CN117972398A - Method and system for extracting fault characteristic information of circuit breaker - Google Patents
Method and system for extracting fault characteristic information of circuit breaker Download PDFInfo
- Publication number
- CN117972398A CN117972398A CN202410362392.XA CN202410362392A CN117972398A CN 117972398 A CN117972398 A CN 117972398A CN 202410362392 A CN202410362392 A CN 202410362392A CN 117972398 A CN117972398 A CN 117972398A
- Authority
- CN
- China
- Prior art keywords
- spectrogram
- training
- fault
- comparison
- determined
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000001228 spectrum Methods 0.000 claims abstract description 81
- 238000009826 distribution Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims description 204
- 230000008569 process Effects 0.000 claims description 28
- 238000012795 verification Methods 0.000 claims description 22
- 238000004364 calculation method Methods 0.000 claims description 21
- 238000005070 sampling Methods 0.000 claims description 18
- 238000010586 diagram Methods 0.000 claims description 13
- 238000000605 extraction Methods 0.000 claims description 12
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 description 14
- 238000004590 computer program Methods 0.000 description 8
- 238000011156 evaluation Methods 0.000 description 8
- 230000002159 abnormal effect Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000003595 spectral effect Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000005303 weighing Methods 0.000 description 2
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 description 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Complex Calculations (AREA)
Abstract
The invention discloses a method and a system for extracting fault characteristic information of a circuit breaker, and relates to the technical field of fault treatment of circuit breakers. The method for extracting the fault characteristic information of the circuit breaker comprises the steps of converting an obtained voltage signal of the circuit breaker into a spectrogram to be determined; identifying a spectrogram to be determined based on a trained frequency spectrum identification model, and obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker; the fault type of the circuit breaker is determined based on the similarity degree of the spectrogram to be determined and the classification spectrogram, fault characteristic information corresponding to the spectrogram to be determined is obtained, the problem can be found in time when the fault occurs by analyzing the spectrogram of the voltage signal and utilizing the trained model, and detailed information about the fault type such as peak value, energy distribution and duration time is provided.
Description
Technical Field
The invention relates to the technical field of breaker fault processing, in particular to a method and a system for extracting breaker fault characteristic information.
Background
The circuit breaker is a switch position capable of closing, carrying and breaking a current under normal loop conditions and capable of closing, carrying and breaking a current under abnormal loop conditions within a specified time, and plays an important role in the power transmission process. In this way, in the management information system of large-scale power grid enterprises in China, the related breaker fault description text input by each substation node forms a huge-scale database. As one of the most important switching equipment in the power system, the circuit breaker consists of a plurality of sub-components, has a complex structure, can know and master the common fault type and characterization information of the circuit breaker and the association relation between the circuit breaker and the components, can help to realize the real-time monitoring of the health state of the circuit breaker, ensures the safe and stable operation of the power system, and has important economic and practical significance.
The Chinese patent with the authority of bulletin number of CN105677833B discloses a method for extracting the fault characteristic information of a circuit breaker based on a text mining technology, the characteristic information is extracted from the semi-structured text data of the fault of the circuit breaker by using the text mining technology, the association relation between the fault type and each part is accurately established, and the intelligent analysis of the fault of the circuit breaker is realized and becomes the practical technical problem to be solved by an information system of a power supply enterprise. Based on the calculation and comparison of the similarity, the fault types of the circuit breakers are clustered and standardized, and the texts are grouped according to the fault types; designing a forward maximum matching word segmentation algorithm to segment the grouped text, labeling part of speech based on a dictionary matching method, and identifying and extracting common characterization information of various types of faults by combining a elimination method; based on the co-occurrence criterion and the statistical method, the association relation between each fault representation and the circuit breaker component and between the components is established, which is favorable for finding and extracting the deep cause causing the fault and provides basis for preventing the circuit breaker fault.
However, the prior art has an inconvenience of finding a problem in time when a fault occurs and providing detailed information about the type of the fault.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a method and a system for extracting fault characteristic information of a circuit breaker, which solve the problems that the prior art is inconvenient to find problems in time when faults occur and provides detailed information about fault types.
In order to achieve the above purpose, the invention is realized by the following technical scheme: a method for extracting fault characteristic information of a circuit breaker comprises the following steps: converting the obtained voltage signal of the circuit breaker into a spectrogram to be determined; identifying a spectrogram to be determined based on a trained frequency spectrum identification model, and obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker; and determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram, and acquiring fault characteristic information corresponding to the spectrogram to be determined, wherein the fault characteristic information comprises peak value, energy distribution and fault duration.
Further, the process of converting the obtained voltage signal of the circuit breaker into the spectrogram to be determined is as follows: preprocessing the discrete data of the acquired voltage signals; applying a window function to the preprocessed discrete data to reduce discontinuities in the voltage signal at the window edge; performing fast Fourier transform on the discrete data processed by the window function, and converting the voltage signal from a time domain to a frequency domain to obtain frequency information of the voltage signal in the frequency domain; the frequency domain information is visualized as a spectrogram.
Further, based on the trained spectrum recognition model, recognizing the to-be-determined spectrogram, and acquiring the similarity degree of the to-be-determined spectrogram and the classified spectrogram is as follows: comparing the spectrogram to be determined with the normal spectrogram to obtain the correlation coefficient of the spectrogram to be determined and the normal spectrogram, wherein the correlation coefficient is used for representing the similarity degree of the spectrogram to be determined and the normal spectrogram; the calculation formula of the correlation coefficient is as follows:
，
in the method, in the process of the invention, Is a correlation coefficient,/>For the number of random sampling points in a spectrogram to be determined,/>，For the total number of random sampling points in the spectrogram to be determined,/>For/>, in the spectrogram to be determinedFrequency to be determined of random sampling points,/>For the frequency average value to be determined of random sampling points in the spectrogram to be determined,/>For the/>, in the normal spectrogram and the spectrogram to be determinedCorresponding normal frequency of random sampling points,/>For the normal spectrogram and the spectrogram to be determined/>Corresponding normal frequency average values of the random sampling points; and comparing the spectrogram to be determined with each fault spectrogram respectively, and obtaining mutual information of the spectrogram to be determined and each fault spectrogram respectively, wherein the mutual information is used for representing the similarity degree of the spectrogram to be determined and each fault spectrogram respectively.
Further, the process of determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram is as follows: judging whether the correlation coefficient is larger than a set normal judging threshold value of the circuit breaker, if so, the circuit breaker has no fault; obtaining breaker fault judging thresholds corresponding to various types of faults, and comparing the mutual information of the spectrogram to be determined and each fault spectrogram with the corresponding breaker fault judging thresholds respectively; and if the mutual information of the spectrogram to be determined and the fault spectrogram is larger than the corresponding breaker fault judgment threshold value, determining that the breaker fault type is the fault type corresponding to the breaker fault judgment threshold value.
Further, the training process of the spectrum recognition model is as follows: inputting the prepared training spectrum atlas into a spectrum recognition model, and acquiring training data of the spectrum recognition model on the training spectrum atlas each time; based on each training data, and combining the comparison data corresponding to each training data to evaluate the proficiency index of each training of the spectrum identification model, wherein the proficiency index is used for representing the identification capability of the spectrum identification model on a training spectrum atlas after training; the calculation formula of the proficiency index is:
，
in the method, in the process of the invention, For training times, number,/>For/>Training proficiency index,/>For/>Accuracy coefficient of secondary training,/>For/>Timeliness coefficient of secondary training,/>Weight factor being accuracy coefficient,/>A weight factor that is a timeliness coefficient; judging whether the proficiency index is larger than a set proficiency threshold value, if so, inputting the prepared verification spectrum atlas into a spectrum recognition model, verifying the spectrum recognition model, if the verification is successful, finishing training, and if not, adjusting the spectrum recognition model.
Further, the process of evaluating the proficiency index of each training of the spectrum recognition model based on each training data and in combination with the corresponding comparison data of each training data is as follows: acquiring training data of each training, wherein the training data of each training comprises training accuracy of each training and fault identification success time of each training; acquiring comparison data corresponding to training data of each training, wherein the comparison data comprises a training comparison accuracy rate and a recognition success comparison time; calculating an accuracy coefficient corresponding to each training based on the training accuracy of each time and the training ratio accuracy corresponding to each training, and calculating a timeliness coefficient corresponding to each training based on the fault identification success time of each time and the identification success comparison time corresponding to the fault identification success time of each time; the calculation formula of the accuracy coefficient is:
；
in the method, in the process of the invention, Numbering of fault types,/>，/>Is the total number of fault types,/>For training the first time/>Training accuracy of successful identification of individual fault types,/>Is AND/>The corresponding training ratio is accurate; the timeliness coefficient is calculated by the following formula:
，
in the method, in the process of the invention, For/>Number of successful number of fault identification in samples of each fault type,/>，/>For the total number of successful fault identification,/>For/>Training of the first time/>First/>, of the fault typeFailure recognition success time of each successfully recognized sample,/>Is AND/>Corresponding recognition success comparison time,/>For/>A weight factor for the recognition success time of the second failure type; and calculating the proficiency index of each training based on the accuracy coefficient and the timeliness coefficient corresponding to each training.
Further, the comparison data corresponding to the first training in the comparison data corresponding to the training data of each training is preset, and the comparison data corresponding to the non-first training is obtained based on the accuracy coefficient and the timeliness coefficient of the last training, wherein the obtaining process is as follows: the accuracy coefficient and the timeliness coefficient of each training are compared with the accuracy comparison coefficient and the timeliness comparison coefficient corresponding to each comparison interval stored in the comparison database, and comparison parameters are obtained, wherein the comparison parameters are used for representing the corresponding degree of the accuracy coefficient and the timeliness coefficient of each training and each comparison interval stored in the comparison database; and comparing the comparison parameter with a set comparison threshold, and if the comparison parameter is smaller than the comparison threshold, acquiring the training comparison accuracy and the recognition success comparison time corresponding to the comparison interval corresponding to the comparison parameter as comparison data corresponding to the training data of the next training.
Further, the calculation formula of the comparison parameter is as follows:
; in the/> For the number of training times,，/>For the total number of training,/>For/>Alignment parameters of secondary training,/>For/>Accuracy coefficient of secondary training,/>For/>Timeliness coefficient of secondary training,/>For the number of the comparison interval,/>，/>For the total number of comparison intervals,/>For/>Accuracy comparison coefficient of individual comparison interval,/>For/>Timeliness comparison coefficient of individual comparison interval,/>As accuracy weighting factor,/>Is a timeliness weight factor, and/>。
Further, the process of verifying the spectrum recognition model is as follows: and acquiring the accuracy and precision of the spectrum identification model on the verification spectrum atlas, and if the accuracy and precision meet the set requirements, finishing the verification.
The utility model provides a circuit breaker fault feature information extraction system, includes spectrogram conversion module, similarity determination module and fault feature information acquisition module, wherein: the frequency spectrum diagram conversion module is used for converting the obtained voltage signal of the circuit breaker into a frequency spectrum diagram to be determined; the similarity degree determining module is used for identifying a spectrogram to be determined based on a trained frequency spectrum identification model, and obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker; the fault characteristic information acquisition module is used for determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram, and acquiring fault characteristic information corresponding to the spectrogram to be determined, wherein the fault characteristic information comprises a peak value, energy distribution and fault duration.
The invention has the following beneficial effects:
(1) According to the circuit breaker fault characteristic information extraction method, through evaluation of accuracy and precision, performance of the model on a verification data set can be comprehensively and objectively evaluated, reliability and generalization capability of the model are verified, if the accuracy and the precision do not reach expectations, the verification process is helpful for analyzing problems existing in the model, guidance is provided for subsequent improvement work, and therefore effects and applicability of the model are improved.
(2) According to the circuit breaker fault characteristic information extraction method, problems can be found in time when faults occur by analyzing the spectrogram of the voltage signal and utilizing the trained model, and detailed information about fault types such as peak value, energy distribution and duration time is provided. This may help maintenance personnel to more effectively locate and resolve the circuit breaker problem, thereby improving the reliability and safety of the power system.
(3) According to the breaker fault characteristic information extraction method, the prepared training frequency spectrum atlas is input into a frequency spectrum recognition model, training of the model is carried out, and model parameters are adjusted through a back propagation algorithm to minimize a loss function on the training frequency spectrum atlas. Based on each training data and corresponding comparison data, a proficiency index is calculated, wherein the proficiency index adopts a linear combination of an accuracy coefficient and a timeliness coefficient, the accuracy coefficient represents the accuracy of the model, and the timeliness coefficient represents the response speed of the model. The weighting factors are used to balance the importance of both.
(4) According to the breaker fault characteristic information extraction method, more comprehensive performance evaluation is provided by considering training accuracy, fault identification success time, training comparison accuracy and identification success comparison time, and the multidimensional evaluation is helpful for deeper understanding of the performances of the model in different aspects. By calculating the accuracy coefficient, the influence of each fault type can be known more carefully, the model can be adjusted more pertinently, and the identification capability of specific fault types is improved.
(5) According to the breaker fault characteristic information extraction method, based on the fault identification success time and the identification success comparison time, the timeliness coefficient of each training is calculated, the time factor of fault identification is considered by the coefficient, and the timeliness of the model is evaluated by weighing the fault identification success time and the corresponding comparison time. For each successful recognition sample of each fault type, calculating the ratio of the fault recognition success time to the corresponding recognition success comparison time, multiplying by a weight factor.
Of course, it is not necessary for any one product to practice the invention to achieve all of the advantages set forth above at the same time.
Drawings
Fig. 1 is a flowchart of a method for extracting fault characteristic information of a circuit breaker according to the present invention.
Fig. 2 is a flow chart of a circuit breaker fault characteristic information extraction system according to the present invention.
Detailed Description
The embodiment of the application realizes timely finding of the problem when the fault occurs and provides detailed information about the fault type through the method and the system for extracting the fault characteristic information of the circuit breaker.
The problems in the embodiment of the application have the following general ideas:
Voltage signal data of the circuit breaker is obtained, and preprocessing steps including filtering, downsampling and the like are carried out to prepare signals for spectrum analysis. The pre-processed voltage signal is converted into a spectrogram reflecting the distribution of the signal over different frequencies, involving the use of fourier transforms or other frequency domain analysis methods. A spectral identification model is constructed and trained that is capable of distinguishing between the spectrograms of a normal circuit breaker and a fault circuit breaker using a training data set containing normal and fault samples.
Classifying the spectrogram to be determined by using a trained model, and calculating the similarity degree of the spectrogram to be determined and the normal and fault spectrograms. Based on the result of the degree of similarity, it is determined whether the breaker has a fault, and if so, the type of fault is further determined. If it is determined that a fault exists, fault characteristic information such as peak value, energy distribution and fault duration is extracted from the spectrogram to be determined.
Referring to fig. 1, the embodiment of the invention provides a technical scheme: a method for extracting fault characteristic information of a circuit breaker comprises the following steps: converting the obtained voltage signal of the circuit breaker into a spectrogram to be determined; identifying a spectrogram to be determined based on the trained frequency spectrum identification model, and obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker; and determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram, and acquiring fault characteristic information corresponding to the spectrogram to be determined, wherein the fault characteristic information comprises peak value, energy distribution and fault duration.
The voltage signal is converted into the spectrogram, the fault is identified through the trained model, the problem can be found in time when the breaker breaks down, the power failure time is reduced, the reliability of a power system is improved, the spectrogram is classified and identified through the machine learning model, the automatic detection and treatment of the breaker fault are realized, the requirement of manual intervention is reduced, and the efficiency is improved.
The fault type can be determined according to the similarity of the spectrograms, fault characteristic information such as peak value, energy distribution and fault duration can be extracted, more detailed fault information is provided for maintenance personnel, and accurate positioning and problem solving are facilitated.
Specifically, the process of converting the obtained voltage signal of the circuit breaker into the spectrogram to be determined is as follows: preprocessing the discrete data of the acquired voltage signals; applying a window function to the preprocessed discrete data to reduce discontinuities in the voltage signal at the window edge; performing fast Fourier transform on the discrete data processed by the window function, and converting the voltage signal from a time domain to a frequency domain to obtain frequency information of the voltage signal in the frequency domain; the frequency domain information is visualized as a spectrogram.
In this embodiment, the voltage signal obtained from the circuit breaker first requires discrete data acquisition and preprocessing to eliminate noise, smooth data, or other necessary data cleaning operations to ensure signal quality. The use of a window function helps to reduce discontinuities in the discrete signal at the edges of the window, reducing the effects of spectral leakage, and the selection of the window function may be made according to specific requirements, such as hanning windows, hamming windows, etc.
And carrying out FFT (fast Fourier transform) on the discrete data processed by the window function, and converting the signal from a time domain to a frequency domain, wherein the FFT is an efficient algorithm for representing the signal in the frequency domain to obtain the frequency spectrum information of the signal. And finally, visualizing the obtained frequency domain information into a spectrogram. The spectrogram generally has a frequency on the horizontal axis and a signal amplitude or power on the vertical axis, and shows the distribution of the signal on different frequencies intuitively through a graph.
The conversion to the frequency domain enables the frequency characteristics of the signal to be more clearly visible, helps to find the components of specific frequencies in the frequency domain, so that the property of the signal is better understood, and the application of the window function and the use of FFT help to reduce the influence of noise and improve the quality and accuracy of the signal.
Specifically, based on the trained spectrum recognition model, recognizing the to-be-determined spectrogram, and acquiring the similarity degree of the to-be-determined spectrogram and the classified spectrogram is as follows: and comparing the spectrogram to be determined with the normal spectrogram to obtain the correlation coefficient of the spectrogram to be determined and the normal spectrogram, wherein the correlation coefficient is used for representing the similarity degree of the spectrogram to be determined and the normal spectrogram.
The calculation formula of the correlation coefficient is as follows: In the above, the ratio of/> Is a correlation coefficient,/>For the number of random sampling points in a spectrogram to be determined,/>，/>For the total number of random sampling points in the spectrogram to be determined,/>For/>, in the spectrogram to be determinedFrequency to be determined of random sampling points,/>For/>, in the spectrogram to be determinedFrequency mean value to be determined of random sampling points,/>For the/>, in the normal spectrogram and the spectrogram to be determinedCorresponding normal frequency of random sampling points,/>For the normal spectrogram and the spectrogram to be determined/>The corresponding normal frequency average of the random sampling points.
Firstly, the spectrogram to be determined is compared with a normal spectrogram, so as to determine whether the spectrogram to be determined accords with the normal condition in the aspect of frequency spectrum characteristics. The correlation coefficient is used to quantify the degree of similarity between the spectrogram to be determined and the normal spectrogram. The correlation coefficient is a statistic used to describe the correlation between two variables, and its value ranges from-1 to 1, 0 indicates no linear correlation, 1 indicates a complete positive correlation, and-1 indicates a complete negative correlation.
The correlation coefficient is used as a similarity index, so that the similarity degree between the spectrogram to be determined and the normal spectrogram can be quantitatively evaluated, an objective quantization index is provided, and whether the spectrogram to be determined accords with the normal condition or not can be accurately judged. The calculation of the correlation coefficient can be used in a frequency spectrum identification model as a discrimination index to distinguish a normal frequency spectrum graph from an abnormal frequency spectrum graph, thereby being beneficial to automatically identifying the abnormal frequency spectrum graph and improving the efficiency and the reliability of the system.
And comparing the spectrogram to be determined with each fault spectrogram respectively, and obtaining mutual information of the spectrogram to be determined and each fault spectrogram respectively, wherein the mutual information is used for representing the similarity degree of the spectrogram to be determined and each fault spectrogram respectively.
The calculation formula of mutual information is: In which, in the process, For the mutual information of the spectrogram to be determined and the fault spectrogram,/>Is a random variable in the spectrogram to be determined, and，/>For the number of random frequency points in a spectrogram to be determined,/>，/>Is the total number of random frequency points,/>For/>Frequency to be determined of random frequency points,/>Is a random variable in the fault spectrogram, and/>For the number of random frequency points in a spectrogram to be determined,/>，Is the total number of random frequency points,/>For/>The random frequency points are to be frequency determined.
In this embodiment, the to-be-determined spectrograms are compared with the fault spectrograms respectively, so as to help determine whether the to-be-determined spectrograms have similarity or correlation with the known fault spectrograms. Mutual information is used as an index for measuring the degree of association between a spectrogram to be determined and a fault spectrogram, and is a measure in an information theory and is used for describing the degree of interdependence between two random variables.
By calculating mutual information between the spectrogram to be determined and each fault spectrogram, abnormal modes or fault characteristics in the spectrogram can be found, and potential fault signals or abnormal conditions can be found early. Mutual information is used as a quantization index, so that the detection accuracy of a frequency spectrum identification model can be improved, and the similarity between a to-be-determined spectrogram and a known fault spectrogram can be estimated more accurately through calculation of the mutual information, so that fault diagnosis and prediction can be performed more reliably.
Specifically, the process of determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram is as follows: judging whether the correlation coefficient is larger than a set normal judging threshold value of the circuit breaker, if so, the circuit breaker has no fault; obtaining breaker fault judging thresholds corresponding to various types of faults, and comparing the mutual information of the spectrogram to be determined and each fault spectrogram with the corresponding breaker fault judging thresholds respectively; and if the mutual information of the spectrogram to be determined and the fault spectrogram is larger than the corresponding breaker fault judgment threshold value, determining that the breaker fault type is the fault type corresponding to the breaker fault judgment threshold value.
The training process of the spectrum recognition model is as follows: inputting the prepared training spectrum atlas into a spectrum recognition model, and acquiring training data of the spectrum recognition model on the training spectrum atlas each time; based on each training data, and combining the corresponding comparison data of each training data to evaluate the proficiency index of each training of the spectrum recognition model, wherein the proficiency index is used for representing the recognition capability of the spectrum recognition model to the training spectrum atlas after training.
The calculation formula of the proficiency index is: In the above, the ratio of/> For training times, number,/>For/>Training proficiency index,/>For/>Accuracy coefficient of secondary training,/>For/>Timeliness coefficient of secondary training,/>Weight factor being accuracy coefficient,/>Is a weight factor for the timeliness coefficient.
Judging whether the proficiency index is larger than a set proficiency threshold value, if so, inputting the prepared verification spectrum atlas into a spectrum recognition model, verifying the spectrum recognition model, if the verification is successful, finishing training, and if not, adjusting the spectrum recognition model.
In this embodiment, first, the prepared training spectral atlas is input to the spectral recognition model for training of the model, including adjusting model parameters by a back propagation algorithm to minimize the loss function on the training spectral atlas. Based on each training data and corresponding comparison data, a proficiency index is calculated, wherein the proficiency index adopts a linear combination of an accuracy coefficient and a timeliness coefficient, the accuracy coefficient represents the accuracy of the model, and the timeliness coefficient represents the response speed of the model. The weighting factors are used to balance the importance of both.
It is determined whether the proficiency index for each training is greater than a set proficiency threshold. If the proficiency index is high enough, the recognition capability of the model on the training spectrum atlas is good, and if the proficiency index meets the set threshold, the prepared verification spectrum atlas is input into the spectrum recognition model for verification. Verification success means that the model performs well on unseen data as well, and training ends.
If the verification is unsuccessful, it may be stated that the model is less generalizable to new data. In this case, the adjustment of the spectrum recognition model may include adjusting super parameters such as model structure and learning rate, so as to improve the generalization performance of the model.
Specifically, the process of evaluating the proficiency index of each training of the spectrum recognition model based on each training data and in combination with the corresponding comparison data of each training data is as follows: acquiring training data of each training, wherein the training data of each training comprises training accuracy of each training and fault identification success time of each training; acquiring comparison data corresponding to training data of each training, wherein the comparison data comprises a training comparison accuracy rate and a recognition success comparison time; and calculating the accuracy coefficient corresponding to each training based on the training accuracy of each training and the training ratio corresponding to each training accuracy.
The calculation formula of the accuracy coefficient is:
In the above, the ratio of/> Numbering of fault types,/>，/>Is the total number of fault types,/>For/>Training of the first time/>Training accuracy of successful identification of individual fault types,/>Is AND/>The corresponding training ratio is the accuracy rate.
First, training data is obtained from each training, including the accuracy of each training and the failure recognition success time, which reflect the performance of the model during the training process. Meanwhile, the comparison data corresponding to the training data of each training is obtained. The comparison data comprises a training comparison accuracy rate and a recognition success comparison time, and is used for comparing the training comparison accuracy rate and the recognition success comparison time with the training data, so that the performance of the model is further evaluated.
According to the training accuracy and the training ratio accuracy, calculating an accuracy coefficient corresponding to each training, wherein the accuracy coefficient calculation formula adopts the relation between the successful accuracy of the identification of each fault type and the corresponding training ratio accuracy. The proficiency index of each training is obtained through a calculation formula of the proficiency index. This index takes into account the model's performance in accuracy and timeliness.
By considering the training accuracy, the fault recognition success time, the training comparison accuracy and the recognition success comparison time at the same time, more comprehensive performance evaluation is provided, and the multidimensional evaluation is helpful for deeper understanding of the performances of the model in different aspects. By calculating the accuracy coefficient, the influence of each fault type can be known more carefully, the model can be adjusted more pertinently, and the identification capability of specific fault types is improved.
And calculating the timeliness coefficient corresponding to each training based on the fault identification success time of each time and the identification success comparison time corresponding to the fault identification success time of each time.
The timeliness coefficient is calculated by the following formula: In the above, the ratio of/> For/>Number of successful number of fault identification in samples of each fault type,/>，/>For the total number of successful fault identification,For/>Training of the first time/>First/>, of the fault typeFailure recognition success time of each successfully recognized sample,Is AND/>Corresponding recognition success comparison time,/>For/>Training of the first time/>Weight factor for the identification success time of each fault type.
And calculating the proficiency index of each training based on the accuracy coefficient and the timeliness coefficient corresponding to each training.
In this embodiment, first, based on the failure recognition success time and the recognition success comparison time, a timeliness coefficient of each training is calculated, the time factor of the failure recognition is considered by the coefficient, and timeliness of the model is evaluated by weighing the failure recognition success time and the corresponding comparison time. For each successful recognition sample of each fault type, calculating the ratio of the fault recognition success time to the corresponding recognition success comparison time, multiplying by a weight factor.
Through comprehensive consideration of accuracy coefficients and timeliness coefficients, comprehensive evaluation of model performance is provided, so that a proficiency index is closer to actual application requirements, accuracy is focused, timeliness is focused, and requirements of some actual scenes are met.
Specifically, the comparison data corresponding to the first training in the comparison data corresponding to the training data of each training is preset, and the comparison data corresponding to the non-first training is obtained based on the accuracy coefficient and timeliness coefficient of the last training, wherein the obtaining process is as follows: the accuracy coefficient and the timeliness coefficient of each training are compared with the accuracy comparison coefficient and the timeliness comparison coefficient corresponding to each comparison interval stored in the comparison database, and comparison parameters are obtained and used for representing the corresponding degree of the accuracy coefficient and the timeliness coefficient of each training and each comparison interval stored in the comparison database; and comparing the comparison parameter with a set comparison threshold, and if the comparison parameter is smaller than the comparison threshold, acquiring the training comparison accuracy and the recognition success comparison time corresponding to the comparison interval corresponding to the comparison parameter as comparison data corresponding to the training data of the next training.
The calculation formula of the comparison parameters is as follows:
; in the/> For the number of training times,，/>For the total number of training,/>For/>Alignment parameters of secondary training,/>For the accuracy coefficient of the first training,/>For/>Timeliness coefficient of secondary training,/>For the number of the comparison interval,/>，/>For the total number of comparison intervals,/>For/>Accuracy comparison coefficient of individual comparison interval,/>For/>Timeliness comparison coefficient of individual comparison interval,/>As accuracy weighting factor,/>Is a timeliness weight factor, and/>。
In this embodiment, the accuracy coefficient and the timeliness coefficient are used to compare with the accuracy comparison coefficient and the timeliness comparison coefficient of each comparison interval in the comparison database through a calculation formula of the comparison parameters. The comparison parameters integrate the difference of accuracy and timeliness, and a parameter representing the corresponding degree is obtained through logarithmic transformation and linear combination.
The calculation of the comparison parameters comprises the difference of accuracy and timeliness, wherein the weight factors of the accuracy and the timeliness can adjust the importance degree of the accuracy and the timeliness. And comparing the comparison parameters with the set comparison threshold. If the comparison parameter is smaller than the comparison threshold, the accuracy and timeliness of the training are relatively good, and the training can be considered as a good training. At this time, the training comparison accuracy and the recognition success comparison time of the comparison interval corresponding to the comparison parameter are obtained.
After the comparison interval meeting the conditions is obtained, the training comparison accuracy and the recognition success comparison time corresponding to the interval are used as comparison data corresponding to training data of the next training. Thus, flexible comparison parameter selection according to the performance of the last training is realized.
The dynamic adjustment of the comparison data of each training is realized through the calculation of the comparison parameters and the setting of the threshold value, the comparison data can be flexibly selected according to the actual performance, and the performance fluctuation of the model in different training stages is adapted. The performance of the model on accuracy and timeliness can be considered more comprehensively through comprehensive comparison of the accuracy coefficient and the timeliness coefficient. And the weight factors are set so that the importance degree of accuracy and timeliness can be adjusted according to actual requirements.
Specifically, the process of verifying the spectrum recognition model is as follows: and acquiring the accuracy and precision of the spectrum identification model on the verification spectrum atlas, and if the accuracy and precision meet the set requirements, finishing the verification.
In the embodiment, through the evaluation of accuracy and precision, the performance of the spectrum identification model on the verification data set can be objectively evaluated, so that the knowledge of how the model performs in an actual application scene can be facilitated, and whether the expected requirement is met or not can be facilitated. By verifying the accuracy and precision of the model on the verification data set, the reliability of the model can be verified, and if the accuracy and precision are high and stable, the model is consistent in performance on different data sets, and the generalization capability is high.
The circuit breaker fault characteristic information extraction system comprises a spectrogram conversion module, a similarity degree determination module and a fault characteristic information acquisition module, as shown in fig. 2, wherein: the frequency spectrum diagram conversion module is used for converting the obtained voltage signal of the circuit breaker into a frequency spectrum diagram to be determined; the similarity degree determining module is used for identifying a spectrogram to be determined based on the trained frequency spectrum identification model, obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker; the fault characteristic information acquisition module is used for determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram, and acquiring fault characteristic information corresponding to the spectrogram to be determined, wherein the fault characteristic information comprises a peak value, energy distribution and fault duration.
In this embodiment, the spectrogram conversion module converts the acquired voltage signal of the circuit breaker into a spectrogram. The spectrogram can represent the signal in the frequency domain, helping to analyze the frequency content of the signal. The similarity degree determining module uses a pre-trained frequency spectrum recognition model to recognize a frequency spectrum diagram to be determined. This model may be a machine learning model, such as a deep learning network, for distinguishing normal spectrograms from fault spectrograms. By comparing the degree of similarity of the spectrogram to be determined and the classified spectrograms (normal and fault spectrograms), the system can determine the state of the spectrogram to be determined.
The fault characteristic information acquisition module determines the fault type of the circuit breaker according to the result of the similarity determination module, and acquires fault characteristic information such as peak value, energy distribution, fault duration and the like corresponding to the spectrogram to be determined.
An electronic device, comprising: a processor; and a memory in which computer program instructions are stored which, when executed by the processor, cause the processor to perform the circuit breaker fault signature information extraction method as described above.
A computer-readable storage medium storing a program which, when executed by a processor, implements the circuit breaker failure feature information extraction method as described above.
In summary, the present application has at least the following effects:
through the evaluation of accuracy and precision, the performance of the model on the verification data set can be comprehensively and objectively evaluated, the reliability and generalization capability of the model are verified, and if the accuracy and precision do not reach the expectations, the verification process is helpful for analyzing the problems of the model, and guidance is provided for subsequent improvement work, so that the effect and applicability of the model are improved.
By analyzing the spectrogram of the voltage signal and using the trained model, problems can be found in time when a fault occurs and detailed information about the type of fault, such as peak value, energy distribution and duration, is provided. This may help maintenance personnel to more effectively locate and resolve the circuit breaker problem, thereby improving the reliability and safety of the power system.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of systems, apparatuses (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. The method for extracting the fault characteristic information of the circuit breaker is characterized by comprising the following steps of:
Converting the obtained voltage signal of the circuit breaker into a spectrogram to be determined;
Identifying a spectrogram to be determined based on a trained frequency spectrum identification model, and obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker;
And determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram, and acquiring fault characteristic information corresponding to the spectrogram to be determined, wherein the fault characteristic information comprises peak value, energy distribution and fault duration.
2. The method for extracting fault characteristic information of a circuit breaker according to claim 1, wherein the process of converting the obtained voltage signal of the circuit breaker into the spectrogram to be determined is as follows:
preprocessing the discrete data of the acquired voltage signals;
applying a window function to the preprocessed discrete data to reduce discontinuities in the voltage signal at the window edge;
Performing fast Fourier transform on the discrete data processed by the window function, and converting the voltage signal from a time domain to a frequency domain to obtain frequency information of the voltage signal in the frequency domain;
the frequency domain information is visualized as a spectrogram.
3. The method for extracting fault characteristic information of a circuit breaker according to claim 1, wherein the process of identifying the spectrogram to be determined based on the trained spectrum identification model and obtaining the similarity degree between the spectrogram to be determined and the classified spectrogram is as follows:
Comparing the spectrogram to be determined with the normal spectrogram to obtain the correlation coefficient of the spectrogram to be determined and the normal spectrogram, wherein the correlation coefficient is used for representing the similarity degree of the spectrogram to be determined and the normal spectrogram;
The calculation formula of the correlation coefficient is as follows:
In the above, the ratio of/> Is a correlation coefficient,/>For the number of random sampling points in a spectrogram to be determined,/>，/>For the total number of random sampling points in the spectrogram to be determined,/>For/>, in the spectrogram to be determinedFrequency to be determined of random sampling points,/>For/>, in the spectrogram to be determinedFrequency mean value to be determined of random sampling points,/>For the/>, in the normal spectrogram and the spectrogram to be determinedCorresponding normal frequency of random sampling points,/>For the normal spectrogram and the spectrogram to be determined/>Corresponding normal frequency average values of the random sampling points;
and comparing the spectrogram to be determined with each fault spectrogram respectively, and obtaining mutual information of the spectrogram to be determined and each fault spectrogram respectively, wherein the mutual information is used for representing the similarity degree of the spectrogram to be determined and each fault spectrogram respectively.
4. A method for extracting fault signature information of a circuit breaker according to claim 3, wherein the process of determining the fault type of the circuit breaker based on the similarity between the spectrogram to be determined and the classification spectrogram is as follows:
Judging whether the correlation coefficient is larger than a set normal judging threshold value of the circuit breaker, if so, the circuit breaker has no fault;
Obtaining breaker fault judging thresholds corresponding to various types of faults, and comparing the mutual information of the spectrogram to be determined and each fault spectrogram with the corresponding breaker fault judging thresholds respectively;
And if the mutual information of the spectrogram to be determined and the fault spectrogram is larger than the corresponding breaker fault judgment threshold value, determining that the breaker fault type is the fault type corresponding to the breaker fault judgment threshold value.
5. The method for extracting fault signature information of a circuit breaker according to claim 1, wherein the training process of the spectrum identification model is as follows:
inputting the prepared training spectrum atlas into a spectrum recognition model, and acquiring training data of the spectrum recognition model on the training spectrum atlas each time;
Based on each training data, and combining the comparison data corresponding to each training data to evaluate the proficiency index of each training of the spectrum identification model, wherein the proficiency index is used for representing the identification capability of the spectrum identification model on a training spectrum atlas after training;
The calculation formula of the proficiency index is: In the above, the ratio of/> For the number of training times,For/>Training proficiency index,/>For/>Accuracy coefficient of secondary training,/>For/>Timeliness coefficient of secondary training,/>Weight factor being accuracy coefficient,/>A weight factor that is a timeliness coefficient;
judging whether the proficiency index is larger than a set proficiency threshold value, if so, inputting the prepared verification spectrum atlas into a spectrum recognition model, verifying the spectrum recognition model, if the verification is successful, finishing training, and if not, adjusting the spectrum recognition model.
6. The method for extracting fault signature of a circuit breaker according to claim 5, wherein the step of evaluating the proficiency index of each training of the spectrum recognition model based on each training data and in combination with the corresponding comparison of each training data is as follows:
acquiring training data of each training, wherein the training data of each training comprises training accuracy of each training and fault identification success time of each training;
Acquiring comparison data corresponding to training data of each training, wherein the comparison data comprises a training comparison accuracy rate and a recognition success comparison time;
Calculating an accuracy coefficient corresponding to each training based on the training accuracy of each time and the training ratio accuracy corresponding to each training, and calculating a timeliness coefficient corresponding to each training based on the fault identification success time of each time and the identification success comparison time corresponding to the fault identification success time of each time;
the calculation formula of the accuracy coefficient is:
In the above, the ratio of/> Numbering of fault types,/>，/>Is the total number of fault types,/>For/>Training of the first time/>Training accuracy of successful identification of individual fault types,/>Is AND/>The corresponding training ratio is accurate;
The timeliness coefficient is calculated by the following formula: In the above, the ratio of/> For/>Number of successful number of fault identification in samples of each fault type,/>，/>For the total number of successful fault identification,/>For/>Training of the first time/>First/>, of the fault typeFailure recognition success time of each successfully recognized sample,/>Is AND/>Corresponding recognition success comparison time,/>For/>Training of the first time/>A weight factor for the identification success time of each fault type;
And calculating the proficiency index of each training based on the accuracy coefficient and the timeliness coefficient corresponding to each training.
7. The method for extracting fault characteristic information of circuit breaker according to claim 6, wherein the comparison data corresponding to the first training of the comparison data corresponding to the training data of each training is preset, and the comparison data corresponding to the non-first training data is obtained based on the accuracy coefficient and timeliness coefficient of the last training, and the obtaining process is as follows:
The accuracy coefficient and the timeliness coefficient of each training are compared with the accuracy comparison coefficient and the timeliness comparison coefficient corresponding to each comparison interval stored in the comparison database, and comparison parameters are obtained, wherein the comparison parameters are used for representing the corresponding degree of the accuracy coefficient and the timeliness coefficient of each training and each comparison interval stored in the comparison database;
And comparing the comparison parameter with a set comparison threshold, and if the comparison parameter is smaller than the comparison threshold, acquiring the training comparison accuracy and the recognition success comparison time corresponding to the comparison interval corresponding to the comparison parameter as comparison data corresponding to the training data of the next training.
8. The method for extracting fault signature information of a circuit breaker according to claim 7, wherein: the calculation formula of the comparison parameters is as follows:
；
in the method, in the process of the invention, For training times, number,/>，/>For the total number of training,/>For/>Alignment parameters of secondary training,/>For/>Accuracy coefficient of secondary training,/>For/>Timeliness coefficient of secondary training,/>For the number of the comparison interval,/>，/>For the total number of comparison intervals,/>For/>Accuracy comparison coefficient of individual comparison interval,/>For/>Timeliness comparison coefficient of individual comparison interval,/>As accuracy weighting factor,/>Is a timeliness weight factor, and/>。
9. The method for extracting fault signature information of a circuit breaker according to claim 1, wherein the process of verifying the spectrum identification model is as follows: and acquiring the accuracy and precision of the spectrum identification model on the verification spectrum atlas, and if the accuracy and precision meet the set requirements, finishing the verification.
10. A circuit breaker fault signature information extraction system for use in a circuit breaker fault signature information extraction method as claimed in any one of claims 1 to 9, comprising a spectrogram conversion module, a similarity determination module and a fault signature information acquisition module, wherein:
The frequency spectrum diagram conversion module is used for converting the obtained voltage signal of the circuit breaker into a frequency spectrum diagram to be determined;
the similarity degree determining module is used for identifying a spectrogram to be determined based on a trained frequency spectrum identification model, and obtaining the similarity degree of the spectrogram to be determined and a classification spectrogram, wherein the classification spectrogram comprises a normal spectrogram corresponding to a voltage signal of a normal circuit breaker and a fault spectrogram corresponding to the voltage signal of a fault circuit breaker;
The fault characteristic information acquisition module is used for determining the fault type of the circuit breaker based on the similarity degree of the spectrogram to be determined and the classification spectrogram, and acquiring fault characteristic information corresponding to the spectrogram to be determined, wherein the fault characteristic information comprises a peak value, energy distribution and fault duration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410362392.XA CN117972398B (en) | 2024-03-28 | 2024-03-28 | Method and system for extracting fault characteristic information of circuit breaker |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410362392.XA CN117972398B (en) | 2024-03-28 | 2024-03-28 | Method and system for extracting fault characteristic information of circuit breaker |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117972398A true CN117972398A (en) | 2024-05-03 |
| CN117972398B CN117972398B (en) | 2024-06-14 |
Family
ID=90848097
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410362392.XA Active CN117972398B (en) | 2024-03-28 | 2024-03-28 | Method and system for extracting fault characteristic information of circuit breaker |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117972398B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118330379A (en) * | 2024-06-17 | 2024-07-12 | 北京金冠智能电气科技有限公司 | Lightning arrester running state detection method, device, system and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109001649A (en) * | 2018-07-21 | 2018-12-14 | 成都光电传感技术研究所有限公司 | A kind of power supply smart diagnostic system and guard method |
| CN112146882A (en) * | 2020-10-12 | 2020-12-29 | 中国人民解放军海军工程大学 | Bearing fault diagnosis method based on transfer learning vibration signal image recognition |
| CN114966312A (en) * | 2022-05-18 | 2022-08-30 | 广东电网有限责任公司 | Power distribution network fault detection and positioning method and system based on migration convolutional neural network |
| CN115185918A (en) * | 2022-06-06 | 2022-10-14 | 浪潮软件集团有限公司 | Method and device for automatically classifying system logs |
| CN116735170A (en) * | 2023-04-25 | 2023-09-12 | 西北工业大学 | Intelligent fault diagnosis method based on self-attention multi-scale feature extraction |
| WO2024046363A1 (en) * | 2022-09-01 | 2024-03-07 | 珠海市伊特高科技有限公司 | Gis partial discharge diagnosis method and apparatus, model training method, and system |
-
2024
- 2024-03-28 CN CN202410362392.XA patent/CN117972398B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109001649A (en) * | 2018-07-21 | 2018-12-14 | 成都光电传感技术研究所有限公司 | A kind of power supply smart diagnostic system and guard method |
| CN112146882A (en) * | 2020-10-12 | 2020-12-29 | 中国人民解放军海军工程大学 | Bearing fault diagnosis method based on transfer learning vibration signal image recognition |
| CN114966312A (en) * | 2022-05-18 | 2022-08-30 | 广东电网有限责任公司 | Power distribution network fault detection and positioning method and system based on migration convolutional neural network |
| CN115185918A (en) * | 2022-06-06 | 2022-10-14 | 浪潮软件集团有限公司 | Method and device for automatically classifying system logs |
| WO2024046363A1 (en) * | 2022-09-01 | 2024-03-07 | 珠海市伊特高科技有限公司 | Gis partial discharge diagnosis method and apparatus, model training method, and system |
| CN116735170A (en) * | 2023-04-25 | 2023-09-12 | 西北工业大学 | Intelligent fault diagnosis method based on self-attention multi-scale feature extraction |
Non-Patent Citations (2)
| Title |
|---|
| YUNYI ZHU等: "Fault Diagnosis of High-Voltage Circuit Breaker Based on Digital Twin", 2021 INTERNATIONAL CONFERENCE ON ADVANCED ELECTRICAL EQUIPMENT AND RELIABLE OPERATION (AEERO), 16 February 2022 (2022-02-16) * |
| 鄢仁武;林穿;高硕勋;罗家满;李天建;夏正邦;: "基于小波时频图和卷积神经网络的断路器故障诊断分析", 振动与冲击, no. 10, 28 May 2020 (2020-05-28) * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118330379A (en) * | 2024-06-17 | 2024-07-12 | 北京金冠智能电气科技有限公司 | Lightning arrester running state detection method, device, system and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117972398B (en) | 2024-06-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117972398B (en) | Method and system for extracting fault characteristic information of circuit breaker | |
| CN109034046B (en) | Method for automatically identifying foreign matters in electric energy meter based on acoustic detection | |
| CN109599120B (en) | Abnormal mammal sound monitoring method based on large-scale farm plant | |
| CN112201260B (en) | Transformer running state online detection method based on voiceprint recognition | |
| CN101995437B (en) | Method for extracting features of crack acoustic emission signal of drawing part | |
| CN116559598B (en) | Smart distribution network fault positioning method and system | |
| CN102426835A (en) | Method for identifying local discharge signals of switchboard based on support vector machine model | |
| CN109448726A (en) | A kind of method of adjustment and system of voice control accuracy rate | |
| US20160322064A1 (en) | Method and apparatus for signal extraction of audio signal | |
| CN114325256A (en) | Power equipment partial discharge identification method, system, equipment and storage medium | |
| CN111045902A (en) | Pressure testing method and device for server | |
| CN110580492A (en) | Track circuit fault precursor discovery method based on small fluctuation detection | |
| CN116778964A (en) | Power transformation equipment fault monitoring system and method based on voiceprint recognition | |
| CN112462355A (en) | Sea target intelligent detection method based on time-frequency three-feature extraction | |
| CN115954017A (en) | HHT-based engine small sample sound abnormal fault identification method and system | |
| CN117668751A (en) | High-low voltage power system fault diagnosis method and device | |
| CN109490776B (en) | Mobile phone vibration motor good and defective product detection method based on machine learning | |
| CN111639583A (en) | Method and system for identifying power quality disturbance of power grid | |
| CN111523317A (en) | Voice quality inspection method and device, electronic equipment and medium | |
| CN115128345A (en) | Power grid safety early warning method and system based on harmonic monitoring | |
| CN114974229A (en) | Method and system for extracting abnormal behaviors based on audio data of power field operation | |
| CN111489736B (en) | Automatic scoring device and method for seat speaking operation | |
| CN109165396A (en) | A kind of equipment remaining life prediction technique of failure evolution trend | |
| CN110321425B (en) | Method and device for judging defect type of power grid | |
| CN117195077A (en) | Unsupervised detection method for fault of voiceprint signal of power transformer |
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 |