic phenomenon parameters and the voltage fluctuation and flicker frequency range specified by the classification standard, and referring to the value range fN5≥25Hz；i5∈[0，N]，i5The initial value of (a) is any random number from 0 to N; j is a function of5∈[0，N]，j5The initial value of (a) is any random number from 0 to N; n iss5，ne5∈[0，N]，ns5And ne5The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]；
Kγ5Is a normalization factor, whose value is:
Figure BDA0002821359770000125
g5is a flickering atom with a normalized coefficient of 1;
(3e) oscillating atom library
Oscillating atom gγ6The expression is as follows:
Figure BDA0002821359770000131
wherein u (t) is a step function, fsThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N; i.e. i6∈[0，N]，i6The initial value of (a) is any random number from 0 to N; j is a function of6∈[0，N]，j6The initial value of (a) is any random number from 0 to N; k is an element of [0, N ]]The initial value of k is any random number from 0 to N; n iss6，ne6∈[0，N]，ns6And ne6The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]；
Kγ6Is a normalization factor, whose value is:
Figure BDA0002821359770000132
g6as oscillating atoms with a normalized coefficient of 1。
The step (4) of extracting the electric energy quality signal features specifically comprises the following steps:
(4a) selecting the best matching atom g of class fundamental wave from class fundamental wave atom library, pulse atom library, harmonic atom library, flicker atom library and oscillation atom library in turn by using LSA algorithmγ2(opt)Best matching pulse atom gγ3(opt)The harmonic best matching atom gγ4(opt)The flash best matching atom gγ5(opt)And oscillation best matching atom gγ6(opt)；
Figure BDA0002821359770000133
Figure BDA0002821359770000134
Figure BDA0002821359770000135
Figure BDA0002821359770000136
Figure BDA0002821359770000137
In the formula (f)sThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N;
Figure BDA0002821359770000138
and
Figure BDA0002821359770000139
j is respectively calculated by LSA algorithm2、ns2And ne2The current optimal solution of;
Figure BDA00028213597700001310
and
Figure BDA00028213597700001311
n being respectively calculated by LSA algorithms3And ne3The current optimal solution of;
Figure BDA0002821359770000141
and
Figure BDA0002821359770000142
i are each determined by the LSA algorithm4And j4The current optimal solution of;
Figure BDA0002821359770000143
and
Figure BDA0002821359770000144
i are each determined by the LSA algorithm5、j5、ns5And ne5The current optimal solution of;
Figure BDA0002821359770000145
k*、
Figure BDA0002821359770000146
and
Figure BDA0002821359770000147
i are each determined by the LSA algorithm6、j6、k、ns6And ne6The current optimal solution of; t is a time variable, t belongs to [0, N/f ]s]；fN5The value of (a) is set according to the voltage fluctuation and flicker frequency range specified by the electric power system electromagnetic phenomenon parameters and classification standards formulated by IEEE, and the reference value range fN5≥25Hz；；f1And
Figure BDA0002821359770000148
respectively extracted by using fundamental atom libraryFrequency and phase;
the best matching atom satisfies:
Figure BDA0002821359770000149
in the formula, gγzRepresenting atoms in a selected atom pool;
Figure BDA00028213597700001410
denotes a residual signal after extracting the disturbance signal, z is 2, 3, 4, 5, 6
(4b) Updating residual signals in sequence:
Figure BDA00028213597700001411
(4c) calculating the amplitude of the disturbance:
Figure BDA00028213597700001412
Figure BDA00028213597700001413
Figure BDA00028213597700001414
Figure BDA00028213597700001415
Figure BDA00028213597700001416
in the formula, A2To A6Sequentially representing the similar fundamental wave disturbance amplitude, the pulse disturbance amplitude, the harmonic disturbance amplitude, the flicker disturbance amplitude andan oscillation disturbance amplitude;
extracted disturbance component:
Figure BDA00028213597700001417
Figure BDA00028213597700001418
Figure BDA00028213597700001419
Figure BDA0002821359770000151
Figure BDA0002821359770000152
in the above formula, V2To V6Sequentially representing a similar fundamental wave disturbance component, a pulse disturbance component, a harmonic disturbance component, a flicker disturbance component and an oscillation disturbance component;
(4d) outputting the extracted disturbance characteristic parameters
Outputting the extracted harmonic disturbance frequency f4Flicker disturbance frequency f5And an oscillation disturbance frequency f6：
Figure BDA0002821359770000153
Outputting extracted similar fundamental wave disturbance initial phase
Figure BDA0002821359770000154
Harmonic disturbance initial phase
Figure BDA0002821359770000155
Initial phase of flicker disturbance
Figure BDA0002821359770000156
And the initial phase of the oscillation disturbance
Figure BDA0002821359770000157
Figure BDA0002821359770000158
Outputting the extracted similar fundamental wave disturbance starting time ts2Start time t of pulse disturbances3Start time t of flicker disturbances5And oscillation disturbance start time ts6：
Figure BDA0002821359770000159
Outputting the extracted similar fundamental wave disturbance termination time te2Pulse disturbance termination time te3End time t of flicker disturbancee5And oscillation disturbance termination time te6：
Figure BDA00028213597700001510
Outputting the extracted oscillation disturbance attenuation coefficient rho6：
Figure BDA00028213597700001511
And outputting the extracted waveform diagram of the disturbance.
When the characteristics of the power quality signal are extracted, firstly, similar fundamental wave disturbance is extracted, and whether the similar fundamental wave disturbance exists in the collected signal to be detected is judged according to an output oscillogram: if the fundamental wave-like disturbance exists, the fundamental wave component extracted in the step (2) is not the true fundamental wave component because the frequency of the fundamental wave-like disturbance is the same as that of the fundamental wave, and the fundamental wave component and the fundamental wave-like disturbance need to be subjected toThe amplitude of the motion is corrected, and the corrected fundamental amplitude
Figure BDA00028213597700001512
VbasicThe amplitude of the fundamental wave component can be directly obtained by using an electric energy quality acquisition device, and the modified similar fundamental wave disturbance amplitude
Figure BDA0002821359770000161
The residual error is re-calculated,
Figure BDA0002821359770000162
when in use
Figure BDA0002821359770000163
When the voltage is positive, the similar fundamental wave disturbance is voltage temporary rise; when in use
Figure BDA0002821359770000164
When the amplitude of the similar fundamental wave disturbance is a negative value, the voltage is interrupted when the amplitude of the similar fundamental wave disturbance is between 0.9 and 1 time of the amplitude of the fundamental wave component, otherwise, the voltage is temporarily dropped; is updated again
Figure BDA0002821359770000165
Repeatedly extracting similar fundamental wave disturbance;
if not, extracting pulse disturbance;
judging whether the acquired signal to be detected has pulse disturbance according to the output oscillogram: if present, then
Figure BDA0002821359770000166
When the voltage is positive, the pulse disturbance is a voltage peak; when in use
Figure BDA0002821359770000167
When the value is negative, the pulse disturbance is a voltage peak; updating
Figure BDA0002821359770000168
Repeatedly extracting pulse disturbance;
if not, extracting harmonic or inter-harmonic disturbance;
judging whether harmonic or inter-harmonic disturbance exists in the acquired signal to be detected according to the output oscillogram:
if the harmonic waves exist, the disturbance is harmonic waves when the output frequency is an integral multiple of the fundamental wave, and the disturbance is inter-harmonic waves when the output frequency is a non-integral multiple; updating
Figure BDA0002821359770000169
Repeatedly extracting harmonic or inter-harmonic disturbance;
if not, extracting flicker disturbance;
judging whether flicker disturbance exists in the acquired signal to be detected according to the output oscillogram:
if so, update
Figure BDA00028213597700001610
Repeatedly extracting flicker disturbance;
if not, extracting oscillation disturbance;
judging whether the acquired signal to be detected has oscillation disturbance according to the output oscillogram:
if so, when p6When the value is equal to 0, the oscillation disturbance is transient harmonic disturbance; when rho6When the frequency is more than 0, the oscillation disturbance is damped oscillation disturbance; when rho6When the frequency is less than 0, the oscillation disturbance is divergent oscillation disturbance; updating
Figure BDA00028213597700001611
Repeatedly extracting oscillation disturbance;
if not, the extraction process is ended.
The LSA algorithm is a novel heuristic optimization algorithm — Lightning Search Algorithm (LSA), provenance: shareef H, Ibrahim A, Mutlag A H.lightning search algorithm [ J ]. Applied Soft Computing, 2015, 36 (S1): 315-333.
The lightning searching algorithm is a meta-heuristic algorithm which is derived from natural phenomena of lightning and is based on a cascade pilot propagation mechanism. When lightning forms, fast particles called "discharges" travel through the atmosphere, creating an initial ionization path or channel by collision and forming a stepped leader. In the algorithm, it is assumed that each discharge creates a step leader and a channel, i.e. a random candidate solution representing a set of problems to be optimized. LSA is mainly achieved by mathematical model simulation of 3 discharges, i.e. a transition discharge, a space discharge and a pilot discharge.
The LSA algorithm selects the best matching atom as follows:
s2011, setting algorithm operation parameters
The population size Num and the maximum iteration number Max _ iter may be set to optimal values according to the result of repeated experiments, where the reference value Num is 50, and Max _ iter is 80; the maximum channel time max _ ch _ time is usually set to 10; the discharge body corresponds to atoms in the fundamental wave atom library, and the energy corresponds to an optimal solution; according to the atom type needing matching, determining an atom index mode and a discharge volume dimension: fundamental atomic index gamma1＝(i1，j1) Dimension dim _1 is 2; fundamental-like atomic index gamma2＝(j2，ns2，ne2) Dimension dim _2 is 3; pulse atom index gamma3＝(ns3，ne3) Dimension dim _3 is 2; harmonic atomic index gamma4＝(i4，j4) Dimension dim _4 is 2; index of atom of flash gamma5＝(i5，j5，ns5，ne5) Dimension dim _5 is 4; oscillating atomic index gamma6＝(i6，j6，k，ns6，ne6) Dimension dim _6 is 5;
the upper limit value up of the search boundary of each variable of the discharge body is N, the lower limit value low of the search boundary is 0, and when the selected atom library is an oscillation atom library, the lower limit value of the search boundary of the variable k is modified to be-N;
initial step leading tip energy setting of each discharge in the population:
Dpointd(xd_1，…，xd_dim_z)＝(rand*(up-low)+low，…，rand*(up-low)+low)
wherein z is 1, 2, 3, 4, 5, 6; d is 1, 2, 3, … Num; rand represents a random number between 0 and 1;
the fitness function is set as follows:
Figure BDA0002821359770000171
xd＝[xd_1，…，xd_dim_z]；
the initial value of the fitness of each discharge body in the population is set as 1010；
S2012, entering a main cycle, substituting the generated initial step leading tip energy of each discharge body into a fitness function to calculate a fitness value, and performing descending order arrangement on the obtained fitness values of each discharge body to determine an optimal individual and a worst individual;
s2013, the initial value of the channel time ch _ time is 0, one is added to the ch _ time every cycle, if the maximum channel time max _ ch _ time is reached, the step leading tip energy of the optimal individual in the current population is given to the worst individual, the channel time is reset, and if the maximum channel time max _ ch _ time is not reached, the next step is carried out;
s2014, updating direction and energy of discharge body
The optimal step leading tip energy obtained in the notation S2012 is Dpointbest＝(y1，…，ydim_z)；
Let Dpointtest＝DpointbestFor DpointtestUpdating the variables in (1):
y′1＝y1+direct(1)*0.005*(up-low)
Figure BDA0002821359770000181
y′dim_z＝ydim_z+direct(dim_z)*0.005*(up-low)
wherein the direction matrix direct is a random positive and negative matrix of a row of dim _ z columns;
substituting the updated result into the fitness function to obtain a new fitness value, if the new fitness value is larger than the previous fitness value, keeping the new fitness value unchanged according to the original direction, and otherwise, changing the direction into direct;
s2015, executing space discharge body and guiding discharge body model
Each step of pilot energy and DpointbestThe difference vector is marked as Dist, whether Dist is a zero vector or not is judged, if yes, the vector is updated:
Figure BDA0002821359770000182
wherein e is 1, … dim _ z;
wherein Energy is 2.05-2 exp (-5 Max)iterNow _ iter)/Max _ iter), Now _ iter is the current iteration number, and norm is a random number generated by a normal distribution function;
if Dist (e) > 0, update energy: dpointtemp(e)＝Dpointd(e)-exprnd(dist(e))；
Where exprand is a random number generated by an exponential distribution function;
if Dist (e) < 0, update energy:
Dpointtemp(e)＝Dpointd(e)-exprnd(abs(dist(e)))；
judging whether the updated energy exceeds the upper limit and the lower limit of the search boundary, if so, setting the updated energy as follows:
Dpointtemp(e)＝rand(1)*(up-low)+low；
s2016, substituting the obtained optimal step leading tip energy into a fitness function to obtain a new fitness value, comparing the new fitness value with the optimal value obtained in the step S3012, if the new fitness value is greater than the optimal value, turning to the step S2017, and otherwise, turning to the step S2018;
s2017, updating the Dpointbest＝DpointtempChecking whether the discharge body is branched:
generating a random number between 0 and 1, if the random number is less than 0.01, considering that the discharge body is forked to obtain a symmetrical channel, wherein the energy of the channel is obtained by subtracting the original channel energy from the sum of the upper limit and the lower limit of a search boundary, substituting two different energies into a fitness function, selecting a channel with a larger fitness value, and keeping the population size unchanged;
S2018, maintain DpointbestIf not, the step goes to step S2019;
s2019, when the maximum iteration times are reached, namely the termination condition of the algorithm is met, the algorithm is stopped, the optimal solution is output, and the optimal matching atom is obtained; otherwise, go to step S2012, increase iteration number and channel time, and continue the next generation search.
The following table is an electric energy quality signal model established in MATLAB, where fundamental frequency f is 50Hz, and u (t) represents a unit step function;
Figure BDA0002821359770000191
Figure BDA0002821359770000201
the following table shows characteristic parameters of C1, C2 … … C11, C1& C4& C9, C2& C5& C7 and C3& C8& C11 extracted by the method
Figure BDA0002821359770000202
Fig. 2 sequentially shows an electric energy quality signal containing only voltage sag disturbance, a fundamental component extracted without noise according to the present invention, a voltage sag component extracted without noise according to the present invention, and a final total residual signal containing noise;
fig. 3 sequentially shows an electric energy quality signal containing only voltage sag disturbance, a fundamental component without noise extracted by the present invention, a voltage sag component without noise extracted by the present invention, and a final total residual signal with noise;
FIG. 4 is a diagram of a power quality signal including only voltage interruption disturbances, a fundamental component extracted without noise according to the present invention, a voltage interruption component extracted without noise according to the present invention, and a final total residual signal with noise in this order;
fig. 5 sequentially shows an electric energy quality signal containing only harmonic disturbance, a fundamental component extracted without noise according to the present invention, a harmonic component extracted without noise according to the present invention, and a final total residual signal containing noise;
fig. 6 sequentially shows a power quality signal containing only inter-harmonic disturbance, a fundamental component extracted without noise according to the present invention, an inter-harmonic component extracted without noise according to the present invention, and a final total residual signal containing noise;
FIG. 7 is a diagram sequentially illustrating a power quality signal including only voltage spike disturbance, a fundamental component extracted without noise according to the present invention, a voltage spike component extracted without noise according to the present invention, and a final total residual signal with noise;
FIG. 8 is a graph of a power quality signal with only voltage notch disturbances, a fundamental component extracted without noise according to the present invention, a voltage notch component extracted without noise according to the present invention, and a final total residual signal with noise in that order;
FIG. 9 is a diagram sequentially showing a power quality signal containing only voltage flicker disturbance, a fundamental component extracted without noise according to the present invention, a voltage flicker component extracted without noise according to the present invention, and a final total residual signal containing noise;
fig. 10 sequentially shows a power quality signal containing only ringing disturbance, a fundamental component extracted without noise according to the present invention, a ringing component extracted without noise according to the present invention, and a final total residual signal containing noise;
fig. 11 is a diagram sequentially showing an electric energy quality signal containing only divergent oscillation disturbances, a fundamental component extracted without noise according to the present invention, a divergent oscillation component extracted without noise according to the present invention, and a final total residual signal containing noise;
fig. 12 sequentially shows an electric energy quality signal containing only transient harmonic disturbance, a fundamental component extracted without noise according to the present invention, a transient harmonic component extracted without noise according to the present invention, and a final total residual signal containing noise;
fig. 13 sequentially shows an electric energy quality signal including a voltage sag disturbance, a harmonic disturbance, and a ringing disturbance, a fundamental component extracted without noise according to the present invention, a voltage sag component extracted without noise according to the present invention, a harmonic component extracted without noise according to the present invention, a ringing component extracted without noise according to the present invention, and a final total residual signal including noise;
fig. 14 sequentially shows an electric energy quality signal including voltage sag disturbance, inter-harmonic disturbance, and voltage shear mark disturbance, a fundamental component extracted without noise according to the present invention, a voltage sag component extracted without noise according to the present invention, an inter-harmonic component extracted without noise according to the present invention, a voltage shear mark component extracted without noise according to the present invention, and a final total residual signal including noise;
fig. 15 shows in sequence a power quality signal including a voltage interruption disturbance, a voltage flicker disturbance, and a transient harmonic disturbance, a noise-free fundamental component extracted by the present invention, a noise-free voltage interruption component extracted by the present invention, a noise-free voltage flicker component extracted by the present invention, a noise-free transient harmonic component extracted by the present invention, and a final noisy total residual signal.
In summary, the method for extracting the characteristics of the power quality signal provided by the invention is realized based on the LSA improved atomic decomposition method. The LSA algorithm is a novel heuristic optimization algorithm provided based on a lightning mechanism, and has the advantages of less adjusting parameters, high convergence precision, strong global optimization capability and the like. The traditional atom decomposition method is realized based on a matching pursuit algorithm, introduces an LSA algorithm into the matching pursuit algorithm, solves the problems of large calculation amount and long operation time of the matching pursuit algorithm, accurately and quickly selects the best matching atom from an atom library, completes atom decomposition and obtains characteristic parameters and waveforms of electric energy quality signals.
The method shows good performance when extracting the characteristics of various complex composite disturbance signals, and the applicable power quality disturbance types are wider. The invention has better anti-noise capability when extracting the characteristics of the power quality signal, and the extracted fundamental wave component and various disturbance signal components do not contain noise.
The invention directly takes the signal acquired by the electric energy quality acquisition device as the original signal to be processed, greatly accelerates the processing process due to the introduction of the LSA algorithm, improves the efficiency of characteristic extraction, and ensures that the requirement of real-time analysis on the electric energy quality is met by utilizing the atomic decomposition method to extract the characteristics of the electric energy quality signal.

Claims (6)

1. A power quality signal feature extraction method is characterized by comprising the following steps: the method comprises the following steps in sequence:
(1) constructing a fundamental wave atom library;
(2) acquiring a power quality signal of a power grid, carrying out sparse decomposition on the power quality signal on a fundamental wave atomic library, and extracting fundamental wave signal characteristics;
(3) constructing five electric energy quality signal atom libraries according to the extracted fundamental wave signal characteristics: namely a similar fundamental wave atom library, a pulse atom library, a harmonic atom library, a flicker atom library and an oscillation atom library;
(4) and (3) carrying out sparse decomposition on the electric energy quality signal with the fundamental wave signal characteristics extracted in the step (2) on the atom library constructed in the step (3) to extract the electric energy quality signal characteristics.
2. The power quality signal feature extraction method according to claim 1, characterized in that: the step (1) specifically comprises the following steps:
fundamental wave atom gγ1The expression is as follows:
Figure FDA0002821359760000011
wherein i1∈[0，N]，i1The initial value of (a) is any random number from 0 to N; n represents the length of the power quality signal f to be measured; j is a function of1∈[0，N]，j1The initial value of (a) is any random number from 0 to N; f. ofNThe maximum frequency extracted by the atom library is set according to the maximum frequency of 50.5Hz of the fundamental wave allowed by the power system specified by the national power quality standard; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal to be measured;
Kγ1is a normalization factor, whose value is:
Figure FDA0002821359760000012
wherein, g1Is the fundamental atom with a normalized coefficient of 1.
3. The power quality signal feature extraction method according to claim 1, characterized in that: and (3) collecting the power quality signals of the power grid in the step (2) at industrial loads, residential loads, transformer substations, photovoltaic power stations, wind power plants, railway traction stations, electric vehicle charging piles and distributed power supply grid-connected positions.
4. The power quality signal feature extraction method according to claim 1, characterized in that: the extracting of the fundamental wave signal features in the step (2) specifically comprises the following steps:
(2a) initialization setting: initial residual signal
Figure FDA0002821359760000013
Equal to the power quality signal f to be measured;
(2b) selecting the best atom g matched with the power quality signal f to be measured in a fundamental wave atom library by utilizing an LSA algorithmγ1(opt)：
Figure FDA0002821359760000021
In the formula, N represents the length of the power quality signal f to be measured;
Figure FDA0002821359760000022
and
Figure FDA0002821359760000023
i are each determined by the LSA algorithm1、j1The optimal solution of (2); t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
initial residual signal
Figure FDA0002821359760000024
The best matching atom satisfies:
Figure FDA0002821359760000025
in the formula, gγ1Representing atoms in a fundamental atom library;
(2c) updating residual signals:
Figure FDA0002821359760000026
(2d) calculating the amplitude of the fundamental wave:
Figure FDA0002821359760000027
extracted fundamental component:
Figure FDA0002821359760000028
(2e) output extracted fundamental frequency:
Figure FDA0002821359760000029
outputting the extracted fundamental wave phase:
Figure FDA00028213597600000210
and outputting a waveform diagram.
5. The power quality signal feature extraction method according to claim 1, characterized in that: the step (3) specifically comprises the following steps:
(3a) class fundamental atom library:
fundamental-like atomic gγ2The expression is as follows:
Figure FDA00028213597600000211
in the formula, u (t) is a step function, and N is the length of the power quality signal f to be measured; f. of1The fundamental frequency extracted by a fundamental atom library is utilized; j is a function of2∈[0，N]，j2The initial value of (a) is any random number from 0 to N; n iss2，ne2∈[0，N]，ns2And ne2The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ2is a normalization factor, whose value is:
Figure FDA0002821359760000031
in the formula, g2The normalized coefficient is a fundamental wave-like atom with 1;
(3b) pulsed atom library
Pulse atom gγ3The expression is as follows:
Figure FDA0002821359760000032
in the formula, u (t) is a step function, and N is the length of the power quality signal f to be measured; n iss3，ne3∈[0，N]，ns3And ne3The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ3is a normalization factor, whose value is:
Figure FDA0002821359760000033
g3is a pulse atom with a normalized coefficient of 1;
(3c) harmonic atom library
Harmonic atom gγ4The expression is as follows:
Figure FDA0002821359760000034
in the formula, N is the length of the power quality signal f to be measured; i.e. i4∈[0，N]，i4The initial value of (a) is any random number from 0 to N; j is a function of4∈[0，N]，j4The initial value of (a) is any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]，fsThe sampling frequency of the power quality signal f to be detected;
Kγ4is a normalization factor, whose value is:
Figure FDA0002821359760000035
g4harmonic atoms with a normalized coefficient of 1;
(3d) pool of mutator atoms
Flash atom gγ5The expression is as follows:
Figure FDA0002821359760000041
wherein u (t) is a step function, fsThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N; f. of1And
Figure FDA0002821359760000042
are respectivelyFrequency and phase extracted by fundamental wave atom library, fN5The value of (a) is set according to the voltage fluctuation and flicker frequency range specified by the electric power system electromagnetic phenomenon parameters and classification standards formulated by IEEE, and the reference value range fN5≥25Hz；i5∈[0，N]，i5The initial value of (a) is any random number from 0 to N; j is a function of5∈[0，N]，j5The initial value of (a) is any random number from 0 to N; n iss5，ne5∈[0，N]，ns5And ne5The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]；
Kγ5Is a normalization factor, whose value is:
Figure FDA0002821359760000043
g5is a flickering atom with a normalized coefficient of 1;
(3e) oscillating atom library
Oscillating atom gγ6The expression is as follows:
Figure FDA0002821359760000044
wherein u (t) is a step function, fsThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N; i.e. i6∈[0，N]，i6The initial value of (a) is any random number from 0 to N; j is a function of6∈[0，N]，j6The initial value of (a) is any random number from 0 to N; k is an element of [0, N ]]The initial value of k is any random number from 0 to N; n iss6，ne6∈[0，N]，ns6And ne6The initial values of all the random numbers are any random number from 0 to N; t is a time variable, t belongs to [0, N/f ]s]；
Kγ6Is a normalization factor, whose value is:
Figure FDA0002821359760000045
g6is an oscillating atom with a normalized coefficient of 1.
6. The power quality signal feature extraction method according to claim 1, characterized in that: the step (4) of extracting the electric energy quality signal features specifically comprises the following steps:
(4a) selecting the best matching atom g of class fundamental wave from class fundamental wave atom library, pulse atom library, harmonic atom library, flicker atom library and oscillation atom library in turn by using LSA algorithmγ2(opt)Best matching pulse atom gγ3(opt)The harmonic best matching atom gγ4(opt)The flash best matching atom gγ5(opt)And oscillation best matching atom gγ6(opt)；
Figure FDA0002821359760000051
Figure FDA0002821359760000052
Figure FDA0002821359760000053
Figure FDA0002821359760000054
Figure FDA0002821359760000055
In the formula (f)sThe sampling frequency of the power quality signal f to be detected is N, and the length of the power quality signal f to be detected is N;
Figure FDA0002821359760000056
and
Figure FDA0002821359760000057
j is respectively calculated by LSA algorithm2、ns2And ne2The current optimal solution of;
Figure FDA0002821359760000058
and
Figure FDA0002821359760000059
n being respectively calculated by LSA algorithms3And ne3The current optimal solution of;
Figure FDA00028213597600000510
and
Figure FDA00028213597600000511
i are each determined by the LSA algorithm4And j4The current optimal solution of;
Figure FDA00028213597600000512
Figure FDA00028213597600000513
and
Figure FDA00028213597600000514
i are each determined by the LSA algorithm5、j5、ns5And ne5The current optimal solution of;
Figure FDA00028213597600000515
k*、
Figure FDA00028213597600000516
and
Figure FDA00028213597600000517
i are each determined by the LSA algorithm6、j6、k、ns6And ne6The current optimal solution of; t is a time variable, t belongs to [0, N/f ]s]；fN5The value of (a) is set according to the voltage fluctuation and flicker frequency range specified by the electric power system electromagnetic phenomenon parameters and classification standards formulated by IEEE, and the reference value range fN5≥25Hz；；f1And
Figure FDA00028213597600000518
respectively extracting the frequency and the phase by using a fundamental wave atom library;
the best matching atom satisfies:
Figure FDA00028213597600000519
in the formula, gγzRepresenting atoms in a selected atom pool;
Figure FDA00028213597600000520
denotes a residual signal after extracting the disturbance signal, z is 2, 3, 4, 5, 6
(4b) Updating residual signals in sequence:
Figure FDA00028213597600000521
(4c) calculating the amplitude of the disturbance:
Figure FDA0002821359760000061
Figure FDA0002821359760000062
Figure FDA0002821359760000063
Figure FDA0002821359760000064
Figure FDA0002821359760000065
in the formula, A2To A6Sequentially representing a similar fundamental wave disturbance amplitude, a pulse disturbance amplitude, a harmonic disturbance amplitude, a flicker disturbance amplitude and an oscillation disturbance amplitude;
extracted disturbance component:
Figure FDA0002821359760000066
Figure FDA0002821359760000067
Figure FDA0002821359760000068
Figure FDA0002821359760000069
Figure FDA00028213597600000610
in the above formula, V2To V6Sequentially representing similar fundamental wave disturbance component and pulse disturbanceThe dynamic component, the harmonic disturbance component, the flicker disturbance component and the oscillation disturbance component;
(4d) outputting the extracted disturbance characteristic parameters
Outputting the extracted harmonic disturbance frequency f4Flicker disturbance frequency f5And an oscillation disturbance frequency f6：
Figure FDA00028213597600000611
Outputting extracted similar fundamental wave disturbance initial phase
Figure FDA00028213597600000612
Harmonic disturbance initial phase
Figure FDA00028213597600000613
Initial phase of flicker disturbance
Figure FDA00028213597600000614
And the initial phase of the oscillation disturbance
Figure FDA00028213597600000615
Figure FDA00028213597600000616
Outputting the extracted similar fundamental wave disturbance starting time ts2Start time t of pulse disturbances3Start time t of flicker disturbances5And oscillation disturbance start time ts6：
Figure FDA0002821359760000071
Outputting the extracted similar fundamental wave disturbance termination time te2Pulse disturbance termination time te3End time of flicker disturbancete5And oscillation disturbance termination time te6：
Figure FDA0002821359760000072
Outputting the extracted oscillation disturbance attenuation coefficient rho6：
Figure FDA0002821359760000073
And outputting the extracted waveform diagram of the disturbance.
CN202011418852.4A 2020-12-07 2020-12-07 Electric energy quality signal feature extraction method Active CN112597816B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011418852.4A CN112597816B (en) 2020-12-07 2020-12-07 Electric energy quality signal feature extraction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011418852.4A CN112597816B (en) 2020-12-07 2020-12-07 Electric energy quality signal feature extraction method

Publications (2)

Publication Number Publication Date
CN112597816A true CN112597816A (en) 2021-04-02
CN112597816B CN112597816B (en) 2022-09-13

Family

ID=75188650

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011418852.4A Active CN112597816B (en) 2020-12-07 2020-12-07 Electric energy quality signal feature extraction method

Country Status (1)

Country Link
CN (1) CN112597816B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070087756A1 (en) * 2005-10-04 2007-04-19 Hoffberg Steven M Multifactorial optimization system and method
CN103124180A (en) * 2013-01-14 2013-05-29 江苏大学 Data reconfiguration and decompression method of power system based on projection pursuit
US20180165550A1 (en) * 2016-12-09 2018-06-14 International Business Machines Corporation Signal classification using sparse representation
CN108287279A (en) * 2018-01-12 2018-07-17 国网冀北电力有限公司秦皇岛供电公司 Direct current energy quality disturbance detection method and device
CN110245321A (en) * 2019-05-09 2019-09-17 东北石油大学 Duration power quality disturbances method and device based on match tracing sparse decomposition
CN111368892A (en) * 2020-02-27 2020-07-03 合肥工业大学 Generalized S transformation and SVM electric energy quality disturbance efficient identification method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070087756A1 (en) * 2005-10-04 2007-04-19 Hoffberg Steven M Multifactorial optimization system and method
CN103124180A (en) * 2013-01-14 2013-05-29 江苏大学 Data reconfiguration and decompression method of power system based on projection pursuit
US20180165550A1 (en) * 2016-12-09 2018-06-14 International Business Machines Corporation Signal classification using sparse representation
CN108287279A (en) * 2018-01-12 2018-07-17 国网冀北电力有限公司秦皇岛供电公司 Direct current energy quality disturbance detection method and device
CN110245321A (en) * 2019-05-09 2019-09-17 东北石油大学 Duration power quality disturbances method and device based on match tracing sparse decomposition
CN111368892A (en) * 2020-02-27 2020-07-03 合肥工业大学 Generalized S transformation and SVM electric energy quality disturbance efficient identification method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
LING AI WONG ET.AL: "Optimal Power Quality Monitors Placement using Improved Lightning Search Algorithm", 《2018 IEEE 7TH INTERNATIONAL CONFERENCE ON POWER AND ENERGY (PECON)》 *
王云静等: "基于PSO-MP算法和RBF神经网络的电能质量扰动识别", 《电测与仪表》 *
袁莉芬: "基于小波包优选的模拟电路故障特征提取方法", 《电工技术学报》 *
龚嫄: "基于原子稀疏分解的电能质量扰动检测与压缩性能研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 *

Also Published As

Publication number Publication date
CN112597816B (en) 2022-09-13

Similar Documents

Publication Publication Date Title
CN105866571B (en) A kind of transient power quality signal analysis method based on high-frequency harmonic compensating iterative EMD
CN110705762A (en) Ubiquitous power Internet of things perception data missing repairing method based on matrix filling
CN107036709A (en) A kind of transformer station&#39;s noise matching separation method
CN111753699B (en) Method for improving digital metering accuracy of direct-current charging pile
CN110543921A (en) cable early fault identification method based on waveform learning
Vasanth et al. Fault location estimation for VSC-HVDC system using Artificial Neural Network
CN108959704B (en) Rewarding and punishing weight type simulation sequence similarity analysis method considering morphological change
CN112597816B (en) Electric energy quality signal feature extraction method
Yoneyama et al. Unified source-filter GAN with harmonic-plus-noise source excitation generation
CN105550450B (en) Electric energy quality interference source characteristic harmonic modeling method
Stanisavljević et al. A comprehensive overview of digital signal processing methods for voltage disturbance detection and analysis in modern distribution grids with distributed generation
CN108680782A (en) The voltage flicker parameter detection method decomposed based on extreme point symmetric pattern
CN108761202A (en) The harmonic detecting method that pole symmetric mode decomposition and Hilbert transform are combined
CN108334822B (en) Kalman and modified wavelet transform filtering method based on electric vehicle charging nonlinear load characteristics
CN113238110B (en) Power quality disturbance diagnosis method
Kong et al. A novel disturbance identification method based on empirical mode decomposition for HVDC transmission line protection
Kumar et al. S-Transform based detection of multiple and multistage power quality disturbances
CN113740678A (en) Noise reduction method and system suitable for corona current measurement data
CN113902136A (en) Load identification method based on electric power fingerprint features and integrated learning mechanism
CN114689991A (en) Method and system for determining fault time of high-voltage cable and related components
CN108564249B (en) Power distribution network confidence peak clipping benefit evaluation method considering distributed photovoltaic randomness
CN110221168B (en) Method for positioning leading harmonic source and tracking harmonic pollution propagation path
Yuan et al. Analysis of power grid harmonics with wavelet network
CN110108985A (en) A kind of recognition methods of the power distribution network internal overvoltage based on Atomic Decomposition algorithm
Patel et al. Wavelet and Machine learning based approach for Fault classification in AC Micro-grid system

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
GR01 Patent grant