CN113238110B - Power quality disturbance diagnosis method - Google Patents
Power quality disturbance diagnosis method Download PDFInfo
- Publication number
- CN113238110B CN113238110B CN202110507182.1A CN202110507182A CN113238110B CN 113238110 B CN113238110 B CN 113238110B CN 202110507182 A CN202110507182 A CN 202110507182A CN 113238110 B CN113238110 B CN 113238110B
- Authority
- CN
- China
- Prior art keywords
- frequency
- power quality
- classifier
- diagnosis
- quality disturbance
- 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
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
Abstract
The invention relates to a power quality disturbance diagnosis method, which comprises the following steps: building a power quality disturbance diagnosis system platform; collecting three-phase voltage and current signals of monitoring points A, B and C in real time through a system platform, and transmitting sampling data to an upper computer; carrying out improved Kaiser window rapid S conversion on the sampled data, setting a window width adjusting factor to obtain a mode time-frequency matrix, and extracting a fundamental frequency amplitude curve and a frequency amplitude envelope curve; extracting characteristic parameters to construct a characteristic vector; determining a classifier hyper-parameter, training a LightGBM classifier, and storing a classifier model; and sending the feature vectors into a trained LightGBM classifier to obtain a diagnosis result. The invention also discloses a power quality disturbance diagnosis system. The method has higher identification precision when power quality disturbance diagnosis is carried out, and the result of building a power quality disturbance diagnosis system platform shows that the average accuracy of the method can reach 99.75 percent, and the method meets the accurate diagnosis requirement of power quality disturbance signals in a field environment.
Description
Technical Field
The invention relates to the technical field of smart power grids, in particular to a power quality disturbance diagnosis method.
Background
With the change of the power grid, power supply and load of the new generation of power system, the quality of electric energy meets new problems and challenges. The ferromagnetic nonlinear device, the power electronic equipment containing the semiconductor switch and the PWM modulation bring harmonic waves or ultra-high harmonic waves to the power generation and transmission link of the power grid; the voltage disturbance caused by the line short-circuit fault and the impulse load start-stop within 10ms is also a new problem of concern. The power quality disturbance not only reduces the utilization and transmission efficiency of the power, but also seriously influences the service life of the electrical equipment, and more power accidents are caused by unqualified power quality in the future, so the correct diagnosis of the power quality disturbance becomes the key for treating and improving the power quality.
The power quality disturbance diagnosis can be divided into two links of disturbance feature extraction and disturbance mode identification, wherein the disturbance feature extraction mainly comprises the following steps: fast Fourier transform, short-time Fourier transform, wavelet transform, hilbert-yellow transform, S transform, generalized S transform, multi-resolution generalized S transform, dual-resolution S transform, fast S transform and the like, wherein the S transform and the variants thereof are most widely applied to the extraction of the power quality disturbance characteristics. Although S transformation has good noise immunity, the operation amount is too large, and the real-time detection requirement under the field environment cannot be met. The generalized S transformation improves the time-frequency resolution to a certain extent by setting a window width adjusting factor, but the electric energy quality disturbance signals are various, the various disturbance differences are large, the requirement of the time-frequency resolution of each frequency band still cannot be met, and the selection of the window width adjusting factor lacks a theoretical basis. The multi-resolution generalized S transform obtains better time-frequency resolution by setting window width adjusting factors in a segmented manner, but the phenomenon of unsmooth transition exists among different frequency segments, and certain interference is brought to later-stage feature extraction. The double-resolution S transformation leads the frequency resolution to be superior to the multi-resolution generalized S transformation by transforming the window function, but the window width of the double-resolution S transformation is still approximately in inverse proportion to the frequency, so that the frequency resolution is still lower under the condition of rich high-frequency components, and even the aliasing phenomenon of the high-frequency components can occur. The fast S transformation is realized, the characteristic frequency points are selected through the FFT frequency spectrum for transformation, unnecessary calculation cost is reduced, the operation speed is improved, but when the disturbance containing transient oscillation signals is analyzed, the characteristic frequency points are inaccurately positioned, and the accurate extraction of disturbance characteristics is influenced.
The disturbance pattern recognition mainly comprises a support vector machine, a neural network, a decision tree, ensemble learning and the like. The integrated learning obtains a better classification effect than a single learner by integrating a plurality of weak learners, wherein the integrated learning algorithm LightGBM using the decision tree as the base classifier obtains a better classification effect in mass electric energy quality disturbance signal recognition due to high training speed, high recognition precision, strong generalization capability and strong robustness.
Disclosure of Invention
The invention aims to provide a power quality disturbance diagnosis method with high diagnosis speed and high identification precision.
In order to achieve the purpose, the invention adopts the following technical scheme: a method for diagnosing disturbances in the quality of electrical energy, the method comprising the sequential steps of:
(1) Building a power quality disturbance diagnosis system platform;
(2) Collecting three-phase voltage and current signals of monitoring points A, B and C in real time through the power quality disturbance diagnosis system platform, and transmitting the sampled data to an upper computer of the power quality disturbance diagnosis system platform;
(3) Carrying out improved Kaiser window rapid S transformation on sampling data received by an upper computer, setting a window width adjusting factor to obtain a module time-frequency matrix, and extracting a fundamental frequency amplitude curve and a frequency amplitude envelope curve from the module time-frequency matrix;
(4) Extracting characteristic parameters from the fundamental frequency amplitude curve and the frequency amplitude envelope curve to construct a characteristic vector;
(5) Determining a classifier hyper-parameter, training a LightGBM classifier, and storing a classifier model;
(6) And sending the feature vectors into a trained LightGBM classifier to obtain a diagnosis result.
The improved Kaiser window rapid S conversion in the step (3) comprises the following specific steps:
(3a) Calculating a discrete fourier transform sequence of the signal x (nT):
in the formula: k =0,1, \8230, N-1; t is a sampling time interval; n is the total sampling point number;
(3b) Performing iterative loop filtering processing on the discrete Fourier transform sequence, and obtaining a processed sequence X i+1 (k/NT) is:
in the formula: i =0,1,2, \8230; k =2,3, \ 8230;, N-3, and the interval center frequency point k is determined according to the formula (2) i Satisfies the following conditions:
taking i as 4 under the condition of comprehensively considering the minimum amplitude of harmonic waves and transient oscillation; xi takes 0.02, in frequency point k i Determining a disturbance frequency interval for extending a plurality of frequency points on the left and right sides of a reference;
(3c) According to the frequency point k of the disturbance interval i Determining window width adjusting factors m and beta in the frequency band;
(3d) Mixing X 0 (k i /NT) translation to X 0 ((k i +l)/NT)；
(3e) Calculating a time domain expression of the modified Kaiser window function discrete Fourier spectrum:
in the formula: t is time; m and beta are window width regulating factors; λ controls the total length of the window function; i is 0 Modifying a zero order Bessel function for the first class; the frequency domain expression of the discrete Fourier spectrum of the improved Kaiser window function calculated from the formula (5) is W K (n,m,β)，n＝0,1,…,N-1；
(3f) Computing improved Kaiser window fast S transform mode time frequency matrix FMKST
In the formula: n =0,1, \ 8230;, N-1; l is a translation factor for controlling the translation of the window function, l =0,1, \8230, N-1.
Setting the window width adjustment factor in the step (3) refers to setting the window width adjustment factors m and beta, wherein m =427 and beta =9.8 are taken from the low frequency range of 0-100 Hz and the high frequency range of above 700 Hz; the middle frequency range is 100-700 Hz, and m =0 is selected; β =12.
The step (4) specifically comprises the following steps: extracting the maximum value A of the fundamental frequency amplitude curve max Minimum value A min Mean value A mean And variance A var And four maximum peaks P of the envelope curve of frequency amplitude 1 、P 2 、P 3 、P 4 And its corresponding frequency value f 1 、f 2 、f 3 、f 4 And constructing a feature vector.
The step (5) specifically comprises the following steps:
(5a) Under a Spyder platform, building a LightGBM classifier model by using an open source learning library Scikit-learn;
(5b) Generating 8 analog signals using a disturbing signal source includes: normal C1, transient C2, transient C3, interruption C4, flicker C5, harmonic C6, transient C7 and transient C8, wherein all disturbance signal parameters are randomly set, the fundamental frequency is 50Hz, and 200 disturbance signals are randomly generated for each disturbance, and the total number is 1600;
(5c) Collecting 1600 analog signals, transmitting the sampled data to an upper computer, and extracting feature vectors as a data set for determining the hyperparameter and model training of the classifier according to the steps (3) and (4);
(5d) Determining the classifier hyperparameters by utilizing a five-fold cross validation method and a grid search method;
(5e) And training the LightGBM classifier by using the data set, and storing the classifier model for later diagnosis.
According to the technical scheme, the invention has the beneficial effects that: firstly, the method has higher identification precision when power quality disturbance diagnosis is carried out, and the result of building a power quality disturbance diagnosis system platform shows that the average accuracy of the method can reach 99.75 percent, so that the method meets the accurate diagnosis requirement of power quality disturbance signals in a field environment; secondly, the method has higher operation speed when power quality disturbance diagnosis is carried out, the single diagnosis time is less than 30ms, and the requirement of on-line diagnosis of mass power quality disturbance signals is met.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a flow chart of an improved Kaiser window fast S-transform algorithm;
FIG. 3 is a hardware structure diagram of a platform of the power quality disturbance diagnosis system;
fig. 4 is a power quality disturbance diagnostic system platform user interface.
Detailed Description
As shown in fig. 1, a power quality disturbance diagnosis method includes the following sequential steps:
(1) Building a power quality disturbance diagnosis system platform;
(2) Collecting three-phase voltage and current signals of monitoring points A, B and C in real time through the power quality disturbance diagnosis system platform, and transmitting sampling data to an upper computer of the power quality disturbance diagnosis system platform;
(3) Carrying out improved Kaiser window rapid S conversion on sampling data received by an upper computer, setting a window width adjusting factor to obtain a mode time-frequency matrix, and extracting a fundamental frequency amplitude curve (a row vector at 50Hz in the mode time-frequency matrix) and a frequency amplitude envelope curve (a vector formed by the maximum value of each row in the mode time-frequency matrix) from the mode time-frequency matrix;
(4) Extracting characteristic parameters from the fundamental frequency amplitude curve and the frequency amplitude envelope curve to construct a characteristic vector;
(5) Determining a classifier hyperparameter, training a LightGBM classifier, and storing a classifier model;
(6) And sending the feature vector into a trained LightGBM classifier to obtain a diagnosis result.
As shown in fig. 2, the step (3) of improving the fast S transformation of the Kaiser window specifically includes:
(3a) Calculating a discrete fourier transform sequence of the signal x (nT):
in the formula: k =0,1, \8230, N-1; t is a sampling time interval; n is the total number of sampling points;
(3b) Performing iterative loop filtering processing on the discrete Fourier transform sequence, and obtaining a processed sequence X i+1 (k/NT) is:
in the formula: i =0,1,2, \ 8230; k =2,3, \ 8230;, N-3, and the interval center frequency point k is determined according to the formula (2) i Satisfies the following conditions:
taking i as 4 under the condition of comprehensively considering the minimum amplitude of harmonic waves and transient oscillation; xi takes 0.02 as frequency point k i Determining a disturbance frequency interval for a plurality of frequency points extended on the left and right of the reference;
(3c) According to the frequency point k of the disturbance interval i Determining window width adjusting factors m and beta in the frequency band; when k is i When the frequency band is above 0-100 Hz or 700Hz, taking m =427 and beta =9.8; when k is i When the frequency band is between 100 and 700Hz, taking m =0 and beta =12;
(3d) Mixing X 0 (k i /NT) translation to X 0 ((k i +l)/NT)；
(3e) Calculating a time domain expression of the modified Kaiser window function discrete Fourier spectrum:
in the formula: t is time; m and beta are window width regulating factors;lambda controls the total length of the window function; i is 0 Modifying a zero order Bessel function for the first class; the frequency domain expression of the discrete Fourier spectrum of the improved Kaiser window function calculated from equation (5) is W K (n,m,β)，n＝0,1,…,N-1；
(3f) Computing improved Kaiser window fast S transform mode time frequency matrix FMKST
In the formula: n =0,1, \8230, N-1; l is a translation factor for controlling the translation of the window function, l =0,1, \8230;, N-1.
Setting the window width adjusting factor in the step (3) refers to setting the window width adjusting factor m and beta, wherein m =427 and beta =9.8 are taken from the low frequency range of 0-100 Hz and the high frequency range of 700 Hz; the middle frequency range is 100-700 Hz, and m =0 is selected; β =12.
The step (4) specifically comprises the following steps: extracting the maximum value A of the fundamental frequency amplitude curve max Minimum value A min Mean value A mean And variance A var And four maximum peaks P of the envelope curve of frequency amplitude 1 、P 2 、P 3 、P 4 And its corresponding frequency value f 1 、f 2 、f 3 、f 4 And constructing a feature vector.
The step (5) specifically comprises the following steps:
(5a) Under a Spyder platform, building a LightGBM classifier model by using an open source learning library Scikit-learn; the Spyder platform is a simple integrated development environment of Python, and compared with other Python development environments, the Spyder platform has the greatest advantage of simulating the 'working space' function of MATLAB and can conveniently observe and modify the value of an array;
(5b) The generation of 8 analog signals by using a disturbance signal source comprises: normal C1, temporary rising C2, temporary falling C3, interruption C4, flicker C5, harmonic C6, temporary rising + harmonic C7 and temporary falling + harmonic C8, wherein all disturbance signal parameters are randomly set, the fundamental frequency is 50Hz, 200 disturbances are randomly generated for each kind, and the total number is 1600; the model of the disturbance signal source is: fluke6105A, which is a standard source of electric energy power, can reproduce some common electric power waveforms, distortion events, and some common electric energy quality disturbance waveforms such as temporary rise, temporary fall, flicker, interruption, harmonic waves and the like;
(5c) Collecting 1600 analog signals by using a power quality disturbance diagnosis system, transmitting the sampled data to an upper computer, and extracting a feature vector as a data set for determining the hyperparameter and model training of the classifier according to the steps (3) and (4);
(5d) Determining the classifier hyperparameters by utilizing a five-fold cross validation method and a grid search method, wherein the parameters are shown in a table 1;
TABLE 1 LightGBM classifier hyper-parameter settings
(5e) The LightGBM classifier is trained using the dataset, and the classifier model is saved for later diagnosis.
To verify the effectiveness of the present invention, the generation of 8 analog signals using a perturbing signal source (Fluke 6105A) includes: the method comprises the following steps of (1) normally (C1), temporarily increasing (C2), temporarily decreasing (C3), interrupting (C4), flickering (C5), harmonic (C6), temporarily increasing + harmonic (C7) and temporarily decreasing + harmonic (C8), wherein parameters of all disturbance signals are randomly set, the fundamental frequency is 50Hz, and 200 disturbance signals are randomly generated for each disturbance signal, and the total number is 1600; 1600 analog signals are collected by the data collecting device, sampled data are transmitted to an upper computer, feature extraction is carried out on the sampled data by different feature extraction methods, the sampled data are sent into a trained LightGBM classifier to be classified, and classification accuracy rates under different feature extraction methods are shown in table 2. As can be seen from the table 2, the average accuracy of the method is 99.75%, which is superior to other two methods, and the characteristic vector extracted by the method has higher distinguishability, and meets the accurate diagnosis requirement of the power quality disturbance signal in the field environment.
TABLE 2 accuracy under different feature extraction methods
To further verify the real-time performance of the present invention, the present invention was compared with other methods at run-time for processing 1600 perturbation signals, as shown in table 3. As can be seen from Table 3, the method is superior to other methods in the aspects of feature extraction time, model training time and classification time, and the advantage is more obvious along with the increase of the number of disturbance signals, which shows that the method meets the requirement of rapidly detecting mass power quality disturbance signals.
TABLE 3 comparison of the run times of the different methods
As shown in fig. 3, the present system includes:
the voltage transformer is used for collecting voltage signals of three phases A, B and C of the monitoring points;
the current transformer is used for collecting current signals of three phases A, B and C of the monitoring points;
the conditioning circuit is used for eliminating jitter, filtering, attenuating and isolating the acquired voltage and current signals;
an A/D converter converting the analog signal into a digital signal;
the DSP processor performs down-sampling on the digital signal and transmits sampled data to the upper computer;
the upper computer is used for extracting the characteristics of the sampled data by improving the Kaiser window rapid S transformation, and sending the characteristic vectors into a trained LightGBM classifier to obtain a diagnosis result;
and the display is used for synchronously displaying the waveform of the detection signal, giving a diagnosis result and early warning through the indicator lamp.
The voltage transformer and the current transformer collect voltage and current signals of three phases A, B and C of monitoring points; the conditioning circuit processes the signals collected by the mutual inductor to make the signals matched with the input signals of an A/D converter (ADS 8556); the A/D converter converts the analog signal into a digital signal; the DSP processor (TSM 320VC 6748) down-samples the digital signal to reduce the size of the data volume; the upper computer is used for determining the hyperparameter of the classifier, training a LightGBM classifier model, storing the model, and simultaneously extracting the characteristics of the sampling data transmitted by the DSP processor to complete disturbance diagnosis; the display synchronously displays the waveform of the detection signal, gives out a diagnosis result and carries out early warning through the indicator light.
As shown in fig. 4, by self-designing a power quality disturbance diagnosis system platform user interface by LabVIEW, it is possible to implement: the method comprises the steps of waveform display of detection signals, display of a fundamental frequency amplitude curve and characteristic parameters thereof, display of a frequency amplitude envelope curve and characteristic parameters thereof, diagnosis results and early warning of an indicator light.
In conclusion, the method has higher identification precision when the power quality disturbance diagnosis is carried out, and the result of building the power quality disturbance diagnosis system platform shows that the average accuracy of the method can reach 99.75 percent, so that the method meets the accurate diagnosis requirement of the power quality disturbance signal in the field environment; the method has higher operation speed when the power quality disturbance diagnosis is carried out, the single diagnosis time is less than 30ms, and the online diagnosis requirement of mass power quality disturbance signals is met.
Claims (1)
1. A power quality disturbance diagnosis method is characterized in that: the method comprises the following steps in sequence:
(1) Building a power quality disturbance diagnosis system platform;
(2) Collecting three-phase voltage and current signals of monitoring points A, B and C in real time through the power quality disturbance diagnosis system platform, and transmitting the sampled data to an upper computer of the power quality disturbance diagnosis system platform;
(3) Carrying out improved Kaiser window rapid S transformation on sampling data received by an upper computer, setting a window width adjusting factor to obtain a module time-frequency matrix, and extracting a fundamental frequency amplitude curve and a frequency amplitude envelope curve from the module time-frequency matrix;
(4) Extracting characteristic parameters from the fundamental frequency amplitude curve and the frequency amplitude envelope curve to construct a characteristic vector;
(5) Determining a classifier hyper-parameter, training a LightGBM classifier, and storing a classifier model;
(6) Sending the feature vectors into a trained LightGBM classifier to obtain a diagnosis result;
the specific steps of improving the Kaiser window fast S transformation in the step (3) are as follows:
(3a) Calculating a discrete fourier transform sequence of the signal x (nT):
in the formula: k =0,1, \ 8230;, N-1; t is a sampling time interval; n is the total sampling point number;
(3b) Performing iterative loop filtering processing on the discrete Fourier transform sequence, and obtaining a processed sequence X i+1 (k/NT) is:
in the formula: i =0,1,2, \ 8230; k =2,3, \ 8230;, N-3, and the interval center frequency point k is determined according to the formula (2) i Satisfies the following conditions:
i is taken as 4; xi takes 0.02 as frequency point k i Determining a disturbance frequency interval for extending a plurality of frequency points on the left and right sides of a reference;
(3c) According to the frequency point k of the disturbance interval i Determining window width adjusting factors m and beta in the frequency band;
(3d) X is to be 0 (k i /NT) translation to X 0 ((k i +l)/NT)；
(3e) Calculating a time domain expression of the modified Kaiser window function discrete Fourier spectrum:
in the formula: t is time; m and beta are window width regulating factors; lambda controls the total length of the window function; i is 0 Modifying a zero order Bessel function for the first class; the frequency domain expression of the discrete Fourier spectrum of the improved Kaiser window function calculated from equation (5) is W K (n,m,β)，n＝0,1,…,N-1；
(3f) Calculating an improved Kaiser window fast S transform mode time frequency matrix FMKST:
in the formula: n =0,1, \ 8230;, N-1; l is a translation factor for controlling the translation of the window function, l =0,1, \8230;, N-1;
setting the window width adjusting factor in the step (3) refers to setting the window width adjusting factor m and beta, wherein m =427 and beta =9.8 are taken from the low frequency range of 0-100 Hz and the high frequency range of 700 Hz; the middle frequency range is 100-700 Hz, and m =0 is selected; β =12;
the step (4) specifically comprises the following steps: extracting the maximum value A of the fundamental frequency amplitude curve max Minimum value A min Mean value A mean Sum variance A var And four peak values P with the largest frequency amplitude envelope curve 1 、P 2 、P 3 、P 4 And its corresponding frequency value f 1 、f 2 、f 3 、f 4 Constructing a feature vector;
the step (5) specifically comprises the following steps:
(5a) Under a Spyder platform, using an open source learning library Scikit-learn to build a LightGBM classifier model;
(5b) The generation of 8 analog signals by using a disturbance signal source comprises: normal C1, temporary rising C2, temporary falling C3, interruption C4, flicker C5, harmonic C6, temporary rising + harmonic C7 and temporary falling + harmonic C8, wherein all disturbance signal parameters are randomly set, the fundamental frequency is 50Hz, 200 disturbances are randomly generated for each kind, and the total number is 1600;
(5c) Collecting 1600 analog signals, transmitting the sampled data to an upper computer, and extracting feature vectors as a data set for determining the hyperparameter and model training of the classifier according to the steps (3) and (4);
(5d) Determining a classifier hyper-parameter by utilizing a five-fold cross validation and a grid search method;
(5e) And training the LightGBM classifier by using the data set, and storing the classifier model for later diagnosis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110507182.1A CN113238110B (en) | 2021-05-10 | 2021-05-10 | Power quality disturbance diagnosis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110507182.1A CN113238110B (en) | 2021-05-10 | 2021-05-10 | Power quality disturbance diagnosis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113238110A CN113238110A (en) | 2021-08-10 |
| CN113238110B true CN113238110B (en) | 2022-10-14 |
Family
ID=77133020
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110507182.1A Active CN113238110B (en) | 2021-05-10 | 2021-05-10 | Power quality disturbance diagnosis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113238110B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114487894A (en) * | 2021-12-24 | 2022-05-13 | 中铁二院工程集团有限责任公司 | System for carrying out real-time quality monitoring on vehicle-mounted power supply equipment |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8539012B2 (en) * | 2011-01-13 | 2013-09-17 | Audyssey Laboratories | Multi-rate implementation without high-pass filter |
| CN103308804B (en) * | 2013-06-17 | 2016-09-14 | 湖南大学 | Based on quick K-S converting electric power quality disturbance signal time and frequency parameter extracting method |
| US10345358B2 (en) * | 2016-04-25 | 2019-07-09 | Qatar University | Smart fault detection device to anticipate impending faults in power transformers |
| CN106597260B (en) * | 2016-12-29 | 2020-04-03 | 合肥工业大学 | Analog circuit fault diagnosis method based on continuous wavelet analysis and ELM network |
| CN107832777B (en) * | 2017-10-12 | 2021-01-26 | 吉林化工学院 | Electric energy quality disturbance identification method adopting time domain compression multi-resolution rapid S-transform feature extraction |
| CN209784432U (en) * | 2019-03-08 | 2019-12-13 | 西华大学 | Electric energy quality disturbance signal acquisition device |
| CN111122941A (en) * | 2019-12-04 | 2020-05-08 | 国网湖南综合能源服务有限公司 | Kaiser window function-based voltage sag characteristic quantity detection method, system and medium for improving S conversion |
| CN111368892B (en) * | 2020-02-27 | 2024-01-30 | 合肥工业大学 | Electric energy quality disturbance efficient identification method for generalized S transformation and SVM |
| CN111325485B (en) * | 2020-03-22 | 2022-03-18 | 东北电力大学 | Light-weight gradient elevator power quality disturbance identification method considering internet-of-things bandwidth constraint |
| CN112163558A (en) * | 2020-10-20 | 2021-01-01 | 腾讯科技（深圳）有限公司 | Time series data feature extraction method, device, equipment and storage medium |
-
2021
- 2021-05-10 CN CN202110507182.1A patent/CN113238110B/en active Active
Non-Patent Citations (2)
| Title |
|---|
| A new window function for signal spectrum analysis and FIR filter design;Mahdi Mottaghi-Kashtiban等;《2010 18th Iranian Conference on Electrical Engineering》;20100708;全文 * |
| High quality low order nonrecursive digital filters design using modified Kaiser window;K. Avci等;《2008 6th International Symposium on Communication Systems, Networks and Digital Signal Processing》;20080829;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113238110A (en) | 2021-08-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11408797B2 (en) | GIL fault on-line monitoring system based on vibration signals and support vector machine | |
| CN103308804B (en) | Based on quick K-S converting electric power quality disturbance signal time and frequency parameter extracting method | |
| Hooshmand et al. | Detection and classification of single and combined power quality disturbances using fuzzy systems oriented by particle swarm optimization algorithm | |
| Tan et al. | Numerical model framework of power quality events | |
| CN110441654B (en) | Power quality disturbance detection method for power distribution network based on improved EWT and CMPE | |
| CN102323480B (en) | Electric energy quality analyzing method based on Hilbert-Huang transform | |
| CN110068759A (en) | A kind of fault type preparation method and device | |
| CN104714075B (en) | A kind of electric network voltage flicker envelope parameters extracting method | |
| CN109633431A (en) | The load ratio bridging switch fault recognition method extracted based on vibration signal characteristics | |
| CN111308260B (en) | Electric energy quality monitoring and electric appliance fault analysis system based on wavelet neural network and working method thereof | |
| CN113238110B (en) | Power quality disturbance diagnosis method | |
| CN111368892A (en) | Generalized S transformation and SVM electric energy quality disturbance efficient identification method | |
| Li et al. | Adaptive S transform for feature extraction in voltage sags | |
| Mahla et al. | Recognition of complex and multiple power quality disturbances using wavelet packet-based fast kurtogram and ruled decision tree algorithm | |
| Stanisavljević et al. | A comprehensive overview of digital signal processing methods for voltage disturbance detection and analysis in modern distribution grids with distributed generation | |
| CN113094983B (en) | Online simulation method for multi-dimensional time-varying characteristics of direct-current fault electric arc of photovoltaic system | |
| Hafiz et al. | An approach for classification of power quality disturbances based on Hilbert Huang transform and Relevance vector machine | |
| Jurado et al. | Application of signal processing tools for power quality analysis | |
| CN111310325A (en) | Dynamic simulation method and system of modular multilevel converter | |
| CN108879680A (en) | Multi-functional gird-connected inverter harmonic wave selectivity compensation method based on sliding fourier transfonn | |
| CN109374968A (en) | A kind of VFTO frequency spectrum analysis method based on STFT-WVD transformation | |
| Pujiantara et al. | Improvement of power quality monitoring based on modified S-transform | |
| Liu et al. | Arc fault diagnosis and analysis based on wavelet neural network | |
| Joga et al. | Harmonic source identification in Microgrid using wavelet time frequency analysis | |
| Panneerselvam et al. | Analysis of sensor trapped power quality indices using empirical wavelet transform and rational dilation wavelet transform to achieve high accuracy and frequency resolution |
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 |