CN115856611A - High-voltage circuit breaker fault diagnosis method based on deep learning - Google Patents
High-voltage circuit breaker fault diagnosis method based on deep learning Download PDFInfo
- Publication number
- CN115856611A CN115856611A CN202211666599.3A CN202211666599A CN115856611A CN 115856611 A CN115856611 A CN 115856611A CN 202211666599 A CN202211666599 A CN 202211666599A CN 115856611 A CN115856611 A CN 115856611A
- Authority
- CN
- China
- Prior art keywords
- voltage circuit
- working conditions
- circuit breaker
- reference signal
- 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
Images
Landscapes
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention provides a high-voltage circuit breaker fault diagnosis method based on deep learning, which comprises the steps of carrying out grouping test on high-voltage circuit breakers, recording data of detection parameters of the high-voltage circuit breakers under various working conditions, and classifying the recorded data according to the working conditions, wherein the working conditions comprise normal working conditions and a plurality of fault working conditions; training according to data under normal working conditions to obtain a reference signal; verifying the reference signal according to data in a plurality of fault conditions; and diagnosing and identifying the high-voltage circuit breaker by using a DTW algorithm according to the reference signal. According to the invention, by verifying the reference signal, the local optimal solution can be effectively avoided, and the reference signal is adjusted and corrected for many times, so that the method is more accurate than the traditional average or random selection method.
Description
Technical Field
The invention belongs to the technical field of breaker fault diagnosis, and particularly relates to a high-voltage breaker fault diagnosis method based on deep learning.
Background
As important high-voltage equipment in a power system, a high-voltage circuit breaker plays a role in control and protection in the power system, and the operation state of the high-voltage circuit breaker is directly related to the safety and stability of a line where the high-voltage circuit breaker is located. When a short-circuit fault occurs in the system, the high-voltage circuit breaker quickly cuts off a large short-circuit current to protect the entire power system from being damaged. However, the high-voltage circuit breakers are wide in variety, large in number, and complicated in structure, and are affected by factors such as self-quality and external operating conditions, and a fault may occur, resulting in a decrease in reliability of the entire power system.
In recent years, with the development of various algorithms and artificial intelligence, the fault diagnosis of a high-voltage circuit breaker is more and more biased to real-time performance and predictability, one of them is a high-voltage circuit breaker fault diagnosis method based on prior knowledge, which compares a measured signal with the prior knowledge to determine the state of the circuit breaker, wherein a dynamic time warping method, i.e., a DTW algorithm, is a relatively mature diagnosis method, and in the DTW algorithm, the prior knowledge is a reference signal, however, in the existing research, when selecting the DTW reference signal, most of all signals are randomly selected from normal state signals as the reference signal or the average of a plurality of signals is used as the reference signal, and a relatively good selection method is not given.
Disclosure of Invention
In view of the above, the present invention is directed to a method for diagnosing a fault of a high voltage circuit breaker based on deep learning, so as to solve the above problems in the prior art.
In order to solve the technical problems, the invention provides the following technical scheme:
a high-voltage circuit breaker fault diagnosis method based on deep learning comprises the following steps:
the method comprises the following steps that S1, grouping test is conducted on the high-voltage circuit breakers, data of detection parameters of the high-voltage circuit breakers under various working conditions are recorded, and the recorded data are classified according to the working conditions, wherein the working conditions comprise normal working conditions and multiple fault working conditions;
s2, training according to data under normal working conditions to obtain a reference signal;
s3, verifying a reference signal according to data in a plurality of fault working conditions;
and S4, diagnosing and identifying the high-voltage circuit breaker by utilizing a DTW algorithm according to the reference signal.
Further, in step S1, the detection parameter of the high-voltage circuit breaker includes one of a coil current, a contact vibration signal, a contact current and a contact moving stroke, and the detection parameters of the high-voltage circuit breaker are all variable parameters with respect to time.
Further, in step S2, the training method of the reference signal includes the following steps:
step S201, sampling data under normal working conditions to obtain a sampling signal X i ＝(i 1 ，i 2 ，...，i m ) I =1, 2, · n, n is the total number of data under normal working conditions, and m is the dimension of the sampling signal;
step S202, traversing all sample data, and acquiring the maximum difference value delta i of each vector of the signal k K =1, 2,. And m, i is selected k ’+Δi k /2 as reference vector element, i k To produce the maximum difference Δ i k Obtaining a reference signal X according to the minimum value of the time 0 ＝(i 1 ’+Δi 1 /2，i 2 ’+Δi 2 /2，...，i m ’+Δi m /2)；
Step S203, according to the action time of the breaker, the reference signal X is transmitted 0 And the sampling signal X i Segmenting, performing feature matching on each segment, selecting the most similar sampling signals for each segment, and finally selecting the most similar sampling signals according to the signal source X with the most similar sampling signals s As a new reference signal, X 0 As the adjustment signal.
Further, in step S203, the number of segments is three or more, and the three segments correspond to an action preparation segment, an action process segment, and an action termination segment.
Further, in step S3, the sample data of each fault condition is processed by using a clustering algorithm to obtain a clustering center signal and a clustering distance maximum value of the sample data of each fault condition, and X is compared s And comparing the distance value with the average value of the clustering distances if the distance value is less than the maximum value of the clustering distances, and using X if the distance value is less than the average value of the clustering distances 0 To X s Performing regulation by using X of non-similar segment 0 Sequential replacement of X with medium data s Corresponding to the vector element.
Further, the adjusting method comprises the following steps:
step S301, setting the number of replaced vector elements, initially setting the number to be 1, wherein the upper limit of the number is the upper limit of the number of the segmented vector elements, and terminating when the number reaches the upper limit;
and S302, arranging the vector elements in a descending order according to the difference values between the vector elements, sequentially replacing the corresponding vector elements, terminating the process if the distance value is larger than the average value of the clustering distance after replacement, and otherwise, adding one to the number of the replaced vector elements, and repeating the step S301.
Further, in step S4, the diagnosis and identification are implemented based on an artificial neural network, the artificial algorithm includes a plurality of neurons, the neurons correspond to identification under different conditions, the identification algorithm under different conditions is a DTW algorithm, the artificial neural algorithm inputs the data to be classified, the artificial neural algorithm outputs a unit vector containing only one element, and the unit vector corresponds to the condition identification result.
Further, the fault diagnosis method further comprises the following steps: and feeding back the sample data obtained after each pair of one sample is identified to the sample database of the corresponding classification, and recalculating the reference signal when the change rate of the sample variable of the sample database is more than 2%.
In summary, the invention provides a high-voltage circuit breaker fault diagnosis method based on deep learning, which comprises the steps of performing grouping test on high-voltage circuit breakers, recording data of detection parameters of the high-voltage circuit breakers under various working conditions, and classifying the recorded data according to the working conditions, wherein the working conditions comprise normal working conditions and a plurality of fault working conditions; training according to data under normal working conditions to obtain a reference signal; verifying the reference signal according to data in a plurality of fault working conditions; and diagnosing and identifying the high-voltage circuit breaker by using a DTW algorithm according to the reference signal. According to the invention, by verifying the reference signal, the local optimal solution can be effectively avoided, and the reference signal is adjusted and corrected for many times, so that the method is more accurate than the traditional average or random selection method.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a deep learning-based fault diagnosis method for a high-voltage circuit breaker according to an embodiment of the present invention.
Detailed Description
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the embodiments described below are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As important high-voltage equipment in a power system, a high-voltage circuit breaker plays a role in controlling and protecting the power system, and the operation state of the high-voltage circuit breaker is directly related to the safety and stability of a circuit where the high-voltage circuit breaker is located. When a short-circuit fault occurs in the system, the high-voltage circuit breaker quickly cuts off a large short-circuit current to protect the entire power system from damage. However, the high-voltage circuit breakers are wide in variety, large in number, and complicated in structure, and are affected by factors such as their own quality and external operating conditions, and a fault occurs in some cases, resulting in a decrease in reliability of the entire power system.
In recent years, with the development of various algorithms and artificial intelligence, the fault diagnosis of a high-voltage circuit breaker is more and more biased to real-time performance and predictability, one of them is a high-voltage circuit breaker fault diagnosis method based on prior knowledge, which compares a measured signal with the prior knowledge to determine the state of the circuit breaker, wherein a dynamic time warping method, i.e., a DTW algorithm, is a relatively mature diagnosis method, and in the DTW algorithm, the prior knowledge is a reference signal, however, in the existing research, when selecting the DTW reference signal, most of all signals are randomly selected from normal state signals as the reference signal or the average of a plurality of signals is used as the reference signal, and a relatively good selection method is not given.
Based on the above, the invention aims to provide a high-voltage circuit breaker fault diagnosis method based on deep learning.
The following describes an embodiment of a deep learning-based fault diagnosis method for a high-voltage circuit breaker in detail.
Referring to fig. 1, the present embodiment provides a method for diagnosing a fault of a high voltage circuit breaker based on deep learning, which includes the following steps:
the method comprises the following steps that S1, grouping test is conducted on the high-voltage circuit breakers, data of detection parameters of the high-voltage circuit breakers under various working conditions are recorded, and the recorded data are classified according to the working conditions, wherein the working conditions comprise normal working conditions and multiple fault working conditions;
s2, training according to data under normal working conditions to obtain a reference signal;
s3, verifying a reference signal according to data in a plurality of fault working conditions;
and S4, diagnosing and identifying the high-voltage circuit breaker by utilizing a DTW algorithm according to the reference signal.
The embodiment provides a high-voltage circuit breaker fault diagnosis method based on deep learning, which comprises the steps of carrying out grouping test on high-voltage circuit breakers, recording data of detection parameters of the high-voltage circuit breakers under various working conditions, and classifying the recorded data according to the working conditions, wherein the working conditions comprise normal working conditions and a plurality of fault working conditions; training according to data under normal working conditions to obtain a reference signal; verifying the reference signal according to data in a plurality of fault conditions; and diagnosing and identifying the high-voltage circuit breaker by using a DTW algorithm according to the reference signal. According to the invention, through verifying the reference signal, the local optimal solution can be effectively avoided, and the reference signal is adjusted and corrected for many times, so that the method is more accurate than the traditional average or arbitrary selection method.
In an optional embodiment, in step S1, the detection parameter of the high voltage circuit breaker includes one of a coil current, a contact vibration signal, a contact current and a contact moving stroke, and the detection parameters of the high voltage circuit breaker are all variable quantity parameters with respect to time.
In an alternative embodiment, in step S2, the training method of the reference signal includes the following steps:
step S201, sampling data under normal working conditions to obtain a sampling signal X i ＝(i 1 ，i 2 ，...，i m ) I =1, 2, · n, n is the total number of data under normal working conditions, and m is the dimension of the sampling signal;
step S202, traversing all sample data, and obtaining the maximum difference value delta i for each vector of the signal k K =1, 2,. And m, selecting i k ’+Δi k [ 2 ] asReference vector element, i k To produce the maximum difference Δ i k Obtaining a reference signal X according to the minimum value of the time 0 ＝(i 1 ’+Δi 1 /2，i 2 ’+Δi 2 /2，...，i m ’+Δi m /2)；
Step S203, converting the reference signal X according to the action time of the breaker 0 And the sampling signal X i Segmenting, performing feature matching on each segment, selecting the most similar sampling signals for each segment, and finally selecting the most similar sampling signals according to the signal source X with the most similar sampling signals s As a new reference signal, X 0 As the adjustment signal.
In an alternative embodiment, in step S203, the number of segments is more than three, and the three segments correspond to an action preparation segment, an action process segment and an action termination segment.
In an optional embodiment, in step S3, the clustering algorithm is used to process the sample data of each fault condition to obtain a clustering center signal and a clustering distance maximum value of the sample data of each fault condition, and X is compared s And comparing the distance value with the average value of the clustering distances if the distance value is less than the maximum value of the clustering distances, and using X if the distance value is less than the average value of the clustering distances 0 To X s Performing regulation by using X of non-similar segment 0 In-sequence replacement of X by data s Corresponding to the vector element.
In an alternative embodiment, the adjustment method comprises the steps of:
step S301, setting the number of replaced vector elements, initially setting the number to be 1, wherein the upper limit of the number is the upper limit of the number of the segmented vector elements, and terminating when the number reaches the upper limit;
and S302, arranging the vector elements in a descending order according to the difference values between the vector elements, sequentially replacing the corresponding vector elements, terminating if the distance value is greater than the average value of the clustering distance after replacement, otherwise, adding one to the number of the replaced vector elements, and repeating the step S301.
In an optional embodiment, in step S4, the diagnosis and identification are implemented based on an artificial neural network, the artificial algorithm includes a plurality of neurons, the neurons correspond to identification under different conditions, the identification algorithm under different conditions is a DTW algorithm, the artificial neural algorithm inputs the data to be classified, the artificial neural algorithm outputs a unit vector containing only one element, and the unit vector corresponds to the condition identification result.
In an optional embodiment, the fault diagnosis method further comprises the steps of: and feeding back the sample data obtained after each pair of one sample is identified to the sample database of the corresponding classification, and recalculating the reference signal when the change rate of the sample variable of the sample database is more than 2%.
Compared with the prior art, the method comprises the steps of firstly carrying out arithmetic solving on sample data to obtain the median, then matching the most similar sample according to the action time subsection of the actual breaker and the action time subsection, selecting the reference signal if the sample is the middle value of the normal working range interval, then comparing the reference signal with the sample centers of other fault working conditions, and further adjusting the reference signal.
The above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (8)
1. A high-voltage circuit breaker fault diagnosis method based on deep learning is characterized by comprising the following steps:
the method comprises the following steps of S1, performing grouping test on the high-voltage circuit breakers, recording data of detection parameters of the high-voltage circuit breakers under various working conditions, and classifying the recorded data according to the working conditions, wherein the working conditions comprise normal working conditions and multiple fault working conditions;
s2, training according to data under normal working conditions to obtain a reference signal;
s3, verifying a reference signal according to data in a plurality of fault working conditions;
and S4, diagnosing and identifying the high-voltage circuit breaker by using a DTW algorithm according to the reference signal.
2. The high-voltage circuit breaker fault diagnosis method based on deep learning of claim 1, characterized in that: in the step S1, the detection parameter of the high-voltage circuit breaker includes one of a coil current, a contact vibration signal, a contact current, and a contact movement stroke, and the detection parameter of the high-voltage circuit breaker is a time-related variation parameter.
3. The high-voltage circuit breaker fault diagnosis method based on deep learning as claimed in claim 2, characterized in that: in step S2, the training method of the reference signal includes the following steps:
step S201, sampling data under normal working conditions to obtain a sampling signal X i ＝(i 1 ，i 2 ，...，i m ) I =1, 2, n is the total number of data under normal working conditions, and m is the dimension of the sampling signal;
step S202, traversing all sample data, and obtaining the maximum difference value delta i for each vector of the signal k K =1, 2,. And m, i is selected k ’+Δi k /2 as reference vector element, i k ' to produce the maximum difference value Δ i k The minimum value of time correspondence is obtained to obtain a reference signal X 0 ＝(i 1 ’+Δi 1 /2，i 2 ’+Δi 2 /2，...，i m ’+Δi m /2)；
Step S203, converting the reference signal X according to the action time of the breaker 0 And the sampling signal X i Segmenting, performing feature matching on each segment, and selecting the most similar sampling information from each segmentNumber, finally according to the signal source X with the most similar sampling signals s As a new reference signal, X 0 As the adjustment signal.
4. The high-voltage circuit breaker fault diagnosis method based on deep learning of claim 3, characterized in that: in step S203, the number of segments is three or more, and the three segments correspond to an action preparation segment, an action process segment, and an action termination segment.
5. The high-voltage circuit breaker fault diagnosis method based on deep learning of claim 4, characterized in that: in the step S3, the clustering algorithm is used for respectively processing the sample data of each fault working condition to obtain a clustering center signal and a clustering distance maximum value of the sample data of each fault working condition, and X is compared s And the distance value of each clustering center signal is compared with the average value of the clustering distances by using the distance value if the distance value is less than the maximum value of the clustering distances, and X is used if the distance value is less than the average value of the clustering distances 0 To X s Performing regulation by using X of non-similar segment 0 Sequential replacement of X with medium data s Corresponding to the vector element.
6. The high-voltage circuit breaker fault diagnosis method based on deep learning of claim 5, characterized in that: the adjusting method comprises the following steps:
step S301, setting the number of the replaced vector elements, initially setting the number as 1, wherein the upper limit of the number is the upper limit of the number of the segmented vector elements, and the method is terminated when the number reaches the upper limit;
and S302, arranging the vector elements in a descending order according to the difference values between the vector elements, sequentially replacing the corresponding vector elements, terminating if the distance value is larger than the average value of the clustering distance after replacement, otherwise, adding one to the number of the replaced vector elements, and repeating the step S301.
7. The high-voltage circuit breaker fault diagnosis method based on deep learning of claim 6, characterized in that: in the step S4, the diagnosis and identification are implemented based on an artificial neural network, the artificial algorithm includes a plurality of neurons, the neurons correspond to identification of different working conditions respectively, the identification algorithm of different working conditions is a DTW algorithm, the artificial neural algorithm inputs data to be classified, the artificial neural algorithm outputs a unit vector including only one element, and the unit vector corresponds to a working condition identification result.
8. The deep learning based high voltage circuit breaker fault diagnosis method according to any one of claims 1-7, characterized in that: the fault diagnosis method further includes the steps of: and feeding back the sample data obtained after each pair of samples is identified to the sample database of the corresponding classification, and recalculating the reference signal when the change rate of the sample variable of the sample database is more than 2%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211666599.3A CN115856611A (en) | 2022-12-23 | 2022-12-23 | High-voltage circuit breaker fault diagnosis method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211666599.3A CN115856611A (en) | 2022-12-23 | 2022-12-23 | High-voltage circuit breaker fault diagnosis method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115856611A true CN115856611A (en) | 2023-03-28 |
Family
ID=85654403
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211666599.3A Pending CN115856611A (en) | 2022-12-23 | 2022-12-23 | High-voltage circuit breaker fault diagnosis method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115856611A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116643163A (en) * | 2023-07-27 | 2023-08-25 | 浙江北岛科技有限公司 | Remote on-line monitoring system of vacuum circuit breaker |
-
2022
- 2022-12-23 CN CN202211666599.3A patent/CN115856611A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116643163A (en) * | 2023-07-27 | 2023-08-25 | 浙江北岛科技有限公司 | Remote on-line monitoring system of vacuum circuit breaker |
| CN116643163B (en) * | 2023-07-27 | 2023-10-20 | 浙江北岛科技有限公司 | Remote on-line monitoring system of vacuum circuit breaker |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Goswami et al. | Predictive model for classification of power system faults using machine learning | |
| CN110569925B (en) | LSTM-based time sequence abnormality detection method applied to power equipment operation detection | |
| CN110161388B (en) | Fault type identification method and system of high-voltage equipment | |
| CN107561997A (en) | A kind of power equipment state monitoring method based on big data decision tree | |
| CN113111591B (en) | Automatic diagnosis method, device and equipment based on internal fault of modular power distribution terminal | |
| CN115856611A (en) | High-voltage circuit breaker fault diagnosis method based on deep learning | |
| CN111999591B (en) | Method for identifying abnormal state of primary equipment of power distribution network | |
| CN111856209A (en) | Power transmission line fault classification method and device | |
| CN115546558A (en) | Electrical equipment insulation fault state classification method and device and storage medium | |
| Xiao et al. | A fault diagnosis method of insulator string based on infrared image feature extraction and probabilistic neural network | |
| CN110716155A (en) | Low-voltage power supply loop abnormal working condition perception algorithm and device | |
| CN118033409B (en) | GCB arc extinguishing chamber switch resistance testing method | |
| CN114091549A (en) | Equipment fault diagnosis method based on deep residual error network | |
| CN113065522A (en) | Transformer partial discharge type identification method based on deep neural network | |
| Chang et al. | Assessment of the insulation status aging in power cable joints using support vector machine | |
| CN110927497A (en) | Point switch fault detection method and device | |
| CN118035909A (en) | Collaborative detection method for abnormal situation of load of platform area based on empirical mode decomposition | |
| CN111898446A (en) | Single-phase earth fault studying and judging method based on multi-algorithm normalization analysis | |
| CN109784777B (en) | Power grid equipment state evaluation method based on time sequence information fragment cloud similarity measurement | |
| CN115792442A (en) | Data-driven comprehensive fault diagnosis method for direct-current switch cabinet for rail transit | |
| CN115859594A (en) | Health evaluation algorithm of power transformation equipment based on hierarchical analysis | |
| CN118294750B (en) | Automatic point-to-point verification system based on relay protection | |
| CN112906489A (en) | Transformer machine winding looseness identification method based on LMD and permutation entropy | |
| CN116796213B (en) | Power distribution network line transformation relation identification method based on clustering algorithm | |
| CN116961072B (en) | Multi-DC chain commutation failure recognition method and system based on space-time convolution 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 |