CN108918994B - Electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting - Google Patents
Electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting Download PDFInfo
- Publication number
- CN108918994B CN108918994B CN201810597746.3A CN201810597746A CN108918994B CN 108918994 B CN108918994 B CN 108918994B CN 201810597746 A CN201810597746 A CN 201810597746A CN 108918994 B CN108918994 B CN 108918994B
- Authority
- CN
- China
- Prior art keywords
- disturbance
- wavelet packet
- power quality
- signal
- opt
- 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.)
- Expired - Fee Related
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Monitoring And Testing Of Exchanges (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a power quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting, and relates to the field of detection of power quality disturbanceFirstly, constructing four different new wavelets, and performing redundant wavelet packet decomposition on node signals to be decomposed one by one; decomposing each node signal to obtain four groups of high-frequency detail signals, and respectively normalizing the four high-frequency detail signals by lpNorm, taking the wavelet corresponding to the minimum norm value as the optimal wavelet matched with the node signal; in the same way, until all node signals of each layer are decomposed, the modulus maximum value of the topmost high-frequency detail signal obtained by decomposing each layer is averaged to obtain the final judgment value of the disturbance starting time and the disturbance ending time; and secondly, forward taking a full-period time corresponding to the power frequency from the disturbance starting time of the signal after noise reduction, taking an absolute ratio of a voltage value of the full-period time without the modulus maximum value and a voltage value of the disturbance starting time, and judging the type and the severity of the disturbance according to a ratio result.
Description
Technical Field
The invention belongs to the technical field of power quality detection, and particularly relates to a power quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting.
Background
The electric energy is used as an important energy source in the development of modern society and the national economic life, and the high-quality, reliable and stable transmission of the electric energy has important significance for guaranteeing the normal running of various industries and the daily life of people.
However, as the equipment and environment become more complex, power quality problems have become more prominent in recent years. In power systems, power quality problems mainly include both steady-state and transient states, and short-time voltage changes in transient power quality problems are becoming the most common typical problems in engineering practice. When a brief voltage change occurs, the amplitude of the voltage will abruptly change at the beginning and end of the disturbance. The wavelet analysis has excellent singularity detection capability, and can be used for positioning and detecting the disturbance starting and stopping moments by combining a modulus maximum algorithm. In the analysis process, wavelets with different characteristics are selected, and the obtained analysis results are different. If a plurality of wavelets with different characteristics are simultaneously selected to analyze the power quality monitoring signals, and then the best one is selected from a plurality of analysis results by combining a self-adaptive algorithm based on a certain criterion or criterion, the accuracy of disturbance positioning is expected to be improved to a great extent. In practice, the power quality monitoring signal often contains background noise, which causes great difficulty in classification identification and severity determination of disturbance. Therefore, an effective algorithm should be used to reduce noise and improve signal-to-noise ratio. Wavelet analysis not only has singularity detection capability, but also the threshold denoising algorithm is widely applied due to strong realizability. After noise is filtered to a certain extent by adopting a wavelet packet threshold noise reduction method, according to the characteristic that the power supply voltage is a sine wave and the definitions of various disturbance types, the whole period time of the power frequency is taken from the beginning time to the front of the disturbance, and the absolute value ratio of the corresponding voltages at the two times is calculated, so that the influence caused by the noise can be further weakened, and the disturbance type and the severity can be more accurately judged.
Disclosure of Invention
The invention aims to: by providing the electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting, the starting and stopping moments of electric energy quality disturbance containing noise are accurately positioned, the disturbance type and the severity are accurately identified, and technical support is provided for effectively solving the electric energy quality problem.
In order to realize the purpose, the invention is realized by adopting the technical scheme that:
a power quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting is used for more accurately identifying power quality transient event disturbance and comprises the following steps:
first, four different new wavelets are constructed using data fitting and lifting algorithms. The method comprises the following specific steps:
(1) determining different basis functions, sample point numbers M and the dimensionality N of the basis functions;
(2) and the length of the predictor is equal to that of the updating operator. And calculating to obtain a prediction operator and an update operator coefficient, and constructing four new wavelets by combining the relationship between the prediction operator and the update operator coefficient and the filter bank coefficient.
Secondly, performing two-layer redundant lifting wavelet packet decomposition on the power quality disturbance signal one by applying four new wavelets. Each node signal x to be decomposedi,j(i is 1,2 is the current decomposition layer number; j is 1, K,2i-1，xi,jJ node signal for i layer) to obtain four groups of high-frequency detail signals di,j,1、di,j,2、di,j,3、di,j,4Respectively normalizing the four samples by lpNorm, taking the high-frequency detail signal with the minimum norm value as xi,jOptimal decomposition result of
di,j,optTo obtain di,j,optCorresponding wavelet is matching xi,jThe other three groups of high-frequency detail signals are abandoned at the same time; after the decomposition of the two-layer lifting wavelet packet is completed by the self-adaptive algorithm, d can be obtained1,1,opt、d2,1,optAnd d2,2,opt。
Thirdly, the two layers of the topmost high-frequency approximation signals d are respectively subjected to1,1,optAnd d2,2,optThe modulus maximum values are averaged to obtain the final decision value t of the two groups of disturbance starting time and ending timestartAnd tendPositioning disturbance is realized;
then, for the disturbance duration tend-tstartAnalyzing, if the transient event can be judged to be the power quality transient event in the standard range, applying a wavelet threshold denoising algorithm to denoise the disturbance signal, and then denoising the denoised signal from tstartForward according to the AC power frequency t of ChinaperiodWhen 1/50 is 0.02(s), i.e. t is tstart-k·tperiod(k∈Z+)。
Finally, judging whether a modulus maximum value appears at the moment t; taking the forward whole period time t without modulus maximumnormalAnd tstartTo obtain an absolute ratio of the voltage values of
And judging the type and the severity of the disturbance according to the ratio.
The four basis functions are:
(1)φ1(x)＝x1.5·k；
(2)φ2(x)＝x0.2·k；
(3)φ3(x)＝xk·e0.25·x；
(4)φ1(x)＝xk·cos(0.1·k)。
where k is 0,1,2, Λ N, N is the dimensionality of the basis function.
The values of the number of sample points M and the number of dimensions N of the basis function are 6 and 5, respectively.
Above lPIn the norm, p takes the value of 0.1.
In the wavelet threshold denoising, a compromise threshold function is selected as
The threshold is generated by the Heursure rule.
The power quality disturbance detection method for the adaptive fitting and wavelet packet noise reduction analysis is used for accurately identifying the disturbance of the power quality transient event. Firstly, four different new wavelets are constructed by applying a data fitting and lifting algorithm, and redundant wavelet packet decomposition is carried out on node signals to be decomposed one by one; decomposing each node signal to obtain four groups of high-frequency detail signals, and respectively normalizing the four high-frequency detail signals by lpNorm, which takes the wavelet corresponding to the minimum norm value as the optimal wavelet matched with the node signal, and retains the group of high-frequency detail signals and discards other three groups of high-frequency detail signals; in the same way, the self-adaptive algorithm is realized until all node signals of each layer are decomposed; averaging the modulus maximum values of the topmost high-frequency detail signals obtained by decomposing each layer to obtain final judgment values of the disturbance starting time and the disturbance ending time, and realizing the positioning of the disturbance; secondly, filtering the noise of the power quality disturbance signal by using a wavelet threshold denoising algorithm, taking a full-period time corresponding to the power frequency from the disturbance starting time forward for the denoised signal, taking an absolute ratio of a voltage value of the full-period time without a modulus maximum value and a voltage value of the disturbance starting time, and judging the type and the severity of the disturbance according to the ratio result. The invention provides a power quality disturbance detection method based on adaptive fitting and wavelet packet noise reduction analysis improvement, which can improve the precision of disturbance positioning and realize judgment on disturbance type and severity.
Drawings
FIG. 1 is a general flow diagram of the present invention;
figure 2 is four different wavelets constructed based on the least squares and lifting algorithm of the data fit.
Detailed Description
The following detailed description of specific embodiments of the invention refers to the accompanying drawings.
As shown in fig. 1, the process of detecting the power quality disturbance signal is mainly divided into the following two steps:
the method comprises the following steps of firstly, acquiring a power supply voltage signal with the sample length of L from a monitoring end, and performing two-layer redundant lifting wavelet packet decomposition by selecting four new wavelets constructed based on data fitting and lifting algorithms. Each node signal x to be decomposedi,j(i is 1,2 is the current decomposition layer number; j is 1, K,2i-1，xi,jIs the jth node signal of the ith layer), four groups of corresponding high-frequency detail signals d can be obtainedi,j,k(k ═ 1,2,3, 4); the normalization of each of the four is determined by the following formulaPNorm Pi,j,k||pComprises the following steps:
take P | |i,j,k||pThe high frequency detail signal with the smallest value is xi,jOptimum decomposition result d ofi,j,optTo obtain di,j,optCorresponding wavelet is matching xi,jThe other three groups of high-frequency detail signals are abandoned at the same time; d can be obtained after the decomposition of the two-layer lifting wavelet packet is completed by the self-adaptive algorithm1,1,opt、d2,1,optAnd d2,2,opt. For two layers of respective topmost high-frequency detail signals d1,1,optAnd d2,2,optTaking a modulus maximum value to obtain two groups of judging values of disturbance starting time and disturbance ending time, namely: from d1,1,optDetermined tend1And tend1From d2,2,optDetermined tstart2And tend2(ii) a Taking the mean value (t)start1+tstart2) [ 2 ] as the disturbance start time tstartMean value (t)end1+tend2) [ 2 ] as disturbance end time tendTo thereby achieve localization of the disturbance。
Four different wavelets are used for carrying out redundancy lifting wavelet packet decomposition on the electric energy quality disturbance signal based on the self-adaptive algorithm, so that the disturbance starting time and the disturbance ending time can be detected more accurately, and the precision of disturbance positioning is improved.
Second, first, the duration t of the disturbance is determinedduration＝tend-tstart. According to the definition of the transient event of the power quality, if t is not more than 0.01s and less than or equal to tdurationIf the time is less than or equal to 60s, judging that no power quality transient event occurs, continuously acquiring power quality monitoring signals, and analyzing again according to the first step; if t is 0.01 s. ltoreq.tdurationLess than or equal to 60s, judging that the power quality transient event occurs, and further applying a compromise threshold function to the transient event
And carrying out wavelet packet threshold denoising on the threshold generated by the Heursure rule. Secondly, combining the characteristic that the ideal power supply voltage is a sine wave, and carrying out noise reduction on the signal from tstartForward power frequency, tperiodThe whole period time of 0.02(s) ═ 1/50(Hz), i.e. t ═ tstart-k·tperiod(k∈Z+) Judging whether the modulus maximum value appears at the acquired moment; for disturbance starting time tstartAnd a forward whole-cycle time t at which no modulo maximum occursnormalThe corresponding voltage values are:
computing
Is the absolute value ratio of
And judging according to the R value:
(1) if R is more than or equal to 1.1 and less than or equal to 1.8, the voltage transient rise event is judged, and the larger R is, the more serious disturbance is;
(2) if R is more than or equal to 0.1 and less than or equal to 0.9, determining a voltage sag event; and the smaller R, the more severe the disturbance;
(3) if R is less than 0.1, determining a short-time voltage interruption event; and the smaller R, the more severe the disturbance.
The wavelet packet threshold denoising processing is carried out on the disturbance signal of the power quality transient event, so that background noise can be effectively filtered, and the interference of the noise on subsequent analysis is reduced; by applying the absolute value ratio of the voltage at the starting moment of disturbance and the forward whole period moment without the occurrence of the modulus maximum, possible misjudgment caused by noise can be further weakened, and the accuracy of disturbance type and severity identification is improved.
As shown in fig. 2, four different new wavelets are constructed for the least squares and lifting algorithm based data fitting: in the figure, the numerical index above the wavelet waveform diagram indicates that the wavelet is obtained by constructing the basis functions with the same index.
Claims (6)
1. A power quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting is characterized by comprising the following steps:
firstly, four different new wavelets are constructed by applying a data fitting and lifting algorithm, and the specific steps are as follows:
(1) determining different basis functions, sample point numbers M and the dimensionality N of the basis functions;
(2) the length of the predictor is equal to that of the updating operator, the coefficients of the predictor and the updating operator are obtained through calculation, and four new wavelets are constructed by combining the relationship between the coefficients of the predictor and the updating operator and the coefficients of the filter bank;
secondly, performing two-layer redundant lifting wavelet packet decomposition on the power quality disturbance signal one by applying four new wavelets, wherein each node signal x to be decomposedi,jWherein, i is 1, and 2 is the current decomposition layer number; j 1.. 2i-1，xi,jFor the jth node signal of the ith layer, decomposing to obtain fourGroup high frequency detail signal di,j,1、di,j,2、di,j,3、di,j,4Respectively normalizing the four samples by lpNorm, taking the high-frequency detail signal with the minimum norm value as xi,jOptimum decomposition result d ofi,j,optTo obtain di,j,optCorresponding wavelet is matching xi,jThe other three groups of high-frequency detail signals are abandoned at the same time; after the decomposition of the two-layer lifting wavelet packet is completed by the self-adaptive algorithm, d can be obtained1,1,opt、d2,1,optAnd d2,2,opt；
Thirdly, the two layers of the topmost high-frequency approximation signals d are respectively subjected to1,1,optAnd d2,2,optThe modulus maximum values are averaged to obtain the final decision value t of the two groups of disturbance starting time and ending timestartAnd tendPositioning disturbance is realized;
then, for the disturbance duration tend-tstartAnalyzing, if the transient event can be judged to be the power quality transient event in the standard range, applying a wavelet threshold denoising algorithm to denoise the disturbance signal, and then denoising the denoised signal from tstartForward according to the AC power frequency t of ChinaperiodWhen 1/50 is 0.02(s), i.e. t is tstart-k·tperiod(k∈Z+)；
Finally, judging whether a modulus maximum value appears at the moment t; taking the forward whole period time t without modulus maximumnormalAnd tstartTo obtain an absolute ratio of the voltage values of
And judging the type and the severity of the disturbance according to the ratio.
2. The method for detecting the power quality disturbance of the adaptive fitting wavelet packet denoising analysis according to claim 1, wherein the method comprises the following steps: the basis functions are respectively as follows:
(1)φ1(x)＝x1.5·k；
(2)φ2(x)＝x0.2·k；
(3)φ3(x)＝xk·e0.25·x；
(4)φ1(x)＝xk·cos(0.1·k)；
where k is 0,1,2, … N, N being the dimensionality of the basis function.
3. The method for detecting the power quality disturbance of the adaptive fitting wavelet packet denoising analysis according to claim 1, wherein the method comprises the following steps: the sample point number M and the dimension N of the basis function are respectively 6 and 5.
4. The method for detecting the power quality disturbance of the adaptive fitting wavelet packet denoising analysis according to claim 1, wherein the method comprises the following steps: the above-mentionedPIn the norm, p takes the value of 0.1.
5. The method for detecting the power quality disturbance of the adaptive fitting wavelet packet denoising analysis according to claim 1, wherein the method comprises the following steps: in the wavelet threshold denoising, a compromise threshold function is selected as
The threshold is generated by the Heursure rule.
6. The method for detecting the power quality disturbance of the adaptive fitting wavelet packet denoising analysis according to claim 1, wherein the method comprises the following steps: for signal after noise reduction from tstartForward power frequency, tperiodThe whole period time of 0.02(s) ═ 1/50(Hz), i.e. t ═ tstart-k·tperiod(k∈Z+) Judging whether the modulus maximum value appears at the acquired moment; for disturbance starting time tstartAnd a forward whole-cycle time t at which no modulo maximum occursnormalThe corresponding voltage values are:
computing
Is the absolute value ratio of
And judging according to the R value:
(1) if R is more than or equal to 1.1 and less than or equal to 1.8, determining that the voltage temporarily rises, wherein the larger R is, the more serious the disturbance is;
(2) if R is more than or equal to 0.1 and less than or equal to 0.9, determining a voltage sag event; and the smaller R, the more severe the disturbance;
(3) if R <0.1, determining a short-time voltage interruption event; and the smaller R, the more severe the disturbance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810597746.3A CN108918994B (en) | 2018-06-11 | 2018-06-11 | Electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810597746.3A CN108918994B (en) | 2018-06-11 | 2018-06-11 | Electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108918994A CN108918994A (en) | 2018-11-30 |
| CN108918994B true CN108918994B (en) | 2021-12-10 |
Family
ID=64418792
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810597746.3A Expired - Fee Related CN108918994B (en) | 2018-06-11 | 2018-06-11 | Electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108918994B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112132021B (en) * | 2020-09-22 | 2023-10-31 | 三峡大学 | Transient power quality disturbance detection method based on WMM and HATF |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101655520A (en) * | 2009-09-11 | 2010-02-24 | 哈尔滨工业大学 | Method for extracting lightning strike signals and transient harmonic signals in power system |
| CN102901855A (en) * | 2012-08-16 | 2013-01-30 | 中国电力科学研究院 | De-noising method for ultra-high-voltage direct-current corona current signal |
| CN106970278A (en) * | 2017-03-28 | 2017-07-21 | 北京印刷学院 | Electrical energy power quality disturbance recognition methods based on fitting Lifting Wavelet and mean analysis |
| CN107356843A (en) * | 2017-04-17 | 2017-11-17 | 武汉科技大学 | The partial discharge of transformer method for diagnosing faults of small echo is synchronously extruded based on gradient threshold |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160124031A1 (en) * | 2014-11-04 | 2016-05-05 | Walid G. Morsi Ibrahim | Smart multi-purpose monitoring system using wavelet design and machine learning for smart grid applications |
-
2018
- 2018-06-11 CN CN201810597746.3A patent/CN108918994B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101655520A (en) * | 2009-09-11 | 2010-02-24 | 哈尔滨工业大学 | Method for extracting lightning strike signals and transient harmonic signals in power system |
| CN102901855A (en) * | 2012-08-16 | 2013-01-30 | 中国电力科学研究院 | De-noising method for ultra-high-voltage direct-current corona current signal |
| CN106970278A (en) * | 2017-03-28 | 2017-07-21 | 北京印刷学院 | Electrical energy power quality disturbance recognition methods based on fitting Lifting Wavelet and mean analysis |
| CN107356843A (en) * | 2017-04-17 | 2017-11-17 | 武汉科技大学 | The partial discharge of transformer method for diagnosing faults of small echo is synchronously extruded based on gradient threshold |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108918994A (en) | 2018-11-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109855874B (en) | Random resonance filter for enhancing detection of weak signals in vibration assisted by sound | |
| CN113325277A (en) | Partial discharge processing method | |
| CN115424635B (en) | Cement plant equipment fault diagnosis method based on sound characteristics | |
| CN112362987A (en) | Lightning arrester fault diagnosis method based on robust estimation | |
| CN108918994B (en) | Electric energy quality disturbance detection method for improving wavelet packet noise reduction analysis through self-adaptive fitting | |
| CN106533361B (en) | A kind of photovoltaic module method for diagnosing faults based on wavelet analysis | |
| CN115840120A (en) | High-voltage cable partial discharge abnormity monitoring and early warning method | |
| CN113551927A (en) | Mechanical equipment fault early warning method and system based on vibration signals | |
| CN109708877B (en) | Mechanical fault analysis method based on wavelet fuzzy recognition and image analysis theory | |
| CN114023353A (en) | Transformer fault classification method and system based on cluster analysis and similarity calculation | |
| Feng et al. | A blind source separation method using denoising strategy based on ICEEMDAN and improved wavelet threshold | |
| Hua et al. | The methodology of modified frequency band envelope kurtosis for bearing fault diagnosis | |
| CN108594156B (en) | Improved current transformer saturation characteristic identification method | |
| Kumar et al. | Classification of voiced and non-voiced speech signals using empirical wavelet transform and multi-level local patterns | |
| CN107505337B (en) | Microwave reflection signal characteristic database for composite insulator defect detection | |
| Lv et al. | A new feature extraction technique for early degeneration detection of rolling bearings | |
| CN112033656A (en) | Mechanical system fault detection method based on broadband spectrum processing | |
| CN110287853B (en) | Transient signal denoising method based on wavelet decomposition | |
| CN117289087A (en) | Series fault arc detection method based on CZT conversion | |
| CN116559643A (en) | GIS partial discharge fault discrimination method based on feature similarity | |
| CN115293219B (en) | Pulse signal denoising method integrating wavelet and kurtosis | |
| CN112104340B (en) | HMM model and Kalman filtering technology-based switching value input module BIT false alarm reduction method | |
| CN112800041A (en) | Neighborhood query-based mechanical monitoring label data quality guarantee method | |
| CN109506936A (en) | Bearing fault degree recognition methods based on flow graph and non-naive Bayesian reasoning | |
| Sun et al. | Research on fault diagnosis method of bearing based on parameter optimization VMD and improved DBN |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20211210 |