CN117436025A - Fault indicator-based non-fault abnormal waveform screening method - Google Patents
Fault indicator-based non-fault abnormal waveform screening method Download PDFInfo
- Publication number
- CN117436025A CN117436025A CN202311763801.9A CN202311763801A CN117436025A CN 117436025 A CN117436025 A CN 117436025A CN 202311763801 A CN202311763801 A CN 202311763801A CN 117436025 A CN117436025 A CN 117436025A
- Authority
- CN
- China
- Prior art keywords
- waveform
- fault
- current
- abnormal
- data
- 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.)
- Pending
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012216 screening Methods 0.000 title claims abstract description 40
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 20
- 230000007246 mechanism Effects 0.000 claims abstract description 17
- 238000012360 testing method Methods 0.000 claims abstract description 17
- 238000003066 decision tree Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 16
- 230000008569 process Effects 0.000 claims description 9
- 238000013145 classification model Methods 0.000 claims description 7
- 238000001514 detection method Methods 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 4
- 238000003672 processing method Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 claims 1
- 238000012795 verification Methods 0.000 abstract description 7
- 238000000605 extraction Methods 0.000 abstract description 3
- 238000010801 machine learning Methods 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000005684 electric field Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 208000025274 Lightning injury Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005070 sampling 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
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
-
- 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
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- 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
Abstract
The invention relates to the technical field of power distribution network ground faults, and discloses a non-fault abnormal waveform screening method based on a fault indicator, which comprises the following steps: acquiring wave recording data when a ground fault occurs in the power distribution network, and marking to obtain a target data set with a label; performing waveform processing on the original waveform data, and extracting characteristics capable of representing waveform categories by using a convolutional neural network; extracting mechanism latent features capable of reflecting non-fault categories; fusing the features extracted by the convolutional neural network with the mechanism latent features to obtain rich features capable of reflecting waveform types, and further utilizing a decision tree algorithm to realize classification; and evaluating the performance of the model by using the test set to realize the verification of the effectiveness of the method. The invention can screen out non-fault abnormal waveforms in the power distribution network ground fault by utilizing the machine learning extraction characteristics and the capability of processing mass data, thereby improving the positioning performance of subsequent fault sections.
Description
Technical Field
The invention relates to the technical field of power distribution network ground faults, in particular to a non-fault abnormal waveform screening method based on a fault indicator.
Background
With the continuous development of power systems, fault problems in the power distribution network frequently occur. In the power system, the fault of the power equipment, external interference and other factors can cause the fault to occur. In order to guarantee domestic electricity effect, the power grid company puts forward higher requirements on safe and stable operation of the power distribution network. A fault indicator is a device that can help quickly identify and locate faults in a power line. However, during fault recording, a series of non-fault abnormal waveform data exist, and the data can influence the operation of a fault positioning algorithm module, so that fault misjudgment or missed judgment is caused, and negative effects are generated on the stability of a power grid, the safety of equipment and the like. Therefore, analysis and screening of non-faulty abnormal waveform data is particularly important. Currently, there are various methods for screening out non-fault abnormal waveform data in fault data, including manual judgment, making fault data judgment rules according to expert experience, waveform comparison, and the like. However, the existing method cannot obtain higher efficiency under the power big data background, and has the problems of low screening precision and the like, so that a non-fault abnormal waveform screening method needs to be researched, the classification of current mass fault data and non-fault data can be realized, and the problem of low identification accuracy is solved.
Disclosure of Invention
Aiming at the defects and drawbacks existing in the prior art, the invention provides a non-fault abnormal waveform screening method based on a fault indicator, which can screen non-fault abnormal waveforms in the ground fault of a power distribution network, thereby improving the positioning performance of subsequent fault sections.
The aim of the invention can be achieved by the following technical scheme:
a fault indicator-based non-fault abnormal waveform screening method, comprising the steps of:
s1, acquiring and storing original waveform data when a ground fault occurs in the operation process of a power distribution network, and marking the acquired original waveform data to obtain a sample label;
s2, performing waveform processing operation on the original waveform data, and extracting characteristics capable of representing waveform types by using a convolutional neural network;
s3, extracting mechanism latent features capable of reflecting waveform types from the original waveform data;
s4, performing back-end fusion on the features extracted by the convolutional neural network and the mechanism latent features to obtain rich information capable of comprehensively reflecting waveform types, and combining the sample label information obtained in the S1, realizing waveform classification by utilizing a decision tree algorithm, and screening out non-fault abnormal waveforms;
s5, evaluating the performance of the trained model by using data in the test set, and verifying the effectiveness of the method.
Further, in the step S1, the original waveform data when the ground fault occurs includes a cycles before the fault occurs and b cycles after the fault occurs;
further, in the step S2, the waveform processing and feature extracting operation of the raw waveform data includes the steps of:
s21, respectively making different normalization thresholds for the current and the voltage, and respectively performing normalization operation for the voltage and the current;
s22, inserting a zero sequence voltage waveform after each phase voltage waveform, and inserting a zero sequence current waveform after each phase current waveform so as to strengthen the spatial relationship between each phase voltage current and the zero sequence voltage current;
s23, aiming at a+b cycles in each phase of voltage and current waveform, the first cycle is unchanged, and the later a+b-1 cycles are respectively different from the first cycle;
s24, inputting the processed waveform characteristics into the built convolutional neural network, training the convolutional neural network, and extracting the waveform characteristics.
Further, in the step S3, the method for extracting the mechanical latent feature includes the following steps:
s31, respectively extracting the voltage effective value and the current effective value of the first cycle and the last cycle in each phase of waveform as mechanism latent characteristics aiming at three-phase voltage and current, wherein the calculation formula of the effective value of a single cycle is as follows:
wherein,for the calculated cycle effective value,as a starting point of the cycle wave,is the total point number of the frequency waves,is waveform at the firstThe value of each point;
s32, calculating the effective values of a+b cycles according to the above formula aiming at the zero sequence current, and selecting various statistical values of the effective values from the effective values as mechanism characteristics.
Further, in S4, the method for performing feature backend fusion and classification further includes the following steps:
s41, taking the output of a straightening layer in the convolutional neural network as an extracted waveform characteristic, and fusing the extracted waveform characteristic with an extracted mechanism latent characteristic;
s42, inputting all the fused features into a decision tree model, training a decision tree classification model, and taking the decision tree classification model as a final non-fault abnormal waveform screening model.
Further, in S5, the method for evaluating a model includes the following steps:
s51, extracting characteristics of waveforms in the test set according to the same processing method as that of the training set;
s52, inputting the extracted test set features into a trained non-fault abnormal waveform screening model to obtain a test set label;
s53, listing detection results, and respectively counting the number of various detection results;
s54, checking the abnormal waveform screening model performance by using the accuracy, the recall and the F1-score.
The beneficial technical effects of the invention are as follows: according to the invention, the rapid classification of massive fault data and non-fault data can be effectively realized through the non-fault abnormal waveform screening model obtained through training, and the trained model can obtain higher identification accuracy rate while guaranteeing the identification efficiency, so that the screening of the non-fault abnormal waveform in the power distribution network ground fault can be realized, and the positioning performance of the subsequent fault section is improved.
Drawings
FIG. 1 is a general flow chart of a fault indicator-based non-fault abnormal waveform screening method according to an embodiment of the present invention.
Fig. 2 is a convolutional neural network frame diagram constructed by a non-fault abnormal waveform screening method based on a fault indicator according to an embodiment of the invention.
FIG. 3 is a graph of a training process loss function of a training set and a verification set of a non-fault abnormal waveform screening method based on a fault indicator according to an embodiment of the present invention.
Fig. 4 is a graph of accuracy of training set and verification set of a non-fault abnormal waveform screening method based on a fault indicator according to an embodiment of the present invention.
FIG. 5 is a hyper-parameter configuration diagram of a decision tree classification model built by a non-fault abnormal waveform screening method based on a fault indicator according to an embodiment of the invention.
Fig. 6 is a diagram of a classification result of a non-fault abnormal waveform screening method on the basis of a fault indicator according to an embodiment of the present invention.
Fig. 7 is a diagram of four types of detection results counted by a non-fault abnormal waveform screening method based on a fault indicator according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Examples: a fault indicator-based non-fault abnormal waveform screening method as shown in fig. 1, comprising the steps of:
s1, acquiring and storing original waveform data when a ground fault occurs in the operation process of a power distribution network, and marking the acquired original waveform data to obtain a sample label;
in the embodiment, the original waveform data obtained during the ground fault comprises 6 cycles before the fault and 10 cycles after the fault, and the waveforms of the 16 cycles are stored; the marked original waveform data comprises fault waveform data and non-fault abnormal waveform data, and the fault waveform data comprises a fault internal waveform and a fault external waveform; the non-fault abnormal waveform data comprises non-fault lightning stroke, weak zero-sequence current, zero-sequence current non-change abnormality, power failure, phase failure, three-phase electric field bias abnormality, three-phase current sampling abnormality and three-phase electric field non-change abnormal waveform.
S2, performing waveform processing operation on the original waveform data, and extracting characteristics capable of representing waveform types by using a convolutional neural network;
in the embodiment, for the three-phase current and the zero-sequence current, a normalization threshold value of 50 is formulated by combining the maximum value appearing in the waveform, and the current waveform preprocessing mode is as follows:
wherein the method comprises the steps ofIn order to process the pre-processed current waveform,is the processed current waveform.
In the embodiment, for the three-phase voltage and the zero-sequence voltage, a normalized threshold value of 1000 is formulated by combining the maximum value appearing in the waveform, and the preprocessing mode of the voltage waveform is as follows:
wherein the method comprises the steps ofIn order to process the pre-processed voltage waveform,is the processed voltage waveform.
And inserting a zero sequence voltage waveform after each phase voltage waveform, and inserting a zero sequence current waveform after each phase current waveform so as to strengthen the spatial relationship between each phase voltage current and the zero sequence voltage current.
In the embodiment, for 16 cycles in each phase of voltage and current waveform, the first cycle is unchanged, and the last 15 cycles are respectively different from the first cycle.
In the embodiment, the processed waveform characteristics are input into a built convolutional neural network, the built convolutional neural network frame is shown in fig. 2, the built convolutional neural network is trained, a waveform characteristic extraction model is obtained, and waveform characteristics are extracted.
In the embodiment, the loss function curves of the training set and the verification set in the training process are drawn as shown in fig. 3, and the accuracy curves of the training set and the verification set in the training process are drawn as shown in fig. 4; as can be seen from fig. 3 and 4, the adjusted model parameters gradually approach the real results, and the deviation between the training set and the verification set is smaller, which indicates that the super-parameter setting is reasonable.
S3, extracting mechanism latent features capable of reflecting waveform types from the original waveform data;
in the embodiment, aiming at three-phase voltage and current, the effective values of the first cycle and the last cycle in each phase of waveform are respectively extracted, and 6 three-phase total extraction current effective values and 6 voltage effective values are used as mechanism latent features; wherein the effective value of a single cycle is calculated as follows:
in the method, in the process of the invention,as a starting point of the cycle wave,is the total point number of the frequency waves,is waveform at the firstThe value of each point.
In the embodiment, for the zero sequence current, effective values of 16 cycles are calculated respectively, and the maximum value and the minimum value of the effective values are selected as mechanism characteristics.
S4, performing back-end fusion on the features extracted by the convolutional neural network and the mechanism latent features to obtain rich information capable of comprehensively reflecting waveform types, and combining the sample label information obtained in the S1, realizing waveform classification by utilizing a decision tree algorithm, and screening out non-fault abnormal waveforms;
in an embodiment, the output of a straightening layer_1 in the convolutional neural network is used as an extracted waveform characteristic; and (3) fusing the output of the straightening layer in the convolutional neural network with 14 mechanism latent features extracted in total in the S3.
In the embodiment, a decision tree classification model is built, the super-parameter configuration of the model is shown in fig. 5, all the fused features are input into the decision tree, the decision tree classification model is trained, and the model is used as a final non-fault abnormal waveform screening model.
S5, evaluating the performance of the trained model by utilizing data in the test set, so as to realize verification of the effectiveness of the method.
Extracting the characteristics of the waveforms in the test set according to the same processing method as the training set; and inputting the extracted test set characteristics into a trained non-fault abnormal waveform screening model to obtain a test set label.
In the embodiment, according to the real label and the predictive label in the test set, the classification result of the non-fault abnormal waveform in the test set is drawn as shown in fig. 6, and in the test set, the non-fault abnormal waveform screening model misconsiders the number of the non-fault abnormal waveforms as 25, misconsiders the fault waveforms as 24, and the average classes of the rest waveforms are correct.
Four types of detection results are listed in the following table, and four types of detection results are counted respectively, as shown in fig. 7, and the performance of the abnormal waveform screening model is checked by using accuracy, recall and F1-score, wherein accuracy = TP/(tp+fp), recall = TP/(tp+fn), and F1-score = 2 x accuracy x recall/(accuracy+recall);
in the embodiment, the number of TP in the test set is 223, the number of TN is 256, the number of FN is 25, and the number of FP is 24, so that the accuracy of the built non-fault abnormal waveform screening model is 0.903, the recall rate is 0.899, and the F1-score is 0.901.
The above embodiments are illustrative of the specific embodiments of the present invention, and not restrictive, and various changes and modifications may be made by those skilled in the relevant art without departing from the spirit and scope of the invention, so that all such equivalent embodiments are intended to be within the scope of the invention.
Claims (6)
1. The fault indicator-based non-fault abnormal waveform screening method is characterized by comprising the following steps of:
s1, acquiring and storing original waveform data when a ground fault occurs in the operation process of a power distribution network, and marking the acquired original waveform data to obtain a sample label;
s2, performing waveform processing operation on the original waveform data, and extracting characteristics capable of representing waveform types by using a convolutional neural network;
s3, extracting mechanism latent features capable of reflecting waveform types from the original waveform data;
s4, performing back-end fusion on the features extracted by the convolutional neural network and the mechanism latent features to obtain rich information capable of comprehensively reflecting waveform types, and combining the sample label information obtained in the S1, realizing waveform classification by utilizing a decision tree algorithm, and screening out non-fault abnormal waveforms;
s5, evaluating the performance of the trained model by using data in the test set, and verifying the effectiveness of the method.
2. The method for screening out abnormal waveforms without failure based on a failure indicator according to claim 1, wherein in S1, the original waveform data when the ground fault occurs includes a cycles before the failure and b cycles after the failure; the marked data set, and the sample label comprises fault waveform data and non-fault waveform data.
3. The fault indicator-based non-fault abnormal waveform screening method according to claim 1, wherein in S2, the waveform processing and feature extracting operations of the raw waveform data include the steps of:
s21, respectively making different normalization thresholds for the current and the voltage, and respectively performing normalization operation for the voltage and the current;
s22, inserting a zero sequence voltage waveform after each phase voltage waveform, and inserting a zero sequence current waveform after each phase current waveform so as to strengthen the spatial relationship between each phase voltage current and the zero sequence voltage current;
s23, aiming at a+b cycles in each phase of voltage and current waveform, the first cycle is unchanged, and the later a+b-1 cycles are respectively different from the first cycle;
s24, inputting the processed waveform characteristics into the built convolutional neural network, training the convolutional neural network, and extracting the waveform characteristics.
4. The fault indicator-based non-fault abnormal waveform screening method according to claim 1, wherein in S3, the method for extracting the mechanical latent feature comprises the steps of:
s31, respectively extracting the voltage effective value and the current effective value of the first cycle and the last cycle in each phase of waveform as mechanism latent characteristics aiming at three-phase voltage and current, wherein the calculation formula of the effective value of a single cycle is as follows:
;
wherein,for the calculated cycle effective value, +.>For the starting point of the cycle, +.>Is the total point number of the frequency, and is->Is waveform in->The value of each point;
s32, calculating the effective values of a+b cycles according to the above formula aiming at the zero sequence current, and selecting various statistical values of the effective values from the effective values as mechanism characteristics.
5. The method for screening out abnormal waveforms without fault indicator according to claim 1, wherein in S4, the feature back-end is fused, and the method for classifying the feature back-end comprises the following steps:
s41, taking the output of a straightening layer in the convolutional neural network as an extracted waveform characteristic, and fusing the extracted waveform characteristic with an extracted mechanism latent characteristic;
s42, inputting all the fused features into a decision tree model, training a decision tree classification model, and taking the decision tree classification model as a final non-fault abnormal waveform screening model.
6. The fault indicator-based non-fault abnormal waveform screening method according to claim 1, wherein in S5 the model evaluation method comprises the steps of:
s51, extracting characteristics of waveforms in the test set according to the same processing method as that of the training set;
s52, inputting the extracted test set features into a trained non-fault abnormal waveform screening model to obtain a test set label;
s53, listing detection results, and respectively counting the number of various detection results;
s54, checking the abnormal waveform screening model performance by using the accuracy, the recall and the F1-score.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311763801.9A CN117436025A (en) | 2023-12-21 | 2023-12-21 | Fault indicator-based non-fault abnormal waveform screening method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311763801.9A CN117436025A (en) | 2023-12-21 | 2023-12-21 | Fault indicator-based non-fault abnormal waveform screening method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117436025A true CN117436025A (en) | 2024-01-23 |
Family
ID=89546492
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311763801.9A Pending CN117436025A (en) | 2023-12-21 | 2023-12-21 | Fault indicator-based non-fault abnormal waveform screening method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117436025A (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111157894A (en) * | 2020-01-14 | 2020-05-15 | 许昌中科森尼瑞技术有限公司 | Motor fault diagnosis method, device and medium based on convolutional neural network |
CN111208387A (en) * | 2020-02-17 | 2020-05-29 | 国电南瑞南京控制系统有限公司 | Distribution network single-phase grounding line selection method based on synchronous phase current fault component comparison |
CN112240965A (en) * | 2019-07-17 | 2021-01-19 | 南京南瑞继保工程技术有限公司 | Grounding line selection device and method based on deep learning algorithm |
CN113159077A (en) * | 2021-05-24 | 2021-07-23 | 南京工程学院 | Three-phase inverter fault identification method based on hybrid convolutional neural network |
-
2023
- 2023-12-21 CN CN202311763801.9A patent/CN117436025A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112240965A (en) * | 2019-07-17 | 2021-01-19 | 南京南瑞继保工程技术有限公司 | Grounding line selection device and method based on deep learning algorithm |
CN111157894A (en) * | 2020-01-14 | 2020-05-15 | 许昌中科森尼瑞技术有限公司 | Motor fault diagnosis method, device and medium based on convolutional neural network |
CN111208387A (en) * | 2020-02-17 | 2020-05-29 | 国电南瑞南京控制系统有限公司 | Distribution network single-phase grounding line selection method based on synchronous phase current fault component comparison |
CN113159077A (en) * | 2021-05-24 | 2021-07-23 | 南京工程学院 | Three-phase inverter fault identification method based on hybrid convolutional neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103745119A (en) | Oil-immersed transformer fault diagnosis method based on fault probability distribution model | |
CN108761287B (en) | Transformer partial discharge type identification method | |
CN110108992B (en) | Cable partial discharge fault identification method and system based on improved random forest algorithm | |
CN110277833B (en) | Transformer substation pressure plate state identification method and terminal equipment | |
CN111160241B (en) | Power distribution network fault classification method, system and medium based on deep learning | |
CN107992665B (en) | Online fault diagnosis and analysis method for alternating current filter of extra-high voltage converter station | |
CN116559598B (en) | Smart distribution network fault positioning method and system | |
CN112464995A (en) | Power grid distribution transformer fault diagnosis method and system based on decision tree algorithm | |
CN116388402B (en) | Abnormality alarm analysis method applied to power transformation equipment | |
CN110794254B (en) | Power distribution network fault prediction method and system based on reinforcement learning | |
CN116401532B (en) | Method and system for recognizing frequency instability of power system after disturbance | |
CN116739829A (en) | Big data-based power data analysis method, system and medium | |
CN110705693A (en) | Unmanned aerial vehicle abnormal behavior recognition module and recognition method thereof | |
CN113313403B (en) | Power distribution network comprehensive evaluation method, device and system based on large-scale high-power electric vehicle charging and discharging and storage medium | |
CN116520068B (en) | Diagnostic method, device, equipment and storage medium for electric power data | |
CN117436025A (en) | Fault indicator-based non-fault abnormal waveform screening method | |
CN112036718A (en) | Electric power system safety risk assessment method considering new energy uncertainty | |
CN116699400A (en) | Generator rotor short-circuit fault monitoring system, method and readable storage medium | |
Tang et al. | Classification for transient overvoltages in offshore wind farms based on sparse decomposition | |
CN114118166A (en) | Transformer fault diagnosis system and method based on intelligent frequency spectrum monitoring | |
Du et al. | Fault diagnosis method of automation equipment in independent and controllable substation based on deep reinforcement learning | |
CN117743815B (en) | Method and system for optimizing signals of multi-path vibration sensor of circuit breaker | |
CN112348074B (en) | Power distribution network power failure event accurate diagnosis method, device and system based on data driving | |
Feng et al. | Infrared Image Recognition and Classification of Typical Electrical Equipment in Substation Based on YOLOv5 | |
CN110070179B (en) | Method and device for identifying partial discharge of high-voltage cable |
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 |