CN111723684B - Identification method for transient overvoltage type in offshore wind farm - Google Patents
Identification method for transient overvoltage type in offshore wind farm Download PDFInfo
- Publication number
- CN111723684B CN111723684B CN202010478764.7A CN202010478764A CN111723684B CN 111723684 B CN111723684 B CN 111723684B CN 202010478764 A CN202010478764 A CN 202010478764A CN 111723684 B CN111723684 B CN 111723684B
- Authority
- CN
- China
- Prior art keywords
- frequency
- transient overvoltage
- offshore wind
- wind farm
- low
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E10/00—Energy generation through renewable energy sources
- Y02E10/70—Wind energy
- Y02E10/76—Power conversion electric or electronic aspects
Abstract
The invention provides a method for identifying the type of transient overvoltage in an offshore wind farm. The method comprises the following steps: determining a new mathematical morphological structural element operator suitable for extracting transient overvoltage high-frequency characteristics in the offshore wind farm; carrying out mathematical morphological decomposition based on a structural element operator, extracting high-frequency information of transient overvoltage, and calculating a high-frequency energy value corresponding to the high-frequency information, thereby establishing a high-frequency characteristic quantity; performing wavelet decomposition on the transient overvoltage signal by using a wavelet transformation method, extracting low-frequency information in the transient overvoltage signal, and calculating a low-frequency energy value corresponding to the low-frequency information; combining the high-frequency energy value and the low-frequency energy value to construct a high-low frequency energy ratio identification index; and comprehensively utilizing the high-frequency characteristic quantity and the high-low frequency energy ratio identification index as identification characteristic quantity, and identifying the transient overvoltage type of the offshore wind farm based on a support vector machine multistage classifier. The method has the advantages of clear identification principle, lower calculation complexity and stronger practicability.
Description
Technical Field
The invention relates to the field of power system fault diagnosis application, in particular to a method for identifying the type of transient overvoltage in an offshore wind farm.
Background
The problems of the wind turbine generator set, the wind power plant equipment, the insulation damage, the operation quit and the like exist in the operation process of the research on the offshore wind power plant, and the problems related to the internal overvoltage protection are more prominent. Since the design of an offshore wind farm has its own characteristics, for example, the offshore wind farm is composed of a large number of identical devices, the internal electrical systems are cable networks, the distance between the fans is far, the capacitive effect of the long cables of the collecting system, the internal electrical devices of the wind farm are frequently opened and closed or switched, etc., which are particularly serious, unlike the internal high-frequency transient overvoltage problem caused by the topological structure and parameter characteristics of the traditional power overhead line. Since offshore wind power generation belongs to an emerging industry in China, attention is paid recently to research on overvoltage inside an offshore wind farm. The overvoltage protection of the offshore wind power system mainly comprises three parts of a wind turbine generator, a booster station, overvoltage protection of an in-site power transmission line, lightning protection and grounding, and the problem of overvoltage in the offshore wind power plant caused by frequent operation of electrical equipment is focused.
After the offshore wind farm equipment suffers from multiple impacts of power frequency, operation, lightning and other overvoltage, sudden faults of the equipment can be caused by accumulation effects, and the safe operation of the wind turbine generator is seriously influenced. According to related researches, the damage of the overvoltage to the offshore wind farm is mainly embodied on the dielectric breakdown of the booster transformer at the fan end, the overvoltage signal of the booster transformer port is rapidly and accurately identified, and the method is beneficial to accurately judging the fault type causing the overvoltage and guiding the improvement of the insulation fit of electrical equipment of the offshore wind farm transformer substation.
There are many existing researches on classifying and identifying overvoltage of a power system, but the existing researches on classifying and identifying overvoltage of an offshore wind farm still have blank. Literature (Lobos T, rezmer J, janik P, et al application of wavelets and Prony method for disturbance detection in fixed speed wind norm.electrical Power and Energy Systems,2009, 31:429-436.) proposes a wavelet-Prony method to extract the frequency, amplitude and relative phase of a wind farm voltage oscillating signal. The documents (Huang Yanling, sima Wenxia, yang Qing, etc. the classification and identification of overvoltage of the power system based on the measured data [ J ]. The power system is automated, 2012, 36 (4): 85-90.) the classification and identification of overvoltage is realized by analyzing the overvoltage of the power system by wavelet multi-resolution transformation. Application of S transformation modular matrices and least squares SVMs in lightning and operational overvoltage identification [ J ] power automation devices, 2012, 32 (8): 35-40.) the literature (Du Lin, li Xin, sima Wenxia, et al) uses S transformation to extract overvoltage fault signatures in high voltage transmission lines. The research shows that the time-frequency domain decomposition algorithm based on integral transformation can realize classification and identification of overvoltage signals in a power system. However, the wavelet transformation and S transformation equal division algorithms have the problems of energy leakage, frequency domain confusion and frequency band characteristic interference when complex nonlinear high-frequency information such as transient overvoltage signals in the offshore wind farm is processed.
The mathematical morphology belongs to a nonlinear mathematical analysis method, is suitable for signal time domain feature analysis, has the advantages of non-attenuation of amplitude and non-deviation of phase in the aspect of signal processing, and has small data window and high calculation speed. When the transient overvoltage in the offshore wind farm is processed by utilizing mathematical morphology, the problem of frequency band decomposition existing in the processing of complex nonlinear high-frequency signals by the method can be solved to a certain extent. The method firstly provides a new morphological structure operator, extracts high-frequency information of transient overvoltage by using a mathematical morphology decomposition method, calculates corresponding high-frequency energy values and constructs a high-frequency characteristic quantity. Then, the low frequency information of the transient overvoltage is extracted through wavelet transformation, corresponding low frequency energy values are calculated, and a high-low frequency energy ratio (HRL, high frequency energy Relative to Low frequency energy) is calculated by combining the high frequency energy values obtained through mathematical morphology. And finally, comprehensively utilizing the high-frequency characteristic quantity and the HRL identification characteristic quantity, and carrying out classification identification on transient overvoltage in different types of offshore wind farms based on a support vector machine multistage classifier model.
Disclosure of Invention
The invention mainly solves the technical problem of providing a method for identifying the type of transient overvoltage in the offshore wind farm based on mathematical morphology and wavelet transformation aiming at the situation of transient overvoltage caused by frequent switching or faults of a vacuum circuit breaker of the offshore wind farm. The identified transient overvoltage waveform refers to figure one.
The object of the invention is achieved by at least one of the following technical solutions.
The method for identifying the type of the transient overvoltage in the offshore wind farm comprises the following steps:
s1, determining a new mathematical morphological structural element operator suitable for extracting transient overvoltage high-frequency characteristics in an offshore wind farm;
s2, carrying out mathematical morphological decomposition based on the structural element operator proposed in the step S1, extracting high-frequency information of transient overvoltage, and calculating a high-frequency energy value corresponding to the high-frequency information so as to establish a high-frequency characteristic quantity;
s3, performing wavelet decomposition on the transient overvoltage signal by using a wavelet transformation method, extracting low-frequency information in the transient overvoltage signal, and calculating a low-frequency energy value corresponding to the low-frequency information;
s4, combining the high-frequency energy value extracted by mathematical morphology and the low-frequency energy value extracted by wavelet transformation to construct a high-low frequency energy ratio (HRL, high frequency energy Relative to Low frequency energy) identification index;
and S5, comprehensively utilizing the high-frequency characteristic quantity and the high-low frequency energy ratio identification index as identification characteristic quantity, and identifying the transient overvoltage type of the offshore wind farm based on a support vector machine multistage classifier.
Further, in step S1, a combination test is performed according to mathematical morphology common structural elements, and in combination with actual waveform characteristics, a new structural element capable of effectively extracting transient overvoltage high-frequency characteristics of the offshore wind farm is provided; the mathematical morphology common structural elements comprise straight lines, triangles and sine and cosine;
the new structural elements are multi-layer mathematical morphology structural elements, and each layer of the multi-layer mathematical morphology structural element operators is similar in shape, so that the amplitude of each layer of structural operator is not smaller than the maximum change amplitude of overvoltage in the power system signal in order to achieve the purposes of keeping steady state information and extracting high-frequency characteristic information as far as possible; the length of each layer of structural elements satisfies the following formula:
L n ＝6·2 n-1 ·m+1,n＝1,2,...,N；
wherein N is the number of corresponding structural layers, N is the selected total number of layers, m is the corresponding length base number, and the specific numerical value is required to be set manually according to actual conditions and is related to the length of the processed data; the longer the data length is, the larger the value of m is;
the construction formula of the new structural element is as follows:
wherein G is n For the operator function of the nth layer structure, A p For maximum variation amplitude of overvoltage signal, L n Length of structural element of n-th layer, m n Is the length radix corresponding to the nth layer.
Further, in step S2, using the new structural elements set forth in step S1, a mathematical morphological multi-layer decomposition is performed on the transient overvoltage signal of the offshore wind farm caused by the fault or the operation condition, and the mathematical morphological operation is specifically as follows:
wherein f and g represent system signals and structural elements, respectively, ∈o is a morphological open operator, +.v is a morphological close operator;
the residual information and high-frequency characteristic information of the transient overvoltage in the time domain are obtained, and the method concretely comprises the following steps:
wherein G is n Respectively an nth layer morphological structure operator; f (F) 0 (x) As an original transient overvoltage signal, F n (x) For the remaining information of the n-th layer decomposition, M (F n-1 ,G n ) (x) high-frequency decomposition information of the n-th layer; y (x) is the sum of the corresponding morphological high frequency decomposition information value results;
then, the corresponding high-frequency energy values of the original transient overvoltage signals A, B, C are calculated, so that a high-frequency characteristic quantity is established, and the method is concretely as follows:
H k ＝E(Y k (x)),k＝A,B,C；
wherein Y is k (x) High frequency information representing mathematical morphological decomposition of the kth phase, E (·) is a function of the calculated voltage energy value, H k A high frequency energy value representing the k-th phase; the high-frequency energy values of A, B, C three phases of the overvoltage signals are ordered from high to low, and the high-frequency characteristic values are defined as follows: ζ= [ H ] 1 H 2 H 3 ]In which H 1 ,H 2 ,H 3 Respectively, a maximum value, a middle value and a minimum value in the three-phase high-frequency energy values.
Further, the function for calculating the voltage energy value is specifically defined as follows:
wherein x is an input signal, D x For the definition domain of the input signal, num is the length of the input signal, E (x) is the voltage energy value of the input signal, threshold processing is adopted for selecting the effective signal to calculate the voltage energy value, and the absolute value of the signal voltage energy value is calculated to be larger than q, wherein q is 0.5kV.
Further, in step S3, performing fast fourier transform decomposition on the transient overvoltage signal inside the offshore wind farm to obtain a signal amplitude-frequency characteristic, thereby obtaining a frequency band where low-frequency information of the transient overvoltage signal inside the offshore wind farm is located; determining the wavelet decomposition layer number according to the sampling frequency of the transient overvoltage signal and the Nyquist sampling theorem so as to acquire low-frequency information of the transient overvoltage signal in the offshore wind farm; the db wavelet decomposition base suitable for processing the power system signal is selected, so that wavelet decomposition is carried out on the transient overvoltage signal to obtain low-frequency information of the overvoltage signal, and the corresponding low-frequency energy value is calculated, wherein the method comprises the following steps of:
the voltage energy value of the overvoltage low-frequency information is used for representing the low-frequency energy value, and is defined as:
L k ＝E(X k (x)),k＝A,B,C；
wherein X is k (x) Overvoltage low frequency information representing wavelet decomposition of the kth phase, E (·) represents a function of the calculated voltage energy value, L k Representing the low frequency energy value of the k-th phase.
Further, in step S4, the high-frequency energy value and the low-frequency energy value obtained in step S2 and step S3 are combined, and the high-frequency energy ratio based on the time domain and frequency domain information is calculated and used as a feature quantity for classifying and identifying transient overvoltage in different types of offshore wind farms, which is specifically as follows:
wherein HRL represents the high-low frequency energy ratio, H A 、H B 、H C A, B, C, L A 、L B 、L C A, B, C low frequency energy values of three phases, respectively.
Further, in step S5, the high-frequency characteristic quantity and the high-low frequency energy ratio recognition index calculated in step S3 and step S4 are used as recognition characteristic quantity, transient overvoltage simulation data inside the offshore wind farm are used as learning samples, and the support vector machine multi-stage classifier is trained and learned; the support vector machine multistage classifier obtained after the training of the transient overvoltage simulation data in the offshore wind farm is used for realizing classification and identification of the actual transient overvoltage data in the offshore wind farm; the simulation data based on PSCAD/EMTDC and the actual measurement data of the transient overvoltage of the offshore wind farm are tested, whether the identification result is accurate and reliable is determined according to the test result, and then various transient overvoltage types of the offshore wind farm are effectively identified.
Further, the training and learning of the support vector machine multi-stage classifier is specifically as follows:
firstly, taking a part of simulation data as a training sample, performing mark-unification pretreatment on input characteristic quantity, and labeling a training sample set; adjusting optimization parameters of the support vector machine multi-stage classifier to obtain a classifier meeting the precision requirement; and finally, using the rest simulation data samples to check whether the acquired classifier precision meets the requirement, thereby obtaining the support vector machine multi-stage classifier.
Compared with the prior art, the invention has the advantages that:
the method is clear in principle, low in calculation complexity and high in practicability, and the proposed overvoltage classification and identification index combines information of actual waveforms in time domain and frequency domain, so that transient overvoltages of different offshore wind power plants can be accurately classified quantitatively.
Drawings
FIG. 1 is a waveform diagram of actual measurement of transient overvoltage of an offshore wind farm with inductive load switching off in an embodiment of the invention;
FIG. 2 is a waveform diagram of actual measurement of transient overvoltage of a closing transient state of an offshore wind farm with inductive load in an embodiment of the invention;
FIG. 3 is a waveform diagram of actual measurement of the no-load brake-off transient overvoltage of the offshore wind farm in an embodiment of the invention;
FIG. 4 is a waveform diagram of actual measurement of the no-load closing transient overvoltage of the offshore wind farm in an embodiment of the invention;
FIG. 5 is a schematic diagram of a structural element operator according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of a transient overvoltage waveform with inductive load switching in an embodiment of the present invention;
FIG. 7 is a first layer exploded view of a transient overvoltage waveform with inductive load shedding in an embodiment of the present invention;
FIG. 8 is a second layer exploded view of a transient overvoltage waveform with inductive load shedding in an embodiment of the present invention;
FIG. 9 is a third layer exploded view of a transient overvoltage waveform with inductive load shedding in an embodiment of the present invention;
FIG. 10 is a total waveform diagram of the difference of transient overvoltage waveforms with inductive load switching in an embodiment of the present invention;
FIG. 11 is a flowchart of a method for identifying transient overvoltage types inside an offshore wind farm in an embodiment of the invention;
FIG. 12 is a diagram showing simulated data classification versus actual tags in an embodiment of the present invention;
FIG. 13 is a graph showing the comparison of the classification of experimental data with the actual labels in the examples of the present invention.
Detailed Description
The following describes the embodiments of the present invention in further detail with reference to examples and drawings, but the embodiments of the present invention are not limited thereto.
Examples:
a method for identifying the type of transient overvoltage in an offshore wind farm is shown in FIG. 11 and comprises the following steps:
s1, determining a new mathematical morphological structural element operator suitable for extracting transient overvoltage high-frequency characteristics in an offshore wind farm;
carrying out combination test according to mathematical morphology common structural elements and combining with actual waveform characteristics, and providing a new structural element capable of effectively extracting transient overvoltage high-frequency characteristics of the offshore wind farm; the mathematical morphology common structural elements comprise straight lines, triangles and sine and cosine;
the new structural elements are multi-layer mathematical morphology structural elements, and each layer of the multi-layer mathematical morphology structural element operators is similar in shape, so that the amplitude of each layer of structural operator is not smaller than the maximum change amplitude of overvoltage in the power system signal in order to achieve the purposes of keeping steady state information and extracting high-frequency characteristic information as far as possible; the length of each layer of structural elements satisfies the following formula:
L n ＝6·2 n-1 ·m+1,n＝1,2,...,N；
wherein N is the number of corresponding structural layers, N is the selected total number of layers, m is the corresponding length base number, and the specific numerical value is required to be set manually according to actual conditions and is related to the length of the processed data; the longer the data length is, the larger the value of m is;
in this embodiment, N is 3; since the length of the transient overvoltage simulation data is smaller than that of the transient overvoltage measured data, m is 2 and 10 respectively when the transient overvoltage simulation data and the transient overvoltage measured data are processed;
the construction formula of the new structural element is as follows:
wherein G is n For the operator function of the nth layer structure, A p For maximum variation amplitude of overvoltage signal, L n Length of structural element of n-th layer, m n Length base number of the corresponding n layer; in this embodiment, N is 3; the mathematical morphology structural element shape is shown in fig. 5.
S2, carrying out mathematical morphological decomposition based on the structural element operator proposed in the step S1, extracting high-frequency information of transient overvoltage, and calculating a high-frequency energy value corresponding to the high-frequency information so as to establish a high-frequency characteristic quantity;
in this embodiment, the transient overvoltage signal of the offshore wind farm is subjected to three-layer mathematical morphological decomposition according to the new structural element proposed in step S1, and the mathematical morphological operation is specifically as follows:
wherein f and g represent system signals and structural elements, respectively, ∈o is a morphological open operator, +.v is a morphological close operator;
the residual information and high-frequency characteristic information of the transient overvoltage in the time domain are obtained, and the method concretely comprises the following steps:
F 1 (x)＝F 0 (x)-M(F,G 1 )(x)
F 2 (x)＝F 1 (x)-M(F 1 ,G 2 )(x)
F 3 (x)＝F 2 (x)-M(F 2 ,G 3 )(x)
Y(x)＝M(F,G 1 )(x)+M(F 1 ,G 2 )(x)+M(F 2 ,G 3 )(x)
wherein G is 1 ,G 2 ,G 3 Morphological structure operators of the first layer, the second layer and the third layer respectively; f (F) 0 (x) As an original transient overvoltage signal, F 1 (x)，F 2 (x)，F 3 (x) Residual information decomposed for the first, second and third layers, respectively, M (F, G 1 )(x)、M(F 1 ,G 2 )(x)、M(F 2 ,G 3 ) (x) high-frequency decomposition information of the first layer, the second layer, and the third layer, respectively; y (x) is the sum of the corresponding morphological high frequency decomposition information value results;
in the present embodiment, a group of switching data waveforms with inductive load circuit breaker is taken as an example, and the decomposition results are shown in fig. 6, 7, 8, 9 and 10. The mathematically morphologically decomposed waveform suppresses and eliminates a stationary portion of the original transient overvoltage, which appears in the time domain as a signal distributed around the zero value. Whereas for steep abrupt changes in the original transient overvoltage, it is embodied in the decomposed waveform as a series of pulse waves in the time domain. The pulse wave accurately captures the time of the original transient overvoltage jump, corresponding to the steep jump part in the transient overvoltage waveform. Compared with the original transient overvoltage, the decomposition waveform has higher recognition degree in the time domain, and the change characteristic of the original transient overvoltage can be intuitively reflected. In addition, as the number of decomposition layers increases, the value of the abrupt signal extracted from the decomposed waveform increases. In this embodiment, when the number of decomposition layers is 3, the difference total waveform at this time is enough to obtain the abrupt signal value of the original transient overvoltage, which can be used to characterize the high-frequency characteristic information of the original transient overvoltage.
Then, the corresponding high-frequency energy values of the original transient overvoltage signals A, B, C are calculated, so that a high-frequency characteristic quantity is established, and the method is concretely as follows:
H k ＝E(Y k (x)),k＝A,B,C；
wherein Y is k (x) High frequency information representing mathematical morphological decomposition of the kth phase, E (·) is a function of the calculated voltage energy value, H k A high frequency energy value representing the k-th phase; the high-frequency energy values of A, B, C three phases of the overvoltage signals are ordered from high to low, and the high-frequency characteristic values are defined as follows: ζ= [ H ] 1 H 2 H 3 ]In which H 1 ,H 2 ,H 3 Respectively, a maximum value, a middle value and a minimum value in the three-phase high-frequency energy values.
The function of the calculated voltage energy value is specifically defined as follows:
wherein x is an input signal, D x For the definition domain of the input signal, num is the length of the input signal, E (x) is the voltage energy value of the input signal, threshold processing is adopted for selecting the effective signal to calculate the voltage energy value, and the absolute value of the signal voltage energy value is calculated to be larger than q, wherein q is 0.5kV.
S3, performing wavelet decomposition on the transient overvoltage signal by using a wavelet transformation method, extracting low-frequency information in the transient overvoltage signal, and calculating a low-frequency energy value corresponding to the low-frequency information;
in this embodiment, considering that the db wavelet base function has a good effect in processing the power system signal, the db16 wavelet base function is finally selected as the decomposition function by comparing the db wavelet base co-correlation coefficients. The transient overvoltage signals are decomposed by adopting a fast Fourier decomposition method, so that low-frequency information can be obtained and mainly concentrated in a frequency band of 0-600 Hz. And determining the wavelet decomposition layer number by the sampling frequency of the transient overvoltage signal and the Nyquist sampling theorem so as to acquire information containing the frequency range of 0-600 Hz. And finally, calculating a corresponding low-frequency energy value according to the overvoltage low-frequency information, wherein the low-frequency energy value is as follows:
the voltage energy value of the overvoltage low-frequency information is used for representing the low-frequency energy value, and is defined as:
L k ＝E(X k (x)),k＝A,B,C；
wherein X is k (x) Overvoltage low frequency information representing wavelet decomposition of the kth phase, E (·) represents a function of the calculated voltage energy value, L k Representing the low frequency energy value of the k-th phase.
S4, combining the high-frequency energy value extracted by mathematical morphology and the low-frequency energy value extracted by wavelet transformation to construct a high-low frequency energy ratio (HRL, high frequency energy Relative to Low frequency energy) identification index;
and (3) combining the high-frequency energy value and the low-frequency energy value acquired in the step (S2) and the step (S3), calculating a high-frequency energy ratio based on time domain and frequency domain information, and using the high-frequency energy ratio and the low-frequency energy ratio as a characteristic quantity for classifying and identifying transient overvoltage inside different types of offshore wind farms, wherein the characteristic quantity is as follows:
wherein HRL represents the high-low frequency energy ratio, H A 、H B 、H C A, B, C, L A 、L B 、L C A, B, C low frequency energy values of three phases, respectively.
S5, comprehensively utilizing the high-frequency characteristic quantity and the high-low frequency energy ratio identification index as identification characteristic quantity, and identifying the type of the transient overvoltage of the offshore wind farm based on a support vector machine multistage classifier;
the high-frequency characteristic quantity and the high-low frequency energy ratio identification index calculated in the step S3 and the step S4 are used as identification characteristic quantity, transient overvoltage simulation data inside the offshore wind farm are used as learning samples, and the support vector machine multistage classifier is trained and learned, specifically as follows:
firstly, taking a part of simulation data as a training sample, performing mark-unification pretreatment on input characteristic quantity, and labeling a training sample set; adjusting optimization parameters of the support vector machine multi-stage classifier to obtain a classifier meeting the precision requirement; finally, the rest simulation data samples are used for checking whether the acquired classifier precision meets the requirement, so that a support vector machine multi-stage classifier is obtained;
the support vector machine multistage classifier obtained after the training of the transient overvoltage simulation data in the offshore wind farm is used for realizing classification and identification of the actual transient overvoltage data in the offshore wind farm; the simulation data based on PSCAD/EMTDC and the actual measurement data of the transient overvoltage of the offshore wind farm are tested, whether the identification result is accurate and reliable is determined according to the test result, and then various transient overvoltage types of the offshore wind farm are effectively identified.
In this embodiment, the processed actual waveforms of four internal transient overvoltages of the offshore wind farm are shown in fig. 1 to 4. Fig. 1, 2, 3 and 4 respectively represent four waveforms of inductive load opening, inductive load closing, no-load opening and no-load closing.
In the embodiment, firstly, modeling is performed on a certain Guangdong offshore wind farm based on PSCAD/EMDTC simulation software, an offshore wind farm equivalent simulation platform is built to obtain 30 groups of four overvoltage type data including inductive load opening, inductive load closing, no-load opening and no-load closing, and the total of 120 groups of simulation sample data are obtained. The 60 groups of simulation sample data are used for training the support vector machine multi-stage classifier, the rest 60 groups of data are used as detection samples, the result shows that the recognition accuracy of the classifier reaches 100%, and the result is shown in fig. 12. And then 8 groups of four types of inductive load opening, inductive load closing, no-load opening and no-load closing are obtained through an experimental platform simulated by a collection system of a Guangdong offshore wind farm, the experimental sample data of 32 groups are taken as test samples, the identification is carried out by using a support vector machine multi-stage classifier trained based on simulation data, the result shows that the identification accuracy of the classifier is 93.75%, as shown in fig. 13, and the test result shows that the invention can accurately judge transient overvoltage of different types of offshore wind farms.
The above-described embodiment is a preferred embodiment of the present invention, and various changes and modifications may be made by the worker in the above description without departing from the technical spirit of the present invention. The technical scope of the present invention is not limited to the description, but must be determined according to the scope of claims.
Claims (5)
1. The method for identifying the type of the transient overvoltage in the offshore wind farm is characterized by comprising the following steps of:
s1, determining a new mathematical morphological structural element operator suitable for extracting transient overvoltage high-frequency characteristics in an offshore wind farm;
s2, carrying out mathematical morphological decomposition based on the structural element operator proposed in the step S1, extracting high-frequency information of transient overvoltage, and calculating a high-frequency energy value corresponding to the high-frequency information so as to establish a high-frequency characteristic quantity; carrying out combination test according to mathematical morphology common structural elements and combining with actual waveform characteristics, and providing a new structural element capable of effectively extracting transient overvoltage high-frequency characteristics of the offshore wind farm; the mathematical morphology common structural elements comprise straight lines, triangles and sine and cosine;
the new structural elements are multi-layer mathematical morphology structural elements, and each layer of the multi-layer mathematical morphology structural element operators is similar in shape, so that the amplitude of each layer of structural operator is not smaller than the maximum change amplitude of overvoltage in the power system signal in order to achieve the purposes of keeping steady state information and extracting high-frequency characteristic information as far as possible; the length of each layer of structural elements satisfies the following formula:
L n ＝6·2 n-1 ·m+1,n＝1,2,...,N；
wherein N is the number of corresponding structural layers, N is the selected total number of layers, m is the corresponding length base number, and the specific numerical value is required to be set manually according to actual conditions and is related to the length of the processed data; the longer the data length is, the larger the value of m is;
the construction formula of the new structural element is as follows:
wherein G is n For the operator function of the nth layer structure, A p For maximum variation amplitude of overvoltage signal, L n Length of structural element of n-th layer, m n Length base number of the corresponding n layer;
s3, performing wavelet decomposition on the transient overvoltage signal by using a wavelet transformation method, extracting low-frequency information in the transient overvoltage signal, and calculating a low-frequency energy value corresponding to the low-frequency information; using the new structural elements proposed in the step S1 to carry out mathematical morphology multi-layer decomposition on transient overvoltage signals of the offshore wind farm caused by faults or operation conditions, wherein the mathematical morphology operation is specifically as follows:
wherein f and g represent system signals and structural elements, respectively,morphological open operator, morphological close operator;
the residual information and high-frequency characteristic information of the transient overvoltage in the time domain are obtained, and the method concretely comprises the following steps:
wherein G is n Respectively an nth layer morphological structure operator; f (F) 0 (x) As an original transient overvoltage signal, F n (x) For the remaining information of the n-th layer decomposition, M (F n-1 ,G n ) (x) high-frequency decomposition information of the n-th layer; y (x) is the sum of the corresponding morphological high frequency decomposition information value results;
then, the corresponding high-frequency energy values of the original transient overvoltage signals A, B, C are calculated, so that a high-frequency characteristic quantity is established, and the method is concretely as follows:
H k ＝E(Y k (x)),k＝A,B,C；
wherein Y is k (x) High frequency information representing mathematical morphological decomposition of the kth phase, E (·) is a function of the calculated voltage energy value, H k A high frequency energy value representing the k-th phase; the high-frequency energy values of A, B, C three phases of the overvoltage signals are ordered from high to low, and the high-frequency characteristic values are defined as follows: ζ= [ H ] 1 H 2 H 3 ]In which H 1 ,H 2 ,H 3 Respectively a maximum value, a middle value and a minimum value in the three-phase high-frequency energy values; the function of the calculated voltage energy value is specifically defined as follows:
wherein x is an input signal, D x For the definition domain of the input signal, num is the length of the input signal, E (x) is the voltage energy value of the input signal, threshold processing is adopted for selecting the effective signal to calculate the voltage energy value, and the absolute value of the signal voltage energy value is calculated to be larger than q, wherein q is 0.5kV;
s4, combining the high-frequency energy value extracted by mathematical morphology and the low-frequency energy value extracted by wavelet transformation to construct a high-low frequency energy ratio identification index;
and S5, comprehensively utilizing the high-frequency characteristic quantity and the high-low frequency energy ratio identification index as identification characteristic quantity, and identifying the transient overvoltage type of the offshore wind farm based on a support vector machine multistage classifier.
2. The method for identifying the transient overvoltage type inside the offshore wind farm according to claim 1, wherein in step S3, fast Fourier transform decomposition is carried out on transient overvoltage signals inside the offshore wind farm to obtain signal amplitude-frequency characteristics, and accordingly a frequency band where low-frequency information of the transient overvoltage signals inside the offshore wind farm is located is obtained; determining the wavelet decomposition layer number according to the sampling frequency of the transient overvoltage signal and the Nyquist sampling theorem so as to acquire low-frequency information of the transient overvoltage signal in the offshore wind farm; the db wavelet decomposition base suitable for processing the power system signal is selected, so that wavelet decomposition is carried out on the transient overvoltage signal to obtain low-frequency information of the overvoltage signal, and the corresponding low-frequency energy value is calculated, wherein the method comprises the following steps of:
the voltage energy value of the overvoltage low-frequency information is used for representing the low-frequency energy value, and is defined as:
L k ＝E(X k (x)),k＝A,B,C；
wherein X is k (x) Overvoltage low frequency information representing wavelet decomposition of the kth phase, E (·) represents a function of the calculated voltage energy value, L k Representing the low frequency energy value of the k-th phase.
3. The method for identifying the type of the transient overvoltage in the offshore wind farm according to claim 1, wherein in the step S4, the high-frequency energy ratio based on the time domain and the frequency domain information is calculated by combining the high-frequency energy value and the low-frequency energy value obtained in the step S2 and the step S3, and the high-frequency energy ratio is used as a characteristic quantity for identifying the transient overvoltage in the offshore wind farm of different types by classification, and the characteristic quantity is as follows:
wherein HRL represents the high-low frequency energy ratio, H A 、H B 、H C A, B, C, L A 、L B 、L C A, B, C low frequency energy values of three phases, respectively.
4. The method for identifying the type of the transient overvoltage in the offshore wind farm according to claim 1, wherein in the step S5, the high-frequency characteristic quantity and the high-low-frequency energy ratio identification index calculated in the step S3 and the step S4 are used as identification characteristic quantity, transient overvoltage simulation data in the offshore wind farm are used as learning samples, and the support vector machine multistage classifier is trained and learned; the support vector machine multistage classifier obtained after the training of the transient overvoltage simulation data in the offshore wind farm is used for realizing classification and identification of the actual transient overvoltage data in the offshore wind farm; the simulation data based on PSCAD/EMTDC and the actual measurement data of the transient overvoltage of the offshore wind farm are tested, whether the identification result is accurate and reliable is determined according to the test result, and then various transient overvoltage types of the offshore wind farm are effectively identified.
5. The method for identifying the transient overvoltage type in the offshore wind farm according to claim 4, wherein the training and learning of the support vector machine multi-stage classifier is specifically as follows:
firstly, taking a part of simulation data as a training sample, performing mark-unification pretreatment on input characteristic quantity, and labeling a training sample set; adjusting optimization parameters of the support vector machine multi-stage classifier to obtain a classifier meeting the precision requirement; and finally, using the rest simulation data samples to check whether the acquired classifier precision meets the requirement, thereby obtaining the support vector machine multi-stage classifier.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010478764.7A CN111723684B (en) | 2020-05-29 | 2020-05-29 | Identification method for transient overvoltage type in offshore wind farm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010478764.7A CN111723684B (en) | 2020-05-29 | 2020-05-29 | Identification method for transient overvoltage type in offshore wind farm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111723684A CN111723684A (en) | 2020-09-29 |
| CN111723684B true CN111723684B (en) | 2023-07-21 |
Family
ID=72565553
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010478764.7A Active CN111723684B (en) | 2020-05-29 | 2020-05-29 | Identification method for transient overvoltage type in offshore wind farm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111723684B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112684282B (en) * | 2020-11-12 | 2022-07-19 | 国网河北省电力有限公司电力科学研究院 | Power distribution network single-phase earth fault identification method and device and terminal equipment |
| CN112858902B (en) * | 2021-02-02 | 2023-01-24 | 南方电网数字电网研究院有限公司 | Miniature circuit breaker monitoring method and device, computer equipment and storage medium |
| CN113343877B (en) * | 2021-06-18 | 2022-02-08 | 上海梦象智能科技有限公司 | Electrical appliance fingerprint feature extraction method based on different Prony methods |
| CN115144647B (en) * | 2022-08-30 | 2022-12-30 | 国网江西省电力有限公司电力科学研究院 | Intelligent overvoltage identification method and system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102135558A (en) * | 2010-11-05 | 2011-07-27 | 重庆市电力公司綦南供电局 | Acquisition and hierarchical identification system of overvoltage data and hierarchical pattern identification method of overvoltage types |
| CN102445620A (en) * | 2011-11-21 | 2012-05-09 | 东北大学 | Transient power quality detection device and method for the same |
| CN103197135A (en) * | 2013-03-29 | 2013-07-10 | 国家电网公司 | Transient overvoltage monitoring method |
| CN103197124A (en) * | 2013-03-14 | 2013-07-10 | 重庆市电力公司电力科学研究院 | Overvoltage identification method based on time-frequency matrix singular value |
| CN105447502A (en) * | 2015-11-05 | 2016-03-30 | 国家电网公司 | Transient power disturbance identification method based on S conversion and improved SVM algorithm |
| CN106227912A (en) * | 2016-07-04 | 2016-12-14 | 华南理工大学 | Obtain and simplify the analysis method of maximum transient overvoltage under wind energy turbine set cable system asymmetry short circuit fault |
| CN106771520A (en) * | 2016-12-15 | 2017-05-31 | 福州大学 | A kind of power distribution network temporary overvoltage classifying identification method and device |
| CN108181107A (en) * | 2018-01-12 | 2018-06-19 | 东北电力大学 | The Wind turbines bearing mechanical method for diagnosing faults of meter and more class objects |
| CN108508268A (en) * | 2018-04-02 | 2018-09-07 | 山西大学 | A kind of novel digital type flickermeter decomposed based on time-frequency domain |
| CN109782158A (en) * | 2019-02-19 | 2019-05-21 | 桂林电子科技大学 | A kind of Analog circuit diagnosis method based on multiclass classification |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6462554B2 (en) * | 2000-02-15 | 2002-10-08 | Intel Corporation | System and method for deterring operation of a system outside a specified frequency range |
-
2020
- 2020-05-29 CN CN202010478764.7A patent/CN111723684B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102135558A (en) * | 2010-11-05 | 2011-07-27 | 重庆市电力公司綦南供电局 | Acquisition and hierarchical identification system of overvoltage data and hierarchical pattern identification method of overvoltage types |
| CN102445620A (en) * | 2011-11-21 | 2012-05-09 | 东北大学 | Transient power quality detection device and method for the same |
| CN103197124A (en) * | 2013-03-14 | 2013-07-10 | 重庆市电力公司电力科学研究院 | Overvoltage identification method based on time-frequency matrix singular value |
| CN103197135A (en) * | 2013-03-29 | 2013-07-10 | 国家电网公司 | Transient overvoltage monitoring method |
| CN105447502A (en) * | 2015-11-05 | 2016-03-30 | 国家电网公司 | Transient power disturbance identification method based on S conversion and improved SVM algorithm |
| CN106227912A (en) * | 2016-07-04 | 2016-12-14 | 华南理工大学 | Obtain and simplify the analysis method of maximum transient overvoltage under wind energy turbine set cable system asymmetry short circuit fault |
| CN106771520A (en) * | 2016-12-15 | 2017-05-31 | 福州大学 | A kind of power distribution network temporary overvoltage classifying identification method and device |
| CN108181107A (en) * | 2018-01-12 | 2018-06-19 | 东北电力大学 | The Wind turbines bearing mechanical method for diagnosing faults of meter and more class objects |
| CN108508268A (en) * | 2018-04-02 | 2018-09-07 | 山西大学 | A kind of novel digital type flickermeter decomposed based on time-frequency domain |
| CN109782158A (en) * | 2019-02-19 | 2019-05-21 | 桂林电子科技大学 | A kind of Analog circuit diagnosis method based on multiclass classification |
Non-Patent Citations (3)
| Title |
|---|
| Internal overvoltages in transformer windings in frequency domain;Marek Florkowski 等;《2010 Annual Report Conference on Electrical Insulation and Dielectric Phenomena》;20101231;1-4 * |
| S变换模矩阵和最小二乘SVM在雷电及操作;杜林 等;《电力自动化设备》;20120831;第32卷(第8期);35-40 * |
| 基于稀疏自编码的过电压特征提取与分类方法;刘明军 等;《High Voltage Apparatus》;20191016;第55卷(第10期);70-75 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111723684A (en) | 2020-09-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111723684B (en) | Identification method for transient overvoltage type in offshore wind farm | |
| CN109444656B (en) | Online diagnosis method for deformation position of transformer winding | |
| CN106443316B (en) | Multi-information detection method and device for deformation state of power transformer winding | |
| CN107167702A (en) | A kind of distribution feeder fault type recognition method and device | |
| CN109444667B (en) | Power distribution network early fault classification method and device based on convolutional neural network | |
| CN106226635A (en) | A kind of distribution feeder fault type recognition method and device | |
| CN109443190B (en) | Transient traveling wave-based transformer winding deformation online monitoring method and device | |
| CN112147462A (en) | Power transmission line fault identification method based on deep learning | |
| CN112462193B (en) | Automatic reclosing judgment method for power distribution network based on real-time fault filtering data | |
| CN111999591B (en) | Method for identifying abnormal state of primary equipment of power distribution network | |
| Li et al. | A fault pattern and convolutional neural network based single-phase earth fault identification method for distribution network | |
| CN113466634A (en) | Ground fault waveform identification method based on fault indicator | |
| Nguyen et al. | Wavelet-based Neural Network for recognition of faults at NHABE power substation of the Vietnam power system | |
| CN107578016B (en) | Sparse representation-based automatic residual current waveform identification method | |
| CN113191192A (en) | Breaker fault detection method based on wavelet analysis and fuzzy neural network algorithm | |
| Zhang et al. | Two terminals protection method for HVDC transmission lines based on maximal overlap discrete wavelet transform and cross-correlation coefficient of traveling wave | |
| Gangwar et al. | Detection and classification of faults on transmission line using time-frequency approach of current transients | |
| Lavanya et al. | A deep learning technique for detecting high impedance faults in medium voltage distribution networks | |
| CN114742295A (en) | Equipment power utilization fault prediction method based on wavelet transform technology | |
| Yican et al. | Feature analysis of overvoltage in offshore wind farms using mathematical morphology | |
| Asman et al. | Fault signature analysis for medium voltage overhead line based on new characteristic indices | |
| Viet et al. | A method for monitoring voltage disturbances based on discrete wavelet transform and adaptive linear neural network | |
| Zeng et al. | High impedance fault detection in distribution network based on S-transform and average singular entropy | |
| Zhang et al. | A fast fault detection method for controlled fault interruption based on wavelet singularities and BP neural network | |
| Jinxiao et al. | Research on single-phase high impedance fault identification of neutral ungrounded system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |