CN113608146B - Fault line selection method suitable for forest fire under high-resistance grounding condition - Google Patents
Fault line selection method suitable for forest fire under high-resistance grounding condition Download PDFInfo
- Publication number
- CN113608146B CN113608146B CN202110902338.6A CN202110902338A CN113608146B CN 113608146 B CN113608146 B CN 113608146B CN 202110902338 A CN202110902338 A CN 202110902338A CN 113608146 B CN113608146 B CN 113608146B
- Authority
- CN
- China
- Prior art keywords
- fault
- zero sequence
- sequence current
- mode
- branch
- 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
- 238000010187 selection method Methods 0.000 title claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 34
- 238000001914 filtration Methods 0.000 claims abstract description 16
- 238000012163 sequencing technique Methods 0.000 claims abstract description 3
- 238000012549 training Methods 0.000 claims description 15
- 230000011218 segmentation Effects 0.000 claims description 10
- 238000000638 solvent extraction Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000000034 method Methods 0.000 abstract description 14
- 238000004088 simulation Methods 0.000 description 5
- 239000012634 fragment Substances 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000001052 transient effect Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000010355 oscillation Effects 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/52—Testing for short-circuits, leakage current or ground faults
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/085—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution lines, e.g. overhead
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/50—Testing of electric apparatus, lines, cables or components for short-circuits, continuity, leakage current or incorrect line connections
- G01R31/58—Testing of lines, cables or conductors
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
Abstract
The invention discloses a high-resistance ground fault line selection mode under forest fire, which comprises the following steps: step1: filtering the acquired zero sequence current of each branch by adopting a wavelet filtering mode to filter the waveform of the zero sequence current of the branch, filtering noise interference, and obtaining the denoised zero sequence currentThe method comprises the steps that a sliding window is adopted on the basis of a shape algorithm for the zero sequence current after denoising of all branches to obtain all possible subsequence sets on a zero sequence current time sequence; for each candidate subsequence C in candidate set candidate, calculating its and each path zero sequence currentDistance of (2)And sequencing; setting different threshold d th The Information Gain of each division mode is calculated, the division mode with the highest Information Gain is used as a fault line selection judging mode, the classification with the highest Information Gain is determined as a final judging mode of fault classification, the class A in the classification with the highest Information Gain is defined as a fault class, and the branch in the class A is defined as a fault branch.
Description
Technical Field
The invention relates to the field of power system fault discrimination, in particular to a fault line selection method suitable for a forest fire under the condition of high resistance grounding.
Background
The fault line is cut off as soon as possible, and the realization of reliable and safe continuous power supply is always an important issue of the power distribution network. The prior grounding line selection device is almost incapable of coping with high-resistance grounding fault working conditions such as forest fires. With the construction of the intelligent power grid, the line erection range and the density are further increased, and the difficulty of fault line selection in forest fires is further improved.
The prior art algorithm for fault line selection extracts features from the zero sequence current for distinguishing between faulty and non-faulty lines, and to make this distinction more obvious, frequency domain feature extraction is applied to fault line selection. The difficulty in line selection under forest fires is that the transition resistance is high, fault characteristics are not obvious, and the noise interference of site sampling is serious. The fault line selection under the forest fire becomes a difficult problem in the power distribution network. However, the long-time fault detection caused by the long-time fault detection and the economic loss and other injuries are unacceptable, so that the development of a new efficient line selection algorithm according to the situation is of great significance.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a fault line selection method suitable for the condition of high-resistance grounding of forest fires, realizes fault line selection aiming at the fault working condition of high-resistance grounding, and has important significance for safe operation of a power distribution network.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
a fault line selection method suitable for a forest fire under the condition of high resistance grounding comprises the following steps:
step one: training a shapelet model algorithm by using zero sequence current history data of a known fault line, and training super-parameters of the shapelet model algorithm until the super-parameters are determined;
step two: the trained shapelet model algorithm is applied to fault line selection, zero-sequence current signals of all the branches are input in real time, the shapelet model algorithm automatically selects optimal candidate subsequences according to super parameters, after distance sorting between the subsequences and the zero-sequence currents of the branches is calculated, all the branches are classified according to trained segmentation threshold values and segmentation modes, and the branches corresponding to the faults are obtained to be the fault line.
The first step comprises the following steps:
step1: filtering the collected zero sequence current historical data of each branch of the known fault line by adopting a wavelet filtering mode to filter the zero sequence current waveform of the branch, filtering noise interference, and obtaining the denoised zero sequence current
Step 2: carrying out normalization treatment on the zero sequence current after denoising of all the branches; obtaining all possible subsequence sets on the zero sequence current time sequence by adopting a sliding window: candidate set candidate;
step3: for each candidate subsequence C in candidate set candidate, calculating its and each path zero sequence currentDistance of->And sequencing;
step4: setting a threshold d th According to the distance threshold, a division mode sp is adopted to divide the branch into two types A and B, and the two types A and B refer to a fault type and a non-fault type respectively.
K is the serial number of each branch
Step5: setting different threshold d th Calculating Information Gain of each division mode, taking the division mode with the highest Information Gain as a fault line selection judging mode, determining the classification with the highest Information Gain as a final judging mode of fault classification, and defining class A in the classification with the highest Information Gain as a fault class, wherein branches in the class A are fault branches;
the information gain is calculated as follows:
zero sequence current set
entropy(I)＝-p(A)log(p(A))-p(B)log(p(B))
A and B are fault class and non-fault class respectively, p (A) is probability distribution, m is total branch number, sp refers to threshold d th The partitioning scheme, entopy (I), is the entropy in the sp partitioning scheme,entropy when not divided;
and training the shape model algorithm by taking the classification threshold value and the classification mode which are obtained by the zero sequence current data of the fault line as the known number so as to determine the super-parameters in the algorithm model, so that the algorithm training is completed.
In step1, the wavelet filtering mode processing method includes: will contain zero sequence current i of noise 0 Converted into a wavelet domain signal s (k) and subjected to wavelet transformationThe wavelet coefficient d of each layer is obtained by conversion j，k Performing threshold processing to obtain processed wavelet coefficientsReconstructing the signal by using the processed wavelet coefficient to obtain the denoised zero sequence current +.>j, k represent the number of layers and the number of levels in the wavelet transform, respectively
Calculating candidate segment C and zero sequence current by using Euclidean distanceDistance between->
The invention has the advantages that: the zero sequence current waveform is extracted by utilizing a wavelet denoising method, so that the zero sequence current waveform is easier to process and is beneficial to line selection; zero-sequence current is processed in the time domain, the fault branch and the normal branch are classified by calculating an information gain mode after training in advance by adopting a shape model algorithm, the correct classification mode of fault classification can be effectively determined according to the magnitude of the information gain, further, a fault line is effectively obtained according to a classification result, and high-resistance grounding line selection is effectively and accurately realized.
Drawings
The contents of the drawings and the marks in the drawings of the present specification are briefly described as follows:
FIG. 1 is a block diagram of a power distribution network system of the present invention;
FIG. 2 is a zero sequence current transient oscillation waveform without denoising;
FIG. 3 is a waveform of transient oscillations of zero sequence current after wavelet denoising;
FIG. 4 is a zero sequence current waveform after adding white Gaussian noise in a simulation example;
fig. 5 is a waveform obtained by denoising zero-sequence current in a simulation example.
Detailed Description
The following detailed description of the invention refers to the accompanying drawings, which illustrate preferred embodiments of the invention in further detail.
The invention relates to a method for selecting a line of a power distribution network ground fault under the condition of natural disasters such as forest fires and the like, and belongs to the field of power system fault discrimination. The existing power distribution network ground fault line selection device has no better solution to high-resistance ground, and the fault characteristics are not obvious when the transition resistance is high, noise interference is serious, and information is covered in noise. After noise is filtered by adopting wavelet, the shape algorithm is applied to fault line selection, and local features in zero sequence current are automatically extracted in a time domain. And calculating the information gain to obtain an optimal dividing mode, and dividing the branch into a fault class and a non-fault class for line selection. The invention provides an effective method for selecting lines aiming at high-resistance ground faults, and improves the power supply reliability of a power distribution network.
The invention relates to a method for selecting a line of a ground fault under the condition of forest fire, which aims at realizing the line selection of the fault under the high-resistance ground fault working condition and has important significance for the safe operation of a power distribution network.
The technical scheme of the invention is as follows:
step one: training a shapelet model algorithm by using zero sequence current history data of a known fault line, and training super-parameters of the shapelet model algorithm until the super-parameters are determined;
step two: the trained shapelet model algorithm is applied to fault line selection, zero-sequence current signals of all the branches are input in real time, the shapelet model algorithm automatically selects optimal candidate subsequences according to super parameters, after distance sorting between the subsequences and the zero-sequence currents of the branches is calculated, all the branches are classified according to trained segmentation threshold values and segmentation modes, and the branches corresponding to the faults are obtained to be the fault line.
The first step is mainly to train the shapelet model algorithm, the hyper-parameters and the like of the shapelet model algorithm after training are confirmed, when faults occur, fault classification can be automatically output, and a branch in the fault classification is a fault branch, and the specific training steps comprise:
step1: filtering the collected zero sequence current historical data of each branch of the known fault line by adopting a wavelet filtering mode to filter the zero sequence current waveform of the branch, filtering noise interference, and obtaining the denoised zero sequence current
And filtering the zero sequence current waveform of the branch by adopting a wavelet filtering mode to filter noise interference. Will contain zero sequence current i of noise 0 Converted into a wavelet domain signal s (k) and subjected to wavelet transformation to obtain wavelet coefficients d of each layer j，k Performing threshold processing to obtain processed wavelet coefficientsReconstructing the signal by using the processed wavelet coefficient to obtain the denoised zero sequence current +.>
Setp2: and carrying out normalization processing on the zero sequence current after denoising of all the branches. The kernel of the Shapelet algorithm is to find a local feature on the time sequence that classifies the set, which appears as a sub-sequence on the time sequence. We will therefore refer to the acquisition of all possible sub-sequence sets over the zero sequence current time sequence using a sliding window as candidate set candidate. The zero sequence current we obtain is a time series with fixed sampling frequency and fixed sampling practice, denoted herein asWherein->Representing the time sequence of the sampled zero sequence current on the kth branch, and +.>Then the sample at time 1 represents zero current in the k branchThe data, here the zero sequence current of each branch is sampled with N time values, respectively.
Shapelet aims at finding a segment of a sequence that differs significantly to distinguish between different sequences. To find this segment that can effectively distinguish between different sequences, we need to choose some candidates over the sequence. Obtaining a candidate fragment C in sequence by using a window with the size of w 1 ＝[i 0，1 ，i 0，2 ，i 0，3 ，...，i 0，w ]Moving the window to obtain the next candidate segment C 2 ＝[i 0，2 ，i 0，3 ，i 0，4 ，...，i 0，w+1 ]The window is continuously moved, and in practice, each movement step may be set for efficiency. And sliding and taking N-w+1 candidate fragments from all the branches to obtain a candidate set.
Step3: for each candidate subsequence C in candidate, calculating its zero sequence current with each branchDistance of->And ordering. The Euclidean distance is generally used for calculating candidate segment C and zero sequence current +.>The distance between the two is calculated in a point-by-point mode, and the square of the difference is added and then square is obtained. And in the calculation, a subsequence with the length of w is taken from the zero sequence current of each branch, the distance between the subsequence and the candidate segment C is calculated, and finally, the minimum distance value is taken.
Step4: setting a threshold d based on the sorting of the step3 th According to the distance threshold, a division mode sp is adopted to divide the branch into two types A and B, and the two types A and B refer to a fault type and a non-fault type respectively.
K is the serial number of each branch
Step5: and (3) calculating the Information Gain of each division mode (each division mode corresponds to a threshold value, different division modes are realized by different thresholds) on the basis of the step (4), and taking the division mode with the highest Information Gain as the judging mode of fault line selection. A and B are fault class and non-fault class respectively, p (A) is probability distribution, m is total branch number, sp refers to division mode, entropy (I) is entropy under sp division mode,is the entropy of the undivided material. The information gain is calculated as follows:
zero sequence current set
entropy(I)＝-p(A)log(p(A))-p(B)log(p(B))
A and B are fault class and non-fault class respectively, p (A) is probability distribution, m is total branch number, sp refers to threshold d th The partitioning scheme, entopy (I), is the entropy in the sp partitioning scheme,is the entropy of the undivided material.
The purpose of the shape is to find a segment that has a large difference in zero sequence current of the faulty line and the non-faulty line, thereby helping us distinguish between the faulty line and the non-faulty lineA faulty line. In step 2 we have chosen a series of candidate fragments, where it is desirable to find the fragment with the best classification. The classification effect is measured by the information gain. Then there are m distance values from the zero sequence current for each candidate segment. After ordering the distance values, the failed class and the non-failed class should ideally be completely separated near a certain segmentation point, but in reality they may be staggered. In step4 we use a threshold d th The m distance values are divided into a fault class and a non-fault class. P (A) and P (B) are adopted to respectively represent the probability that the fault class and the non-fault class are correctly divided in an sp division mode, so that the division mode that each candidate segment can obtain the maximum information gain is calculated, and finally the candidate segment with the maximum information gain, namely the best classification effect, is selected. From this point on, the shapelet has been trained and we have found candidate segments and best segmentation points that can be used to distinguish between faulty and non-faulty categories.
And training the shape model algorithm by taking the classification threshold value and the classification mode which are obtained by the zero sequence current data of the fault line as the known number so as to determine the super-parameters in the algorithm model, so that the algorithm training is completed. The trained shape model algorithm determines the mode that the information segmentation mode meets the super parameters through the determined super parameters, calculates the information gain, and finally outputs two categories of fault classification and non-fault classification and branches corresponding to the categories, and the method for dividing the zero sequence current set of the unknown fault line into fault types and non-fault types is obtained.
The beneficial effects of the invention are as follows: the zero sequence current waveform is extracted by utilizing a wavelet denoising method, so that the zero sequence current waveform is easier to process and is beneficial to line selection; the invention processes zero-sequence current in time domain, which is an effective mode of high-resistance grounding line selection
In order to verify the effectiveness of the line selection mode, the method adopts the following mode to carry out simulation verification, firstly, a simulation model of a power distribution network shown in figure 1 is established by using a simulink, a 35kV/10kV substation is provided with 6 return lines and a feeder L 1 、L 5 Is a pure cable L 3 、L 4 Is an overhead line L 2 、L 6 Is a hybrid line. The positive sequence impedance of the overhead feeder is as follows: r is R 1 ＝0.27Ω/km，L 1 ＝0.352Ω/km，C 1 =178 uS/km, zero sequence impedance is: r is R 0 ＝0.42Ω/km，L 0 ＝0.304Ω/km，C 0 =110 uS/km; the positive sequence impedance of the cable is: r is R 1 ＝0.157Ω/km，L 1 ＝0.076Ω/km，C 1 =132 uS/km, zero sequence impedance is: r is R 0 ＝0.307Ω/km，L 0 ＝0.304Ω/km，C 0 =110 uS/km; the neutral point of the power distribution system is led out from a grounding transformer of a bus, is grounded through an arc suppression coil, the compensation mode of the arc suppression coil is overcompensation, and the sampling frequency of the system is 1MHz.
(1) Feed line L of resonant grounded distribution network shown in figure 1 through electromagnetic transient simulation 1 Setting single-phase earth fault as fault feeder line, setting fault initial phase angle as 90 deg, simulating time as 0.2s, fault occurring in 0.05s, transitional resistance as 10k omega, adding Gaussian white noise to each branch line, extracting zero sequence current of each feeder line as shown in figure 4;
(2) Noise reduction is performed by using a wavelet denoising mode, and zero sequence current is extracted as shown in fig. 5.
(3) Performing fault line selection by using trained shapelet algorithm to obtain a fault line L 1 . The trained shapelet algorithm refers to the shapelet algorithm that has selected the optimal segment and segmentation point based on the existing fault data.
The algorithm is utilized to perform fault selection, the software logic is edited to process data and then simulate to obtain the line L1 in the class A in fault classification, the line L1 is matched with the fault line L1 set by the method, and the line selection is successful.
It is obvious that the specific implementation of the present invention is not limited by the above-mentioned modes, and that it is within the scope of protection of the present invention only to adopt various insubstantial modifications made by the method conception and technical scheme of the present invention.
Claims (3)
1. A fault line selection method suitable for a forest fire under the condition of high resistance grounding is characterized by comprising the following steps: comprising the following steps:
step one: training a shapelet model algorithm by using zero sequence current history data of a known fault line, and training super-parameters of the shapelet model algorithm until the super-parameters are determined;
step two: applying the trained shapelet model algorithm to fault line selection, inputting zero sequence current signals of all branches in real time, automatically selecting optimal candidate subsequences according to super parameters by the shapelet model algorithm, sorting all branches according to trained segmentation threshold values and segmentation modes after calculating distance sequences of the subsequences and the zero sequence currents of the branches, and obtaining branches corresponding to fault types as fault lines;
the first step comprises the following steps:
step1: filtering the collected zero sequence current historical data of each branch of the known fault line by adopting a wavelet filtering mode to filter the zero sequence current waveform of the branch, filtering noise interference, and obtaining the denoised zero sequence current
Step 2: carrying out normalization treatment on the zero sequence current after denoising of all the branches; obtaining all possible subsequence sets on the zero sequence current time sequence by adopting a sliding window: candidate set candidate;
step3: for each candidate subsequence C in candidate set candidate, calculating its and each path zero sequence currentDistance of->And sequencing;
step4: setting a threshold d th Dividing the branch into two classes A and B according to the distance threshold by adopting a dividing mode sp, wherein the classes A and B respectively refer to a fault class and a non-fault class;
k is the serial number of each branch
Step5: setting different threshold d th Calculating Information Gain of each division mode, taking the division mode with the highest Information Gain as a fault line selection judging mode, determining the classification with the highest Information Gain as a final judging mode of fault classification, and defining class A in the classification with the highest Information Gain as a fault class, wherein branches in the class A are fault branches;
the information gain is calculated as follows:
zero sequence current setk＝1,…,m
entropy(I)＝-p(A)log(p(A))-p(B)log(p(B))
A and B are fault class and non-fault class respectively, p (A) is probability distribution, m is total branch number, sp refers to threshold d th The partitioning scheme, entopy (I), is the entropy in the sp partitioning scheme,entropy when not divided;
and training the shape model algorithm by taking the classification threshold value and the classification mode which are obtained by the zero sequence current data of the fault line as the known number so as to determine the super-parameters in the algorithm model, so that the algorithm training is completed.
2. The fault line selection method suitable for the condition of high resistance grounding of forest fires as claimed in claim 1, wherein: in step1, the smallThe wave filtering mode processing method comprises the following steps: will contain zero sequence current i of noise 0 Converted into a wavelet domain signal s (k) and subjected to wavelet transformation to obtain wavelet coefficients d of each layer j,k Performing threshold processing to obtain processed wavelet coefficientsReconstructing the signal by using the processed wavelet coefficient to obtain the denoised zero sequence current +.>j, k represent the number of layers and the number of stages in the wavelet transform, respectively.
3. The fault line selection method suitable for the condition of high resistance grounding of forest fires as claimed in claim 1, wherein: calculating candidate segment C and zero sequence current by using Euclidean distanceDistance between->
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110902338.6A CN113608146B (en) | 2021-08-06 | 2021-08-06 | Fault line selection method suitable for forest fire under high-resistance grounding condition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110902338.6A CN113608146B (en) | 2021-08-06 | 2021-08-06 | Fault line selection method suitable for forest fire under high-resistance grounding condition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113608146A CN113608146A (en) | 2021-11-05 |
| CN113608146B true CN113608146B (en) | 2023-12-19 |
Family
ID=78307488
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110902338.6A Active CN113608146B (en) | 2021-08-06 | 2021-08-06 | Fault line selection method suitable for forest fire under high-resistance grounding condition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113608146B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104020845A (en) * | 2014-03-27 | 2014-09-03 | 浙江大学 | Acceleration transducer placement-unrelated movement recognition method based on shapelet characteristic |
| DE102014105994A1 (en) * | 2013-04-29 | 2014-10-30 | Technische Universität Dresden | Method for detecting signal patterns in spectroscopically generated data sets |
| CN109161931A (en) * | 2018-10-22 | 2019-01-08 | 中南大学 | Aluminium electrolysis anode electric current classification method based on shapelet conversion |
| CN110097134A (en) * | 2019-05-08 | 2019-08-06 | 合肥工业大学 | A kind of mechanical breakdown method of early diagnosis based on time series |
| CN111814897A (en) * | 2020-07-20 | 2020-10-23 | 辽宁大学 | Time series data classification method based on multi-level shape |
| CN113162037A (en) * | 2021-04-23 | 2021-07-23 | 清华大学 | Power system transient voltage stability self-adaptive evaluation method and system |
-
2021
- 2021-08-06 CN CN202110902338.6A patent/CN113608146B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE102014105994A1 (en) * | 2013-04-29 | 2014-10-30 | Technische Universität Dresden | Method for detecting signal patterns in spectroscopically generated data sets |
| CN104020845A (en) * | 2014-03-27 | 2014-09-03 | 浙江大学 | Acceleration transducer placement-unrelated movement recognition method based on shapelet characteristic |
| CN109161931A (en) * | 2018-10-22 | 2019-01-08 | 中南大学 | Aluminium electrolysis anode electric current classification method based on shapelet conversion |
| CN110097134A (en) * | 2019-05-08 | 2019-08-06 | 合肥工业大学 | A kind of mechanical breakdown method of early diagnosis based on time series |
| CN111814897A (en) * | 2020-07-20 | 2020-10-23 | 辽宁大学 | Time series data classification method based on multi-level shape |
| CN113162037A (en) * | 2021-04-23 | 2021-07-23 | 清华大学 | Power system transient voltage stability self-adaptive evaluation method and system |
Non-Patent Citations (1)
| Title |
|---|
| 基于改进 shapelet 挖掘的风电并网系统 暂态功角稳定评估;梁咪咪 等;《能源工程》;第37-43页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113608146A (en) | 2021-11-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109307824B (en) | Clustering-based power distribution network single-phase earth fault section positioning method | |
| CN107329046B (en) | Direct current overhead line lightning stroke identification method based on modulus analysis | |
| CN111650470B (en) | Rapid self-adaptive fault detection and identification method for micro-grid line section | |
| CN113466634B (en) | Ground fault waveform identification method based on fault indicator | |
| CN104101817A (en) | PSO (Particle Swarm Optimization) improved atomic decomposition method based lightning interference and fault identification method | |
| CN110244122A (en) | Resonant Overvoltage in Power Systems Detection & Controling method | |
| CN113985194A (en) | Power distribution network fault positioning method based on stack self-encoder | |
| CN106841915A (en) | A kind of power transmission line fault locating method based on compressed sensing | |
| CN112462193A (en) | Power distribution network automatic reclosing judgment method based on real-time fault filtering data | |
| CN109541395B (en) | Distribution network section positioning method based on Hausdorff under characteristic frequency band | |
| CN113608146B (en) | Fault line selection method suitable for forest fire under high-resistance grounding condition | |
| CN110988606B (en) | Sharpening processing-based small current ground fault line selection method | |
| CN111896890B (en) | Micro-grid line fault diagnosis method and system based on Hilbert-Huang transform | |
| CN113534006B (en) | Single-phase earth fault line selection method based on CEEMD and autocorrelation threshold denoising | |
| CN114755529A (en) | Multi-feature fusion single-phase earth fault type identification method based on deep learning | |
| CN113759278B (en) | Ground fault line selection method suitable for small-current grounding system | |
| Chen et al. | New algorithm for EHV transmission line faulted-phase selection based on wavelet transforms and artificial intelligence | |
| CN115047296B (en) | Power distribution network fault section positioning method | |
| CN115575783B (en) | Multiple Lei Raoji and counterattack recognition method and system for alternating-current power transmission line | |
| Yang et al. | Detection of high impedance faults in distribution system | |
| CN113989201B (en) | Axle center track identification method based on cloud computing and LSTM | |
| CN114062880B (en) | Direct-current fault arc detection method based on adaptive signal processing and CART tree ensemble learning | |
| CN113189446B (en) | Low-current ground fault line selection method | |
| CN117741505A (en) | Power distribution network successive fault detection method and system based on kurtosis and Teager energy operators | |
| Zhou et al. | Single-ended fault location and early warning method of transmission line based on back propagation neural network |
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 |