CN112132021B - Transient power quality disturbance detection method based on WMM and HATF - Google Patents

Transient power quality disturbance detection method based on WMM and HATF Download PDF

Info

Publication number
CN112132021B
CN112132021B CN202011004374.2A CN202011004374A CN112132021B CN 112132021 B CN112132021 B CN 112132021B CN 202011004374 A CN202011004374 A CN 202011004374A CN 112132021 B CN112132021 B CN 112132021B
Authority
CN
China
Prior art keywords
wavelet
signal
decomposition
disturbance
wmm
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.)
Active
Application number
CN202011004374.2A
Other languages
Chinese (zh)
Other versions
CN112132021A (en
Inventor
程杉
严潇
徐敬伟
王睿
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.)
China Three Gorges University CTGU
Original Assignee
China Three Gorges University CTGU
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 China Three Gorges University CTGU filed Critical China Three Gorges University CTGU
Priority to CN202011004374.2A priority Critical patent/CN112132021B/en
Publication of CN112132021A publication Critical patent/CN112132021A/en
Application granted granted Critical
Publication of CN112132021B publication Critical patent/CN112132021B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06F2218/06—Denoising by applying a scale-space analysis, e.g. using wavelet analysis
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Signal Processing (AREA)
  • General Engineering & Computer Science (AREA)
  • Indication And Recording Devices For Special Purposes And Tariff Metering Devices (AREA)

Abstract

Sampling disturbance signals to obtain a sampling sequence based on a transient power quality disturbance detection method of WMM and HATF; selecting a proper wavelet base; determining the maximum number of decomposition layers by obtaining the minimum frequency of the signalNThe method comprises the steps of carrying out a first treatment on the surface of the According to the energy distribution characteristics of each decomposition layer of wavelet transformation, an adjustable factor is establishedμAnd calculating the adjustment factors of each decomposition level according to the model. Calculating threshold values of decomposition layersλ. And obtaining the self-adaptive threshold function of each decomposition layer, and processing the detail coefficient after wavelet decomposition. And carrying out wavelet reconstruction on the approximation coefficients and the detail coefficients of each decomposition layer. And (3) starting electric energy quality disturbance detection, positioning mutation points, and solving a mode maximum value and the position of the mode maximum value, wherein the mode maximum value point is the mutation point of a signal. Recording the moment and time interval corresponding to the maximum value point of the mode, namely the starting and stopping moment and duration of the disturbance. The method of the invention reduces noise of the disturbance signal, better keeps the information of the mutation position, and improves the accuracy of transient disturbance detection under noise interference.

Description

Transient power quality disturbance detection method based on WMM and HATF
Technical Field
The invention belongs to the field of transient power quality, and particularly relates to a transient power quality disturbance detection method based on WMM and HATF.
Background
In the prior art, renewable energy sources, electric automobiles and the like are networked through nonlinear power electronic devices, so that the electric energy quality of a power system is seriously influenced, and the transient electric energy quality problem is particularly serious. The premise of treating and improving the transient power quality is to effectively detect the transient disturbance signal, but the transient disturbance signal has randomness and uncertainty, and the acquired signal to be detected is often polluted by noise due to factors such as sampling errors, external interference in the channel transmission process and the like.
The sensitivity of the traditional transient power quality disturbance detection method is greatly affected in a strong noise environment, and how to effectively reduce noise is a problem which needs to be solved when disturbance signals are detected. The earliest proposal of Donoho et al was to use wavelet hard and soft threshold functions for noise reduction, but due to the great limitations in practical application, many studies have focused on improvements in the threshold functions and their application. In order to reduce noise of disturbance signals and better keep information of mutation positions, accuracy of transient disturbance detection under noise interference is improved.
Disclosure of Invention
The invention provides a transient power quality disturbance detection method based on WMM and HATF, which can reduce noise of disturbance signals and better retain information of mutation positions, thereby effectively improving the accuracy of transient disturbance detection under noise interference.
The technical scheme adopted by the invention is as follows:
the transient power quality disturbance detection method based on WMM and HATF comprises the following steps:
step one: sampling 100 points every cycle of the disturbance signal to obtain a sampling sequence s (t);
step two: denoising the sampling sequence s (t) and selecting a proper wavelet basis;
step three: by obtaining the minimum frequency f of the sampling sequence s (t) min Determining the maximum decomposition layer number N, namely:
wherein deltat is the sampling period; f (f) 0 The center frequency of the wavelet basis function;
step four: establishing a mathematical model of an adjustable factor mu according to the energy distribution characteristics of each decomposition layer of wavelet transformation, and calculating the adjustable factor mu of each decomposition layer according to the model;
step five: calculating an optimal threshold lambda of each decomposition layer;
step six: constructing HATF, bringing the adjustment factor mu and the optimal threshold lambda to obtain self-adaptive threshold functions of all decomposition layers, and decomposing the detail coefficient d after wavelet n Processing;
step seven: approximation coefficient and detail coefficient d of each decomposition layer n Performing wavelet reconstruction to obtain a reconstructed signal, and completing denoising;
step eight: starting electric energy quality disturbance detection, determining wavelet base and decomposition scale, and selecting db4 wavelet to perform 5-layer decomposition on the sampling sequence due to suitability of the db4 wavelet to obtaind 1 ；
Step nine: locating the mutation points: find d 1 The maximum value of the model and the position thereof are shown as abrupt points of the signal;
step ten: recording the moment and time interval corresponding to the maximum value point of the mode, namely the starting and stopping moment and duration of the disturbance.
The invention discloses a transient power quality disturbance detection method based on WMM and HATF, which has the following technical effects:
1) The invention relates to a novel transient power quality disturbance detection method based on the combination of a wavelet mode maximum value (Wavelet Modulus Maximum, WMM) and a layered self-adaptive threshold function (Hierarchical Adaptive Threshold Function, HATF). According to different energy proportion in each decomposition layer of the wavelet, adjustable parameters are introduced to adaptively obtain the HATF of each layer, and the noise reduction purpose is achieved by processing the signal to be detected through the HATF, so that the influence of noise interference on the WMM detection method is reduced.
2) The WMM detection algorithm is simple, the real-time performance is good, and the detection effect on noise-free single disturbance and composite disturbance signals is good by determining the position positioning disturbance of the detail coefficient mode maximum value point.
3) HTAF has a significant noise reduction effect compared to hard, soft, and modified threshold functions.
4) The detection method based on WMM and HTAF has wide application range, has good detection effect on single disturbance and composite disturbance, and can keep higher positioning accuracy in 15dB noise environment.
Drawings
Fig. 1 is a HATF-based noise reduction flow diagram.
FIG. 2 (1) is a plot of the original signal of a voltage sag disturbance signal;
FIG. 2 (2) is a graph of a noisy signal for a voltage sag disturbance signal;
fig. 2 (3) is a graph of the noise reduced signal of the voltage sag disturbance signal.
FIG. 3 (1) is a plot of the original signal of the voltage sag disturbance signal;
FIG. 3 (2) is a graph of a noisy signal for a voltage sag disturbance signal;
fig. 3 (3) is a graph of the noise reduced signal of the voltage sag disturbance signal.
FIG. 4 (1) is a plot of the original signal of a voltage interrupt disturbance signal;
FIG. 4 (2) is a graph of a noisy signal for a voltage-interrupted disturbance signal;
fig. 4 (3) is a graph of the noise reduced signal of the voltage interruption disturbance signal.
FIG. 5 (1) is a plot of the original signal of a transient pulse disturbance signal;
FIG. 5 (2) is a plot of the noise-containing signal of the transient pulse disturbance signal;
fig. 5 (3) is a graph of the noise reduced signal of the transient pulse disturbance signal.
FIG. 6 (1) is a plot of the original signal of a transient oscillation disturbance signal;
FIG. 6 (2) is a graph of the noise-containing signal of a transient oscillating disturbance signal;
fig. 6 (3) is a graph of the noise reduced signal of the transient oscillation disturbance signal.
FIG. 7 (1) is a plot of the original signal of a voltage dip versus transient oscillation composite disturbance signal;
FIG. 7 (2) is a graph of the noise-containing signal of the voltage dip and transient oscillation composite disturbance signal;
fig. 7 (3) is a graph of the noise reduced signal of the voltage dip and transient oscillation composite disturbance signal.
FIG. 8 (1) is a graph of the original signal localization result (based on hard threshold function and WMM) of the voltage sag disturbance signal;
FIG. 8 (2) is a graph of the noisy signal localization result (based on soft threshold function and WMM) of the voltage sag disturbance signal;
FIG. 8 (3) is a graph of the post-noise reduction signal localization result of a voltage sag disturbance signal (based on the modified threshold function and WMM);
fig. 8 (4) is a diagram of the post-noise reduction signal localization result of the voltage sag disturbance signal (based on HATF and WMM).
FIG. 9 (1) is a graph of the original signal localization result (based on hard threshold function and WMM) of the voltage sag disturbance signal;
FIG. 9 (2) is a graph of the noisy signal localization result (based on soft threshold function and WMM) of the voltage sag disturbance signal;
FIG. 9 (3) is a graph of the post-noise reduction signal localization result of the voltage sag disturbance signal (based on the modified threshold function and WMM);
fig. 9 (4) is a diagram of the post-noise reduction signal localization result of the voltage sag disturbance signal (based on HATF and WMM).
FIG. 10 (1) is a graph of the original signal localization result (based on hard threshold function and WMM) of the voltage interrupt disturbance signal;
FIG. 10 (2) is a graph of the noisy signal localization result (based on soft threshold function and WMM) of the voltage-interrupted disturbance signal;
FIG. 10 (3) is a graph of the post-noise reduction signal localization result of the voltage interrupt disturbance signal (based on the modified threshold function and WMM);
fig. 10 (4) is a diagram of the post-noise reduction signal localization result of the electric voltage interruption disturbance signal (based on HATF and WMM).
FIG. 11 (1) is a graph of the original signal localization results (based on hard threshold function and WMM) of the transient pulse disturbance signal;
FIG. 11 (2) is a graph of the noisy signal localization result (based on soft threshold function and WMM) of the transient pulse disturbance signal;
FIG. 11 (3) is a graph of the post-noise reduction signal localization result of the transient pulse disturbance signal (based on the modified threshold function and WMM);
fig. 11 (4) is a diagram of the post-noise reduction signal localization result of the transient pulse disturbance signal (based on HATF and WMM).
FIG. 12 (1) is a graph of the original signal localization results (based on hard threshold function and WMM) of transient oscillation disturbance signals;
FIG. 12 (2) is a graph of the noisy signal localization result (based on soft threshold function and WMM) of a transient oscillation disturbance signal;
FIG. 12 (3) is a graph of the post-noise reduction signal localization results of transient oscillation disturbance signals (based on the modified threshold function and WMM);
fig. 12 (4) is a diagram of the post-noise reduction signal localization result of the transient oscillation disturbance signal (based on HATF and WMM).
FIG. 13 (1) is a graph of the original signal localization result (based on hard threshold function and WMM) of the voltage sag and transient pulse composite disturbance signal;
FIG. 13 (2) is a graph of the noise-containing signal localization result (based on soft threshold function and WMM) of the voltage sag and transient pulse composite disturbance signal;
FIG. 13 (3) is a graph of the post-noise reduction signal localization result of the voltage sag and transient pulse composite disturbance signal (based on the modified threshold function and WMM);
fig. 13 (4) is a diagram of the post-noise reduction signal localization result of the voltage sag and transient pulse composite disturbance signal (based on HATF and WMM).
Detailed Description
The transient power quality disturbance detection method based on WMM and HATF comprises the following steps:
step one: sampling 100 points per cycle of the disturbance signal, namely setting the sampling frequency to be 5kHz, and obtaining a sampling sequence s (t);
step two: denoising the sampling sequence s (t) is started, and an appropriate wavelet basis is selected. Due to good similarity and symmetry, sym4 wavelets were chosen as wavelet basis functions.
Sym4 may act as a basic wavelet function. The wavelet sequence can be obtained by stretching and translating the wavelet sequence:
wherein Sym (t) is a basic wavelet function, sym α,β And (t) is a wavelet function after expansion and translation, and alpha and beta are scale factors and expansion factors respectively. For any signal Sym (t) ε L in the time domain 2 (R) the wavelet transform may be expressed as:
step three: by obtaining the minimum frequency f of the sampling sequence s (t) min Determining the maximum decomposition layer number N, namely:
wherein deltat is the sampling period; f (f) 0 Is the center frequency of the wavelet basis function.
Step four: according to the energy distribution characteristics of each decomposition layer of wavelet transformation, establishing a mathematical model of an adjustable factor mu, and calculating the adjustable factor mu of each decomposition layer according to the model:
the noise-containing signal s (t) comprises a clean signal x (t), a noise signal n (t), i.e.:
s(t)＝x(t)+n(t) (4)；
according to the energy distribution characteristics of each decomposition layer of n (t) and s (t) wavelet transformation, establishing a mathematical model of an adjustable factor mu:
wherein E is nj And E is sj The energy of n (t) and s (t) in the j-th layer decomposition are respectively, and the noise-containing signal s (t) and the noise signal n (t) in the first layer decomposition are basically consistent in waveform, so E n1 ≈E s1 The rest are E sj ＝2 j E nj The value range of mu on each decomposition layer is obtained by mathematical calculation: 1 to 11.
Step five: calculating an optimal threshold lambda of each decomposition layer:
according to the unbiased likelihood estimation principle, an optimal threshold lambda is obtained:
each element in the signal s (t) is subjected to absolute value sequencing from small to large, and each element is subjected to squaring, so that a new signal sequence y (k) is obtained:
y(k)＝[sort(|s|)] 2 ,k＝0,1,...,N-1 (6)；
wherein: the sort is the order command in MATLAB. S represents the absolute value of each element in the signal s (t), y (k) represents the new signal sequence, and k represents the sequence number.
Taking the square root of the kth element with a threshold of y (k), namely:
wherein: MAD is the absolute mean of the wavelet coefficient estimates on the best scale, σ represents the threshold coefficient, and a factor of 0.6755 is the gaussian correction choice.
The risk Rish (k) that this value generates is:
according to the obtained risk curve Rish (k), the minimum risk point is calculated as r min The definition of the optimal threshold λ is:
step six: constructing HATF, bringing the adjustment factor mu and the optimal threshold lambda to obtain self-adaptive threshold functions of all decomposition layers, and decomposing the detail coefficient d after wavelet n And (3) performing treatment:
in order to keep more detail information after the noise of the signal is reduced, constructing the HATF as shown in the formula (8), bringing the adjustment factor mu and the optimal threshold lambda, and obtaining the self-adaptive threshold function of each decomposition layer.
Wherein: t represents an adaptive function of each decomposition level, and x represents a sampled signal value.
Step seven: approximation coefficient and detail coefficient d of each decomposition layer n And carrying out wavelet reconstruction to obtain a reconstructed signal, and finishing denoising. Wavelet reconstruction process:
let t be L 2 (R) Fourier transform of the complex is psi (omega), if the condition is satisfied:
then ψ (t) may be a basic wavelet function. The wavelet sequence can be obtained by stretching and translating the wavelet sequence:
wherein, alpha and beta are scale factors and expansion factors respectively. For any signal f (t) e L in the time domain 2 (R) the wavelet transform may be expressed as:
the reconstruction formula is as follows:
step eight: and starting power quality disturbance detection, determining a wavelet base and a decomposition scale, and selecting the wavelet base and the decomposition scale to carry out 5-layer decomposition on a sampling sequence due to the suitability of db4 wavelets. The information of the local abrupt change point of the signal is mainly based on the detail coefficient d of the first layer of wavelet decomposition 1 So the first layer detail coefficient d of the decomposition is selected 1 . And (4) selecting db4 wavelets by using an MATLAB wavelet toolbox, so as to decompose.
Step nine: locating the mutation points: find d 1 The modulus maximum point is the abrupt point of the signal:
the mode maximum value point of the disturbance signal after wavelet transformation can reflect the abrupt change information of the signal, so that the starting and ending moment of the disturbance can be positioned by determining the position of the mode maximum value point, and the deduction is as follows:
given a smooth low-pass function θ (x), let its first derivative be:
thenThe band-pass function can be used as a basic wavelet, and then the wavelet function under the scale factor alpha is as follows:
in the method, in the process of the invention,represents the expansion and contraction of θ (x) at the scale factor α.
The wavelet transform of the signal f (x) on the scale α can be expressed as:
where W (α, x) represents the first derivative of the signal f (x) after wavelet transformation at the scale α, and its mode maxima correspond to local discontinuities of the signal.
Step ten: determining the starting and stopping moments of disturbance: recording the moment and time interval corresponding to the maximum value point of the mode, namely the starting and stopping moment and duration of the disturbance.
FIG. 2 (1) is a plot of the original signal of a voltage sag disturbance signal;
FIG. 2 (2) is a graph of a noisy signal for a voltage sag disturbance signal;
fig. 2 (3) is a graph of the noise reduced signal of the voltage sag disturbance signal.
As shown in fig. 2 (1) to fig. 2 (3), the HATF noise reduction algorithm can effectively denoise the voltage transient rise disturbance signal.
FIG. 3 (1) is a plot of the original signal of the voltage sag disturbance signal;
FIG. 3 (2) is a graph of a noisy signal for a voltage sag disturbance signal;
fig. 3 (3) is a graph of the noise reduced signal of the voltage sag disturbance signal.
As shown in fig. 3 (1) to 3 (3), the HATF noise reduction algorithm can effectively denoise the voltage sag disturbance signal.
FIG. 4 (1) is a plot of the original signal of a voltage interrupt disturbance signal;
FIG. 4 (2) is a graph of a noisy signal for a voltage-interrupted disturbance signal;
fig. 4 (3) is a graph of the noise reduced signal of the voltage interruption disturbance signal.
As shown in fig. 4 (1) to fig. 4 (3), the HATF noise reduction algorithm can effectively denoise the voltage interruption disturbance signal.
FIG. 5 (1) is a plot of the original signal of a transient pulse disturbance signal;
FIG. 5 (2) is a plot of the noise-containing signal of the transient pulse disturbance signal;
fig. 5 (3) is a graph of the noise reduced signal of the transient pulse disturbance signal.
As shown in fig. 5 (1) to 5 (3), the HATF noise reduction algorithm can effectively denoise the transient pulse disturbance signal.
FIG. 6 (1) is a plot of the original signal of a transient oscillation disturbance signal;
FIG. 6 (2) is a graph of the noise-containing signal of a transient oscillating disturbance signal;
fig. 6 (3) is a graph of the noise reduced signal of the transient oscillation disturbance signal.
The contrast of fig. 6 (1) to fig. 6 (3) shows that the HATF noise reduction algorithm can effectively denoise the transient oscillation disturbance signal.
FIG. 7 (1) is a plot of the original signal of a voltage dip versus transient oscillation composite disturbance signal;
FIG. 7 (2) is a graph of the noise-containing signal of the voltage dip and transient oscillation composite disturbance signal;
fig. 7 (3) is a graph of the noise reduced signal of the voltage dip and transient oscillation composite disturbance signal.
Compared with the graphs from the graph (1) to the graph (3), the HATF noise reduction algorithm can effectively remove noise from the voltage sag and transient oscillation composite disturbance signal.
FIG. 8 (1) is a graph of the localization result of the voltage sag disturbance signal based on the hard threshold function and WMM;
FIG. 8 (2) is a graph of the positioning result of the voltage sag disturbance signal based on the soft threshold function and WMM;
FIG. 8 (3) is a graph of the localization result of the voltage sag disturbance signal based on the improvement threshold function and WMM;
fig. 8 (4) is a diagram of the localization result of the voltage sag disturbance signal based on HATF and WMM.
As shown in fig. 8 (1) to fig. 8 (4), the HATF and WMM methods have more accurate disturbance signal recognition capability and positioning capability than the other three methods when positioning the voltage transient increase disturbance signal.
FIG. 9 (1) is a graph of the localization result of the voltage sag disturbance signal based on the hard threshold function and WMM;
FIG. 9 (2) is a graph of the positioning result of the voltage sag disturbance signal based on the soft threshold function and WMM;
FIG. 9 (3) is a graph of the localization result of the voltage sag disturbance signal based on the improvement threshold function and WMM;
fig. 9 (4) is a diagram of the positioning result of the voltage sag disturbance signal based on HATF and WMM.
As shown in fig. 9 (1) to fig. 9 (4), the HATF and WMM methods have more accurate disturbance signal recognition capability and positioning capability than the other three methods when positioning the voltage sag disturbance signal.
FIG. 10 (1) is a graph of the voltage interrupt disturbance signal localization result based on a hard threshold function and WMM;
FIG. 10 (2) is a graph of the voltage interrupt disturbance signal localization result based on the soft threshold function and WMM;
FIG. 10 (3) is a graph of the voltage interrupt disturbance signal localization result based on the modified threshold function and WMM;
fig. 10 (4) is a diagram of the voltage interruption disturbance signal localization result based on HATF and WMM.
As shown in fig. 10 (1) to fig. 10 (4), the HATF and WMM methods have more accurate disturbance signal recognition capability and positioning capability than the other three methods when positioning the voltage interruption disturbance signal.
FIG. 11 (1) is a graph of the transient pulse disturbance signal localization result based on a hard threshold function and WMM;
FIG. 11 (2) is a graph of the transient pulse disturbance signal localization result based on the soft threshold function and WMM;
FIG. 11 (3) is a graph of the transient pulse disturbance signal localization result based on the modified threshold function and WMM;
fig. 11 (4) is a diagram of the transient pulse disturbance signal localization result based on HATF and WMM.
As shown in fig. 11 (1) to 11 (4), the HATF and WMM methods have more accurate disturbance signal recognition capability and positioning capability than the other three methods when positioning the transient pulse disturbance signal.
FIG. 12 (1) is a graph of transient oscillation disturbance signal localization results based on a hard threshold function and WMM;
FIG. 12 (2) is a graph of transient oscillation disturbance signal localization results based on soft threshold function and WMM;
FIG. 12 (3) is a graph of transient oscillation disturbance signal localization results based on the modified threshold function and WMM;
fig. 12 (4) is a diagram of the transient oscillation disturbance signal localization result based on HATF and WMM.
As shown in fig. 12 (1) to fig. 12 (4), the HATF and WMM methods have more accurate disturbance signal recognition capability and positioning capability than the other three methods when positioning the transient oscillation disturbance signal.
FIG. 13 (1) is a graph of disturbance signal localization results based on a hard threshold function and a voltage sag and transient pulse of WMM;
FIG. 13 (2) is a graph of disturbance signal localization results based on soft threshold function and WMM voltage sag and transient pulse;
FIG. 13 (3) is a graph of disturbance signal localization results based on a modified threshold function and a voltage sag and transient pulse of WMM;
fig. 13 (4) is a diagram of the disturbance signal localization result based on the voltage sag and transient pulse of HATF and WMM.
As shown in fig. 13 (1) to fig. 13 (4), the HATF and WMM methods have more accurate disturbance signal recognition capability and positioning capability than the other three methods when positioning the disturbance signals of the voltage sag and the transient pulse.
Table 1 six transient disturbance signal positioning accuracy rates
Table 1 shows six transient disturbance signal positioning accuracy comparison tables, and can be obtained through Table 1, the detection accuracy of the four detection methods is not very different under the 30dB weak noise environment, but the positioning accuracy of the other three methods is obviously reduced along with the increase of the noise intensity, and HATF+WMM only slightly slips down. Under the 15dB strong noise interference condition, the positioning accuracy of the HATF+WMM on the voltage sag, the voltage dip, the voltage interruption, the transient pulse, the transient oscillation and the voltage dip+transient pulse composite disturbance signal can be maintained at 95.9%, 96.1%, 95.9%, 94.4%, 94.5% and 93.5% respectively.

Claims (6)

1. The transient power quality disturbance detection method based on WMM and HATF is characterized by comprising the following steps:
step one: sampling the disturbance signal every cycle to obtain a sampling sequence s (t);
step two: denoising the sampling sequence s (t), and selecting a proper wavelet basis;
step three: by obtaining the minimum frequency f of the sampling sequence s (t) min Determining the maximum decomposition layer number N, namely:
wherein deltat is the sampling period; f (f) 0 The center frequency of the wavelet basis function;
step four: establishing a mathematical model of an adjustable factor mu according to the energy distribution characteristics of each decomposition layer of wavelet transformation, and calculating the adjustable factor mu of each decomposition layer according to the model;
step five: calculating an optimal threshold lambda of each decomposition layer;
step six: constructing HATF, bringing in the regulating factor mu and the optimal threshold lambda to obtain eachAdaptive thresholding of decomposition level and decomposing the wavelet decomposed detail coefficients d n Processing;
step six, constructing the HATF as shown in the formula (8), and bringing the adjustment factor mu and the optimal threshold lambda to obtain the self-adaptive threshold function of each decomposition layer;
wherein: t represents the self-adaptive function of each decomposition layer, and x represents the sampled signal value;
step seven: approximation coefficient and detail coefficient d of each decomposition layer n Performing wavelet reconstruction to obtain a reconstructed signal, and completing denoising;
step eight: starting electric energy quality disturbance detection, determining wavelet base and decomposition scale, and selecting db4 wavelet to carry out 5-layer decomposition on the sampling sequence due to suitability of the db4 wavelet to obtain d 1 ；
Step nine: locating the mutation points: find d 1 The maximum value of the model and the position thereof are shown as abrupt points of the signal;
in the step nine, the mode maximum value point of the disturbance signal after wavelet transformation can reflect the abrupt change information of the signal, so that the start and stop moment of disturbance can be positioned by determining the position of the mode maximum value point, and the deduction is as follows:
given a smooth low-pass function θ (x), let its first derivative be:
thenThe band-pass function can be used as a basic wavelet, and then the wavelet function under the scale factor alpha is as follows:
in the method, in the process of the invention,represents the expansion and contraction of θ (x) under the scale factor α;
the wavelet transform of the signal f (x) on the scale α can be expressed as:
wherein W (alpha, x) represents the first derivative of the signal f (x) after wavelet transformation under the scale alpha, and the mode maximum point corresponds to the local mutation point of the signal;
step ten: determining the starting and stopping moments of disturbance: recording the moment and time interval corresponding to the maximum value point of the mode, namely the starting and stopping moment and duration of the disturbance.
2. The method for detecting transient power quality disturbance based on WMM and HATF according to claim 1, wherein: selecting Sym4 wavelet as a wavelet basis function;
sym4 may act as a basic wavelet function; the wavelet sequence can be obtained by stretching and translating the wavelet sequence:
wherein Sym (t) is a basic wavelet function, sym α,β (t) is a wavelet function after expansion and translation, and alpha and beta are scale factors and expansion factors respectively; for any signal Sym (t) ε L in the time domain 2 (R) the wavelet transform may be expressed as:
3. the method for detecting transient power quality disturbance based on WMM and HATF according to claim 1, wherein: in the fourth step, the noise-containing signal s (t) includes the clean signal x (t) and the noise signal n (t), namely:
s(t)＝x(t)+n(t)(4)；
according to the energy distribution characteristics of each decomposition layer of n (t) and s (t) wavelet transformation, establishing a mathematical model of an adjustable factor mu:
wherein E is nj And E is sj The energy of n (t) and s (t) in the j-th layer decomposition are respectively, and the noise-containing signal s (t) and the noise signal n (t) in the first layer decomposition are basically consistent in waveform, so E n1 ≈E s1 The rest are E sj ＝2 j E nj The value range of mu on each decomposition layer is obtained by mathematical calculation: 1 to 11.
4. The method for detecting transient power quality disturbance based on WMM and HATF according to claim 1, wherein: in the fifth step, according to the unbiased likelihood estimation principle, an optimal threshold lambda is obtained:
each element in the signal s (t) is subjected to absolute value sequencing from small to large, and each element is subjected to squaring, so that a new signal sequence y (k) is obtained:
y(k)＝[sort(|s|)] 2 ,k＝0,1,...,N-1 (6)；
wherein: the sort is an order command in MATLAB; s represents the absolute value of each element in the signal s (t), y (k) represents the new signal sequence, k represents the sequence number;
taking the square root of the kth element with a threshold of y (k), namely:
wherein: MAD is the absolute mean of the wavelet coefficient estimates on the optimal scale, σ represents the threshold coefficient, and the factor 0.6755 is the gaussian distribution correction choice;
the risk Rish (k) that this value generates is:
according to the obtained risk curve Rish (k), the minimum risk point is calculated as r min The definition of the optimal threshold λ is:
5. the method for detecting transient power quality disturbance based on WMM and HATF according to claim 1, wherein: in the seventh step, the wavelet reconstruction process:
let t be L 2 (R) Fourier transform of the complex is psi (omega), if the condition is satisfied:
then ψ (t) may be a basic wavelet function; the wavelet sequence can be obtained by stretching and translating the wavelet sequence:
wherein, alpha and beta are scale factors and expansion factors respectively; for any signal f (t) e L in the time domain 2 (R) the wavelet transform may be expressed as:
the reconstruction formula is as follows:
6. the method for detecting transient power quality disturbance based on WMM and HATF according to claim 1, wherein:
in the eighth step, power quality disturbance detection is started, a wavelet base and a decomposition scale are determined, and the db4 wavelet is selected to carry out 5-layer decomposition on a sampling sequence due to the suitability of the db4 wavelet; the information of the local abrupt change point of the signal is mainly based on the detail coefficient d of the first layer of wavelet decomposition 1 So the first layer detail coefficient d of the decomposition is selected 1 The method comprises the steps of carrying out a first treatment on the surface of the And (4) selecting db4 wavelets by using an MATLAB wavelet toolbox, so as to decompose.
CN202011004374.2A 2020-09-22 2020-09-22 Transient power quality disturbance detection method based on WMM and HATF Active CN112132021B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011004374.2A CN112132021B (en) 2020-09-22 2020-09-22 Transient power quality disturbance detection method based on WMM and HATF

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011004374.2A CN112132021B (en) 2020-09-22 2020-09-22 Transient power quality disturbance detection method based on WMM and HATF

Publications (2)

Publication Number Publication Date
CN112132021A CN112132021A (en) 2020-12-25
CN112132021B true CN112132021B (en) 2023-10-31

Family

ID=73842486

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011004374.2A Active CN112132021B (en) 2020-09-22 2020-09-22 Transient power quality disturbance detection method based on WMM and HATF

Country Status (1)

Country Link
CN (1) CN112132021B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113176431A (en) * 2021-03-26 2021-07-27 东南大学溧阳研究院 Voltage sag initial time positioning analysis method and device
CN117562561B (en) * 2024-01-17 2024-04-26 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) Surface electromyographic signal denoising method, system, equipment and computer storage medium

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101551853A (en) * 2008-11-14 2009-10-07 重庆大学 Human ear detection method under complex static color background
CN101661616A (en) * 2009-09-29 2010-03-03 北京科技大学 Method for enhancing images based on multi-scale edge detection in wavelet reconstruction
CN102014296A (en) * 2010-12-10 2011-04-13 北京中科大洋科技发展股份有限公司 Video consistency monitoring technology based on self-adaptive edge matching and local stream processing algorithm
CN108732440A (en) * 2018-04-17 2018-11-02 北京理工大学 A kind of transient power quality detection method and system
CN108918994A (en) * 2018-06-11 2018-11-30 北京印刷学院 A kind of duration power quality disturbances method of auto-adapted fitting Lifting Wavelet packet Reduction Analysis
CN108985179A (en) * 2018-06-22 2018-12-11 福建和盛高科技产业有限公司 A kind of electric energy quality signal denoising method based on improvement wavelet threshold function
CN110967599A (en) * 2019-12-11 2020-04-07 贵州电网有限责任公司 Electric energy quality disturbance detection and positioning algorithm

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101551853A (en) * 2008-11-14 2009-10-07 重庆大学 Human ear detection method under complex static color background
CN101661616A (en) * 2009-09-29 2010-03-03 北京科技大学 Method for enhancing images based on multi-scale edge detection in wavelet reconstruction
CN102014296A (en) * 2010-12-10 2011-04-13 北京中科大洋科技发展股份有限公司 Video consistency monitoring technology based on self-adaptive edge matching and local stream processing algorithm
CN108732440A (en) * 2018-04-17 2018-11-02 北京理工大学 A kind of transient power quality detection method and system
CN108918994A (en) * 2018-06-11 2018-11-30 北京印刷学院 A kind of duration power quality disturbances method of auto-adapted fitting Lifting Wavelet packet Reduction Analysis
CN108985179A (en) * 2018-06-22 2018-12-11 福建和盛高科技产业有限公司 A kind of electric energy quality signal denoising method based on improvement wavelet threshold function
CN110967599A (en) * 2019-12-11 2020-04-07 贵州电网有限责任公司 Electric energy quality disturbance detection and positioning algorithm

Also Published As

Publication number Publication date
CN112132021A (en) 2020-12-25

Similar Documents

Publication Publication Date Title
CN112395992B (en) Power harmonic signal denoising method based on improved wavelet threshold
CN113378661B (en) Direct-current electric energy signal denoising method based on improved wavelet threshold and related detection
CN108985179B (en) Electric energy quality signal denoising method based on improved wavelet threshold function
CN112132021B (en) Transient power quality disturbance detection method based on WMM and HATF
CN110865357B (en) Laser radar echo signal noise reduction method based on parameter optimization VMD
Sun et al. Cross-correlation analysis and time delay estimation of a homologous micro-seismic signal based on the Hilbert–Huang transform
CN109557429A (en) Based on the GIS partial discharge fault detection method for improving wavelet threshold denoising
CN104182946A (en) Spectral denoising method based on index soft and hard averaging threshold function
CN113642417A (en) Improved wavelet algorithm-based denoising method for partial discharge signals of insulated overhead conductor
CN112307959A (en) Wavelet denoising method for electrocardiosignal analysis
CN109409281A (en) A kind of noise-reduction method based on improved wavelet threshold function
CN110032968B (en) Denoising method based on dual-tree complex wavelet and self-adaptive semi-soft threshold method
CN115061018A (en) High-voltage switch cabinet partial discharge monitoring method
CN110703089B (en) Wavelet threshold denoising method for low-frequency oscillation Prony analysis
Sun et al. Wavelet denoising method based on improved threshold function
CN110287853B (en) Transient signal denoising method based on wavelet decomposition
CN107015124A (en) A kind of Partial discharge signal disturbance restraining method decomposed based on framing adaptive sparse
CN108020761B (en) A kind of Denoising of Partial Discharge
CN115310478A (en) Method for extracting effective Raman spectrum of natural gas based on wavelet optimal decomposition layer number
CN111723677B (en) Wavelet denoising method based on self-adaptive threshold
CN114814897A (en) Combined filtering method for weakening GNSS multipath error
CN113702821A (en) Method and system for extracting GIS partial discharge signal
Feng et al. Voice activity detection based on ensemble empirical mode decomposition and teager kurtosis
CN108764092B (en) Micro-fluidic chip signal denoising method based on energy element dual-threshold
CN103376108B (en) Star sensor noise reduction method

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