CN109541344B - SVM-based fault identification method for modular multi-level direct current power transmission system - Google Patents

SVM-based fault identification method for modular multi-level direct current power transmission system Download PDF

Info

Publication number
CN109541344B
CN109541344B CN201811387279.8A CN201811387279A CN109541344B CN 109541344 B CN109541344 B CN 109541344B CN 201811387279 A CN201811387279 A CN 201811387279A CN 109541344 B CN109541344 B CN 109541344B
Authority
CN
China
Prior art keywords
fault
value
amax
current
identification
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201811387279.8A
Other languages
Chinese (zh)
Other versions
CN109541344A (en
Inventor
胡伟
黄萌
雷杨
刘浴霜
宿磊
苏昊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
State Grid Hubei Electric Power Co Ltd Electric Power Research Institute
State Grid Corp of China SGCC
Wuhan University WHU
Original Assignee
State Grid Hubei Electric Power Co Ltd Electric Power Research Institute
State Grid Corp of China SGCC
Wuhan University WHU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by State Grid Hubei Electric Power Co Ltd Electric Power Research Institute, State Grid Corp of China SGCC, Wuhan University WHU filed Critical State Grid Hubei Electric Power Co Ltd Electric Power Research Institute
Priority to CN201811387279.8A priority Critical patent/CN109541344B/en
Publication of CN109541344A publication Critical patent/CN109541344A/en
Application granted granted Critical
Publication of CN109541344B publication Critical patent/CN109541344B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements 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 provides a fault identification method of a modular multilevel DC power transmission system based on a support vector machine, which is suitable for fault identification of the modular multilevel DC power transmission system. The actual verification result shows that the fault type identified by the model is consistent with the actual fault type, the identification time is short, and quick, accurate and efficient technical support can be provided for fault detection of the modular multilevel direct current transmission system.

Description

SVM-based fault identification method for modular multi-level direct current power transmission system
Technical Field
The invention relates to the field of flexible direct current transmission, in particular to a fault identification method of a modular multi-level direct current transmission system based on a support vector machine.
Background
Modular Multilevel Converter Based high voltage Direct Current transmission (MMC-HVDC) has become a development trend in the field of high voltage Direct Current transmission of voltage source converters due to its specific advantages.
In the actual operation of MMC-HVDC, a plurality of factors can cause the voltage unbalance of an alternating current system. The unipolar ground fault causes the voltage to earth of the direct current non-fault pole and the phase voltage of the alternating current side of the converter station to increase when seen from the direct current side; the single-pole grounding fault can cause the bridge arm of the converter station to generate a serious overcurrent phenomenon; the single-pole disconnection fault can cause a large direct-current voltage change rate of the rectifier station, and serious direct-current overvoltage is caused; from the perspective of the ac side system, there are mainly: the three-phase voltage of the system is asymmetric, an alternating current system has asymmetric faults, the inductance values of reactors (or converter transformers) of each phase of valves are unequal, the switching losses of bridge arms of the converter are inconsistent, and the like. However, if the current MMC-HVDC system in China operates under an abnormal working condition or the system fails, the failure type and the failure position of the system are difficult to accurately judge because the failure has the characteristics of large harm, complex type, high identification difficulty and the like.
At present, in the aspect of fault identification of an MMC-HVDC system, the basic principle, typical fault characteristics, fault identification of an MMC sub-module, direct current fault identification of the MMC-HVDC system and construction of a corresponding protection control strategy are mainly analyzed from the aspect of an electrical mechanism, and the research of further realizing the fault identification according to the running state of the MMC-HVDC system is almost blank. How to extract the characteristics of the electrical information in the MMC station by using the transient signal characteristic extraction method and judge the running state of the MMC-HVDC system and identify the alternating current fault by combining a new artificial intelligence method based on the characteristic extraction result is a new direction for future MMC-HVDC technical research.
The invention combines a machine learning method with the fault identification of a modular multilevel direct current transmission system, provides a fault identification method of the modular multilevel direct current transmission system based on a support vector machine, and can provide quick, accurate and efficient technical support for the fault detection of the modular multilevel direct current transmission system.
Disclosure of Invention
The invention provides a fault identification method of a modular multilevel direct current transmission system based on a support vector machine, which combines a machine learning method with the fault identification of the modular multilevel direct current transmission system, and improves the fault identification method by carrying out classification modeling on characteristic value data on the basis, thereby realizing more accurate and efficient fault identification function.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a fault identification method for a modular multi-level direct current power transmission system based on a support vector machine comprises the following steps:
step one, collecting a characteristic value of a training system from a PSCAD simulation system;
step two, judging the magnitude of a negative peak value of three-phase alternating current in the characteristic value, dividing fault types, and training through a support vector machine to obtain a fault identification model;
and step three, collecting characteristic value data of the modular multilevel direct current transmission system to be identified as a test sample, dividing fault types through fault characteristic values, selecting a corresponding fault identification model, and identifying faults of the corresponding fault identification model.
Further, in the step one, characteristic values of the training system are collected from PSCAD simulation, and the collected content comprises three-phase alternating current voltage amplitude Vam,Vbm,VcmMaximum value V of DC voltagedcmax and minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value IdcminThirteen feature values are counted.
Further, the fault type division criterion in step two is as follows: when Iamax--Ibmax-I and Iamax--Icmax-When | is less than 0.1, classify it as a first type of fault type, when | Iamax--Ibmax-I or Iamax--Icmax-When | is greater than or equal to 0.1, classifying the fault into a second fault type, wherein the two fault types respectively representAnd after the three-phase balanced fault and the three-phase unbalanced fault are subjected to fault type division, carrying out normalization processing on data, and carrying out data training through a support vector machine to obtain fault identification models corresponding to the two types of fault types.
Further, the characteristic value data of the modular multilevel direct-current transmission system to be identified, which is input by the fault identification model in the step three, is the same as that in the step one, namely the amplitude V of the three-phase alternating-current voltageam,Vbm,VcmMaximum value V of DC voltagedcmax and minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value IdcminAnd thirteen characteristic values are counted.
Further, the fault type identification criterion in the third step is the same as that in the second step, namely, the I is the current Iamax--Ibmax-I and Iamax--Icmax-When | is less than 0.1, classify it as a first type of fault type, when | Iamax--Ibmax-I or Iamax--Icmax-And when the | is more than or equal to 0.1, classifying the | into a second type of fault type, selecting a corresponding fault identification model according to the fault type, and identifying the fault.
Further, the time of the fault occurrence in the second step and the third step is judged, specifically, the time is judged by the direct current voltage VdcAnd performing sliding window iteration to serve as a starting criterion of fault detection, namely using the latest real-time sampling data to participate in detection and simultaneously removing the earliest sampling data so as to update the sampling data more quickly.
Further, by applying a DC voltage VdcThe starting criterion for performing sliding window iteration as fault detection is specifically as follows:
(1) setting the width of a sliding window according to the frequency of data sampling, wherein the data sampling is to acquire one data every 0.1ms, the sampling frequency is 10kHz, so that the initial width of the sliding window is 5ms, the frequency is 200Hz, the sliding window can transversely cover 50 discrete waveform points, and the discrete signal point subinterval covered by the initial sliding window is
S=[S1,S2,…,S50]
(2) Selecting a first point S in a sliding window1As the initial reference point, sequentially subtracting the initial reference point from the value of the rest point in the sliding window, if the initial reference point is satisfied
Can be regarded as SNThe point corresponds to the time when the fault occurs, and characteristic values within 10ms after the fault occurs are collected;
if the waveform point in the initial sliding window does not satisfy the condition that
The sliding window is moved to the right and a new waveform point S is read51And eliminating the earliest waveform point S1Forming new sub-intervals
S=[S2,S3,…,S51]
(3) And continuing to perform the judgment until a fault occurrence point is found.
Further, the bad values are eliminated in the second step and the third step by using the Grubbs criterion.
Further, a Grubbs criterion is used for removing bad values, and the specific flow is as follows:
(1) taking mean square error of measurement result of a certain characteristic valueAccording to Bessel's formula, there are
Wherein n represents the number of measurements, ViRepresenting the difference between the measured value and the average value, i.e. the residual error;
(2) using alphos-HermettThe periodic system error is judged according to the criterion, and the criterion is comparisonAndsize of (1), ifThe measured value has no periodic system error, and the influence is not generated at the moment;
(3) removing bad values from the measured values, and obtaining the mean square error according to Bessel formulaUsing the concept of confidence interval to eliminate bad value, under a certain confidence level, inquiring the Grabbs coefficient table to find out the coefficient G, if the measured value x isiSatisfy the requirement ofOrThe measured value xiJudging as bad value, rejecting it
The method comprises the steps of collecting characteristic values of training samples, classifying the training samples into three-phase balanced faults and unbalanced faults according to data characteristics of the training samples, establishing corresponding models, collecting characteristic values of an electric power system to be identified, carrying out normalization processing in the corresponding models after fault type identification, identifying the fault type of the system through a support vector machine method, and outputting the fault type. The actual verification result shows that the fault type identified by the model is consistent with the actual fault type, the identification time is short, and quick, accurate and efficient technical support can be provided for fault detection of the modular multilevel direct current transmission system
Drawings
FIG. 1 is a schematic flow chart of a fault identification method of a modular multi-level direct current power transmission system based on a support vector machine according to the invention;
fig. 2 is a schematic diagram of a sliding window iterative fault detection start criterion taking a dc-side single-pole ground short-circuit fault as an example.
Detailed Description
The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of a fault identification method for a modular multi-level direct current power transmission system based on a support vector machine, the method collects characteristic values of training samples, classifies the training samples into three-phase balanced faults and unbalanced faults according to data characteristics of the training samples, establishes corresponding models, collects characteristic values of an electric power system to be identified, performs normalization processing in the corresponding models after fault type identification, and identifies fault types of the system through the support vector machine method.
The specific implementation mode of the invention is as follows:
(1) collecting characteristic values of a training system from PSCAD simulation, wherein the collected content comprises three-phase alternating voltage amplitude Vam,Vbm,VcmMaximum value V of DC voltagedcmax and minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value Idcmin
(2) When the three phases of the system are balanced, the positive and negative peak values of the three-phase alternating current are consistent, and if the peak values are inconsistent, the three phases of the system are unbalanced, so that the fault judgment criterion is that I isamax--Ibmax-I and Iamax--Icmax-When | is less than 0.1, classify it as a first type of fault type, when | Iamax--Ibmax-I or Iamax--Icmax-And when the | is more than or equal to 0.1, classifying the | into a second fault type, wherein the two fault types respectively represent a three-phase balanced fault and a three-phase unbalanced fault. After the thirteen characteristic values are normalized, the fault is obtained through the training of a support vector machineIdentifying a model;
(3) inputting characteristic values of the modular multilevel direct-current power transmission system to be identified to the fault identification model, wherein the characteristic values comprise three-phase alternating-current voltage amplitude values, positive peak values and negative peak values of three-phase alternating-current, maximum values and minimum values of direct-current voltage and maximum values and minimum values of direct-current;
(4) preliminarily judging the fault type through the fault characteristic value of the modular multilevel direct current transmission system to be identified, wherein the criterion is Iamax--Ibmax-I and Iamax--Icmax-When | is less than 0.1, classify it as a first type of fault type, when | Iamax--Ibmax-I or Iamax--Icmax-And when the | is more than or equal to 0.1, classifying the | into a second type of fault type, selecting a corresponding fault identification model according to the fault type, and inputting the fault identification model into the corresponding model for fault identification.
In this scheme, a DC voltage V is useddcPerforming sliding window iteration to serve as a starting criterion of fault detection, namely using the latest real-time sampling data to participate in detection and simultaneously removing the earliest sampling data so as to update the sampling data more quickly, wherein the specific flow is as follows:
(1) firstly, the width of a sliding window is set according to the frequency of data sampling, the data sampling is to acquire one data every 0.1ms, the sampling frequency is 10kHz, so that the initial width of the sliding window can be 5ms, the frequency is 200Hz, and at the moment, the sliding window can transversely cover 50 discrete waveform points. The discrete signal point subinterval covered by the initial sliding window is
S=[S1,S2,…,S50]
(2) Selecting a first point S in a sliding window1As the initial reference point, sequentially subtracting the initial reference point from the value of the rest point in the sliding window, if the initial reference point is satisfied
Can be regarded as SNAnd (4) the point corresponds to the time when the fault occurs, and characteristic values within 10ms after the fault occurs are collected. The reason why the selection criterion is 2.0kV here is thatThe system is enabled to operate in a stable state, the voltage value of the direct current side of the system still has small fluctuation, and a criterion larger than a fluctuation interval is selected to prevent the malfunction of the fault detection system.
If the waveform point in the initial sliding window does not satisfy the condition that
The sliding window is moved to the right and a new waveform point S is read51And eliminating the earliest waveform point S1Forming new sub-intervals
S=[S2,S3,…,S51]
(3) And continuing to perform the judgment until a fault occurrence point is found.
Fig. 2 shows an example of a single-pole ground short circuit fault on a dc side, where a window slides from an initial position window 1, the window 2 is a window in the sliding process, and when the window slides to a position of a window 3, a criterion is satisfied, and an available fault occurs at 1s, from which characteristic value data within 10ms is collected for subsequent fault identification.
In the scheme, a Grubbs criterion is applied to eliminate bad values, and the specific process is as follows:
(1) taking mean square error of measurement result of a certain characteristic valueAccording to Bessel's formula, there are
Wherein n represents the number of measurements, ViRepresenting the difference between the measured value and the average value, i.e. the residual.
(2) Periodic system errors are judged by using Alps-Hermett criterion, which is comparisonAndthe size of (2). If it isThere is no periodic systematic error in the measured values and no effect is produced.
(3) Removing bad values from the measured values, and obtaining the mean square error according to Bessel formulaTo reject bad values using the concept of confidence intervals. Under a certain confidence level, inquiring the Grabbs coefficient table to find out the coefficient G, if the measured value xi satisfies the requirementOrThe measured value xi is judged as a bad value and is rejected.
Aiming at the invention, a certain practical modularized multi-level direct current transmission system is taken as an example to verify the effectiveness of the provided fault identification method and the accuracy of the identification result.
(1) 13 feature values of the training system are collected. Establishing a PSCAD/EMTDC model of the system, obtaining seven working conditions of system operation under different parameters by changing system parameters (such as alternating-current side voltage, sub-module capacitance, direct-current side voltage, bridge arm inductance and the like) for multiple times, giving a label to each working condition type, and respectively collecting three-phase alternating-current voltage amplitude V at a rectification side within 10ms after a fault occursam,Vbm,VcmMaximum value V of DC voltagedcmax and minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value IdcminThese thirteen items of data of the system failure characteristic values. Obtaining 4 groups of samples under each operating condition, and obtaining the total28 omics study samples were obtained for 7 operating conditions.
(2) The method is characterized in that three-phase current negative peak values are classified according to the relation between the three-phase current negative peak values, as the fault types are more, misjudgment is easy to occur when all training samples are integrally modeled, and in order to improve the identification efficiency and accuracy of a fault identification model, seven types of faults needing to be identified are firstly divided into two main types through simple current amplitude judgment: observing the data characteristics of the training sample, the method can easily find that when the system works under four working conditions of normal work, single-pole grounding fault, single-pole disconnection fault and three-phase symmetrical grounding fault, no difference exists among three phases of alternating current, and at the moment, the three phases of the system are balanced, namely, the labels are 0, 1, 3 and 4; under three working conditions of a bipolar short-circuit fault, an alternating-current side single-phase earth fault and a two-phase short-circuit fault, due to the fact that three-phase currents are unbalanced in a short time after the faults occur, the peak values of the three-phase alternating currents are greatly different, namely the situation that the labels are 2, 5 and 6. After the classification, the two sets of data were normalized separately as shown in the attached table 1.
TABLE 1
Data in the table 1 are read in Matlab by using the libsvread function of Libsvm, namely the data can be changed into training samples and trained, and a first model and a second model for fault recognition are obtained after training.
(3) And collecting characteristic value data of the modular multilevel direct current transmission system to be identified as a test sample, and identifying the fault of the test sample. The method comprises the following specific steps: the modular multilevel direct-current transmission system corresponding to the data of the attached table 1 is simulated in the PSCAD, and because the test sample and the training sample are obtained through simulation, the time of fault occurrence and the data after the fault can be directly read without judging through a sliding window criterion. Extract normal work thereofThree-phase alternating current voltage amplitude V at rectification side within 10ms after seven faults of action, single-pole earth fault, single-pole broken line fault, three-phase symmetrical earth fault, double-pole short-circuit fault, single-phase earth fault at alternating current side and two-phase short-circuit fault occuram,Vbm,VcmMaximum value V of DC voltagedcmaxAnd a minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value IdcminThese thirteen items of system failure characteristic values. Of these thirteen system failure feature values, | I before normalization processing is comparedamax--Ibmax-I and Iamax--Icmax-L. When the system measures | Iamax--Ibmax-I and Iamax--Icmax-When the | is less than 0.1, the test sample is input into the model to be normalized, so that the current operation condition of the system is judged in the categories with labels of 0, 1, 3 and 4 through Libsvm; when the system measures | Iamax--Ibmax-I or Iamax--Icmax-And when the | is more than or equal to 0.1, inputting the test sample into a second model and carrying out normalization processing, so that the current operation condition of the system is judged in the categories with labels of 2, 5 and 6 through Libsvm. The verification results are shown in the attached table 2.
TABLE 2
The result shows that the fault type identified by the model is consistent with the actual fault type, the accuracy of the method for identifying six faults on a single direct current side or alternating current side is up to 100%, the training time is only 0.008134s, the fault type can be identified after 0.018134s after the fault occurs by combining 0.01s required by reading the characteristic value after the fault, and the method plays an extremely important role in actual engineering. In addition, the improved fault identification method of the modular multi-level direct current power transmission system based on the support vector machine is also suitable for other alternating current and direct current power transmission projects, and can provide quick, accurate and efficient technical support for fault detection of other types of alternating current and direct current power transmission systems.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (5)

1. A fault identification method for a modular multi-level direct current power transmission system based on a support vector machine is characterized by comprising the following steps:
step one, collecting a characteristic value of a training system from a PSCAD simulation system;
step two, judging the magnitude of a negative peak value of three-phase alternating current in the characteristic value, dividing fault types, and training through a support vector machine to obtain a fault identification model;
collecting characteristic value data of the modular multilevel direct current transmission system to be identified as a test sample, dividing fault types through fault characteristic values, selecting a corresponding fault identification model, and identifying faults of the corresponding fault identification model;
in the first step, characteristic values of a training system are collected from PSCAD simulation, and the collected content comprises three-phase alternating-current voltage amplitude Vam,Vbm,VcmMaximum value V of DC voltagedcmaxAnd a minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value IdcminThirteen feature values in total;
the fault type division criterion in the step two is as follows: when Iamax--Ibmax-I and Iamax--Icmax-When | is less than 0.1, classify it as a first type of fault type, when | Iamax--Ibmax-I or Iamax--Icmax-When | is greater than or equal to 0.1, dividing the fault into a second type of fault type, wherein the two types of fault types respectively represent three-phase balanced fault and three-phase unbalanced fault, carrying out fault type division, then carrying out normalization processing on data, and carrying out data training through a support vector machine to obtain fault identification models corresponding to the two types of fault types;
the characteristic value data of the modular multilevel direct-current power transmission system to be identified input by the fault identification model in the third step is the same as that in the first step, namely the amplitude V of the three-phase alternating-current voltageam,Vbm,VcmMaximum value V of DC voltagedcmaxAnd a minimum value VdcminPositive peak and negative peak of three-phase AC current Iamax+、Ibmax+、Icmax+、Iamax-、Ibmax-、Icmax-And maximum value of DC current IdcmaxAnd a minimum value IdcminThirteen feature values in total;
the fault type identification criterion in the third step is the same as that in the second step, namely, the I is the current Iamax--Ibmax-I and Iamax--Icmax-When | is less than 0.1, classify it as a first type of fault type, when | Iamax--Ibmax-I or Iamax--Icmax-And when the | is more than or equal to 0.1, classifying the | into a second type of fault type, selecting a corresponding fault identification model according to the fault type, and identifying the fault.
2. The method for fault identification of a modular multilevel current-flat transmission system based on a support vector machine according to claim 1, characterized in that: judging the time of the fault in the second step and the third step, specifically, judging the time of the fault by the DC voltage VdcAnd performing sliding window iteration to serve as a starting criterion of fault detection, namely using the latest real-time sampling data to participate in detection and simultaneously removing the earliest sampling data so as to update the sampling data more quickly.
3. The method for fault identification of a modular multilevel current-flat transmission system based on a support vector machine according to claim 2, characterized in that: by pairsDC voltage VdcThe starting criterion for performing sliding window iteration as fault detection is specifically as follows:
(1) setting the width of a sliding window according to the frequency of data sampling, wherein the data sampling is to acquire one data every 0.1ms, the sampling frequency is 10kHz, so that the initial width of the sliding window is 5ms, the frequency is 200Hz, the sliding window can transversely cover 50 discrete waveform points, and the discrete signal point subinterval covered by the initial sliding window is
S=[S1,S2,…,S50]
(2) Selecting a first point S in a sliding window1As the initial reference point, sequentially subtracting the initial reference point from the value of the rest point in the sliding window, if the initial reference point is satisfied
Can be regarded as SNThe point corresponds to the time when the fault occurs, and characteristic values within 10ms after the fault occurs are collected;
if the waveform point in the initial sliding window does not satisfy the above condition, i.e.
The sliding window is moved to the right and a new waveform point S is read51And eliminating the earliest waveform point S1Forming new sub-intervals
S=[S2,S3,…,S51]
(3) And continuing to perform the judgment until a fault occurrence point is found.
4. The method for fault identification of a modular multilevel current-flat transmission system based on a support vector machine according to claim 1, characterized in that: and step two and step three, the bad values are removed by using the Grubbs criterion.
5. The method for fault identification of a modular multilevel current-flat transmission system based on a support vector machine according to claim 4, characterized in that: and (3) removing bad values by using a Grubbs criterion, wherein the specific process is as follows:
(1) taking mean square error of measurement result of a certain characteristic valueAccording to Bessel's formula, there are
Wherein n represents the number of measurements, ViRepresenting the difference between the measured value and the average value, i.e. the residual error;
(2) periodic system errors are judged by using Alps-Hermett criterion, which is comparisonAndsize of (1), ifThe measured value has no periodic system error, and the influence is not generated at the moment;
(3) removing bad values from the measured values, and obtaining the mean square error according to Bessel formulaUsing the concept of confidence interval to eliminate bad value, under a certain confidence level, inquiring the Grabbs coefficient table to find out the coefficient G, if the measured value x isiSatisfy the requirement ofOrThe measured value xiIf the value is bad, the value is removed.
CN201811387279.8A 2018-11-21 2018-11-21 SVM-based fault identification method for modular multi-level direct current power transmission system Active CN109541344B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811387279.8A CN109541344B (en) 2018-11-21 2018-11-21 SVM-based fault identification method for modular multi-level direct current power transmission system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811387279.8A CN109541344B (en) 2018-11-21 2018-11-21 SVM-based fault identification method for modular multi-level direct current power transmission system

Publications (2)

Publication Number Publication Date
CN109541344A CN109541344A (en) 2019-03-29
CN109541344B true CN109541344B (en) 2020-10-27

Family

ID=65848925

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811387279.8A Active CN109541344B (en) 2018-11-21 2018-11-21 SVM-based fault identification method for modular multi-level direct current power transmission system

Country Status (1)

Country Link
CN (1) CN109541344B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110333427B (en) * 2019-08-16 2020-04-28 西南交通大学 IGBT open-circuit fault diagnosis method for sending end converter of flexible direct-current transmission system

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102005755A (en) * 2010-10-29 2011-04-06 昆明理工大学 Extra-high voltage direct current transmission line boundary element forming method based on support vector machine
WO2014124660A1 (en) * 2013-02-12 2014-08-21 Abb Technology Ltd Method and device for detection of a fault in a protected unit
CN104865499A (en) * 2015-05-11 2015-08-26 昆明理工大学 Super-high voltage direct-current power transmission line region internal and external fault identification method
CN106093678A (en) * 2016-06-15 2016-11-09 国网福建省电力有限公司 A kind of method quick and precisely diagnosing flexible direct current power transmission system converter fault

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102005755A (en) * 2010-10-29 2011-04-06 昆明理工大学 Extra-high voltage direct current transmission line boundary element forming method based on support vector machine
WO2014124660A1 (en) * 2013-02-12 2014-08-21 Abb Technology Ltd Method and device for detection of a fault in a protected unit
CN104865499A (en) * 2015-05-11 2015-08-26 昆明理工大学 Super-high voltage direct-current power transmission line region internal and external fault identification method
CN106093678A (en) * 2016-06-15 2016-11-09 国网福建省电力有限公司 A kind of method quick and precisely diagnosing flexible direct current power transmission system converter fault

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
Fault Diagnosis for HVDC Converter Based on Support Vector Machine;Chen TangXian等;《第27届中国控制与决策会议论文集(下册)》;20151130;全文 *

Also Published As

Publication number Publication date
CN109541344A (en) 2019-03-29

Similar Documents

Publication Publication Date Title
Johnson et al. Complete protection scheme for fault detection, classification and location estimation in HVDC transmission lines using support vector machines
Biswal et al. Supervisory protection and automated event diagnosis using PMU data
CN101968525B (en) Fault positioning method for power distribution network by combining simulation calculation and real-time monitoring
Sidhu et al. Design, implementation and testing of an artificial neural network based fault direction discriminator for protecting transmission lines
Song et al. Artificial neural-network-based protection scheme for controllable series-compensated EHV transmission lines
Styvaktakis et al. Expert system for classification and analysis of power system events
CN102854437B (en) Fault line selection method of low current grounding system using time-frequency atom decomposition theory
CN103576053B (en) A kind of voltage sag source localization method based on limited electric energy quality monitoring point
RU2632989C2 (en) Method and device for determining location of single-phase-to-ground fault in distributing network based on wavelet transformation of transitional signals
Mora-Florez et al. Fault location in power distribution systems using a learning algorithm for multivariable data analysis
Korkali et al. Optimal deployment of wide-area synchronized measurements for fault-location observability
Janik et al. Automated classification of power-quality disturbances using SVM and RBF networks
Dalstein et al. Neural network approach to fault classification for high speed protective relaying
Mohamed et al. A neural network-based scheme for fault diagnosis of power transformers
CN102508076B (en) Fault diagnosis device and method based on multi-agent system and wavelet analysis
Shaik et al. Power quality assessment and event detection in hybrid power system
Singh et al. Transmission line fault detection and classification
CN104820157B (en) A kind of flexible direct current power transmission system direct current monopolar grounding fault judgment method
CN104865499B (en) A kind of extra high voltage direct current transmission line internal fault external fault recognition methods
CN109119977B (en) Multiterminal flexible direct current power grid DC line fast protection method and system based on single ended voltage
CN100335912C (en) Small Current earthing line selecting method based on model parametric recognition
CN102969888B (en) Design method for multi media card (MMC) self-defining submodule based on real time digital system (RTDS)
Khoshkhoo et al. On-line dynamic voltage instability prediction based on decision tree supported by a wide-area measurement system
Saravanababu et al. Transmission line faults detection, classification, and location using discrete wavelet transform
Casagrande et al. Data mining approach to fault detection for isolated inverter-based microgrids

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