A method for extracting low-voltage series arc fault components in a power collection terminal

Abstract

The present invention provides a method for extracting low-voltage series arc fault components of an electric power acquisition terminal, and relates to the technical field of current signal arc fault detection. The method comprises: obtaining a load current signal and a series arc fault current signal; performing modal decomposition on the load current signal and the series arc fault current signal to obtain different frequency intrinsic modal components of the original current signal; calculating the correlation coefficients of the load current signal and the series arc fault current signal with their respective intrinsic modal components according to the different frequency intrinsic modal components, and determining the fundamental frequency components of the load current signal and the series arc fault current signal according to the correlation coefficient. The present invention is conducive to the improvement of the real-time performance and accuracy of fault feature expression and fault detection through the original current signal decomposition, high-frequency band intrinsic modal component selection, and fault sensitive component accurate extraction provided by the present invention.

Description

Technical Field

[0001] The present invention relates to the technical field of current signal arc fault detection, and in particular to a method for extracting low-voltage series arc fault components for a power collection terminal. Background Art

[0002] When a series arc fault occurs in a low-voltage line, the line current signal usually contains components such as normal load current, line noise current and arc fault current. It is difficult to directly analyze and extract fault features from the original signal containing the arc fault current component.

[0003] The main reason is that the arc fault current component usually appears as a weak fault signal, which is easily submerged in the main signal of the load current; and because there are usually some inherent noise current components in the line, these components are easy to couple with the arc fault current component to form a mixed signal, making the fault signal less significant. Therefore, it is extremely difficult to extract fault features and accurately identify series arc faults. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method for extracting low-voltage series arc fault components of a power collection terminal. Through the original current signal decomposition, high-frequency intrinsic mode component selection, and accurate extraction of fault sensitive components provided by the present invention, it is beneficial to improve the real-time and accuracy of fault feature expression and fault detection.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0006] In a first aspect, a method for extracting low-voltage series arc fault components of a power collection terminal is provided, the method comprising:

[0007] Obtaining load current signal and series arc fault current signal;

[0008] Perform modal decomposition on the load current signal and the series arc fault current signal to obtain the different frequency intrinsic modal components of the original current signal;

[0009] According to the different frequency eigenmode components, the correlation coefficients between the load current signal and the series arc fault current signal and their respective eigenmode components are calculated, and the fundamental frequency components of the load current signal and the series arc fault current signal are determined according to the correlation coefficients;

[0010] According to the fundamental frequency component, each intrinsic mode component of the normal load current signal and the series arc fault current signal is divided into frequency bands to obtain divided frequency bands;

[0011] According to the divided frequency bands, the corresponding characteristic components are extracted;

[0012] Calculate the similarity between feature components to construct a similarity measurement matrix;

[0013] According to the similarity measurement matrix, the corresponding key components are screened;

[0014] The key components are compared and analyzed with the preset similarity threshold to obtain the sensitive fault component signal caused by the series arc fault.

[0015] Furthermore, the load current signal and the series arc fault current signal are modally decomposed to obtain the different frequency intrinsic modal components of the original current signal, including:

[0016] pass The load current signal and the series arc fault current signal are decomposed to obtain the intrinsic mode components of the original current signal, where: is the total number of different modal components decomposed, It is eigenmode components, is the residual component obtained by decomposition, is the intrinsic mode component of the original current signal;

[0017] According to the intrinsic mode components of the original current signal, Transform to obtain the different frequency eigenmode components of the original current signal, where: is the index of the time domain sampling point, is the frequency, is the length of each short period, It is The eigenmode components are at the sampling point The value at is the index of the eigenmode component, It is The result of Fourier transform of a short period of time is It is the overlap point between short periods; is a complex exponential function, representing the basis function of frequency domain transformation.

[0018] Further, according to the different frequency eigenmode components, the correlation coefficients between the load current signal and the series arc fault current signal and their respective eigenmode components are calculated, and the fundamental frequency components of the load current signal and the series arc fault current signal are determined according to the correlation coefficients, including:

[0019] By Calculate the correlation coefficient of each intrinsic mode component of the load current signal and the series arc fault current signal ,in, is the original current signal, It is eigenmode components, is the original current signal sequence length, is the mean of the original signal sequence, It is The mean of the sequence of eigenmode components, is the time index, is the index of the eigenmode component;

[0020] The correlation coefficients of the intrinsic modal components of the load current signal and the series arc fault current signal are calculated. The signals are sorted to obtain the fundamental frequency components of the load current signal and the series arc fault current signal.

[0021] Further, according to the fundamental frequency component, each intrinsic mode component of the normal load current signal and the series arc fault current signal is divided into frequency bands to obtain divided frequency bands, including:

[0022] With the fundamental frequency component as the center, the frequency range is expanded to both sides to form a frequency band including the fundamental frequency;

[0023] For each eigenmode component, it is classified into a corresponding frequency band according to its frequency characteristics to obtain a series of divided frequency bands, each of which contains a group of eigenmode components with similar frequency characteristics.

[0024] Furthermore, the similarity between the feature components is calculated to construct a similarity measurement matrix, including:

[0025] Performing standardization on the characteristic components to obtain standardized characteristic components;

[0026] For each pair of normalized feature components, Calculate the similarity value between each pair of standardized feature components, where and are two feature vectors, representing the two data sets whose similarity needs to be calculated. Each vector contains Sample values; Indicates the number of samples; is an index, traversing from 1 to Each sample of and Represents the eigenvectors and the eigenvector Middle The value of samples; and The eigenvectors are and the eigenvector The mean of and The eigenvectors are and the eigenvector The standard deviation of Represents the calculation of eigenvector and the eigenvector The function of the cosine similarity between ;

[0027] Create a size of N × N The matrix of N is the number of eigencomponents;

[0028] Traverse all feature component pairs and fill the similarity values ​​into the corresponding positions of the matrix. Each element of the matrix ( i , j ) indicates the i The feature components and j The similarity between the feature components.

[0029] Furthermore, according to the similarity measurement matrix, the corresponding key components are screened, including:

[0030] According to the similarity measurement matrix, traverse each value in the similarity measurement matrix, according to each high-frequency intrinsic mode component of the normal load current signal, to obtain the similarity with all fault current components;

[0031] The similarities with all fault current components are sorted in descending order and screened to obtain the corresponding key components.

[0032] Furthermore, the key components are compared and analyzed with the preset similarity threshold to obtain the sensitive fault component signal caused by the series arc fault, including:

[0033] Setting a modal component similarity threshold, and comparing the value in the similarity measurement matrix with the modal component similarity threshold to obtain a comparison result;

[0034] The results are analyzed and compared to obtain the sensitive fault component signal caused by the series arc fault.

[0035] In a second aspect, a low-voltage series arc fault component extraction system for a power collection terminal includes:

[0036] An acquisition module is used to acquire a load current signal and a series arc fault current signal; perform modal decomposition on the load current signal and the series arc fault current signal to acquire different frequency intrinsic modal components of the original current signal; calculate the correlation coefficients between the load current signal and the series arc fault current signal and their respective intrinsic modal components according to the different frequency intrinsic modal components, and determine the fundamental frequency components of the load current signal and the series arc fault current signal according to the correlation coefficients;

[0037] The processing module is used to divide the intrinsic mode components of the normal load current signal and the series arc fault current signal into frequency bands according to the fundamental frequency components to obtain the divided frequency bands; extract the corresponding characteristic components according to the divided frequency bands; calculate the similarity between the characteristic components to construct a similarity measurement matrix; screen the corresponding key components according to the similarity measurement matrix; compare and analyze the key components with the preset similarity threshold to obtain the sensitive fault component signal caused by the series arc fault.

[0038] According to a third aspect, a computing device includes:

[0039] one or more processors;

[0040] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method described.

[0041] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described is implemented.

[0042] The above solution of the present invention includes at least the following beneficial effects:

[0043] By performing fine modal decomposition on the load current signal and the series arc fault current signal, the sensitive fault components that cause the series arc fault can be accurately extracted, which helps to accurately express the fault characteristics and thus improve the accuracy of fault detection; by providing the original current signal decomposition, high-frequency intrinsic modal component selection, and accurate extraction of fault sensitive components, the sensitive fault components that cause the series arc fault can be separated from the original line current, and the current signal can be quickly processed and analyzed, and the fault can be discovered and located in time. This is crucial for power systems that require rapid response, which can reduce the impact of faults on system operation and improve the stability and reliability of the system. Extracting and analyzing fault components in an automated manner reduces the need for manual intervention, improves work efficiency, and reduces the possibility of human error. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flow chart of a method for extracting low-voltage series arc fault components of a power collection terminal provided by an embodiment of the present invention.

[0045] Figure 2 It is a schematic diagram of a low-voltage series arc fault component extraction system of a power collection terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in a form and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0047] like Figure 1 As shown, an embodiment of the present invention provides a method for extracting low-voltage series arc fault components of a power collection terminal, the method comprising the following steps:

[0048] Step 11, obtaining a load current signal and a series arc fault current signal;

[0049] Step 12, performing modal decomposition on the load current signal and the series arc fault current signal to obtain different frequency intrinsic modal components of the original current signal;

[0050] Step 13, calculating the correlation coefficients between the load current signal and the series arc fault current signal and their respective eigenmode components according to the eigenmode components of different frequencies, and determining the fundamental frequency components of the load current signal and the series arc fault current signal according to the correlation coefficients;

[0051] Step 14, dividing each intrinsic mode component of the normal load current signal and the series arc fault current signal into frequency bands according to the fundamental frequency component to obtain divided frequency bands;

[0052] Step 15, extracting corresponding characteristic components according to the divided frequency bands;

[0053] Step 16, calculating the similarity between the feature components to construct a similarity measurement matrix;

[0054] Step 17, screening corresponding key components according to the similarity measurement matrix;

[0055] Step 18: Compare and analyze the key component with the preset similarity threshold to obtain a sensitive fault component signal caused by the series arc fault.

[0056] In the embodiment of the present invention, the current signal of the system in normal and faulty states can be comprehensively collected and analyzed. Through modal decomposition, the intrinsic structure and characteristics of the current signal can be deeply understood, which helps to reveal the complexity and nonlinear behavior of the signal. Accurately identifying the fundamental frequency component helps to understand the main components of the current signal. Through frequency band division, the characteristics of the current signal in different frequency ranges can be analyzed more carefully, and the components that are irrelevant to the fault or have little impact can be removed. The components of the first frequency band are highlighted, which helps to focus on the fault characteristics. Through similarity measurement, the similarity between each component signal can be quantitatively analyzed, and the fault component with the greatest difference from the normal state can be effectively identified. By setting a threshold and performing comparative analysis, the effectiveness of the screening component can be verified to ensure that the final sensitive fault component signal has a high degree of accuracy and reliability.

[0057] In a preferred embodiment of the present invention, the above step 11 may include:

[0058] Collecting normal load current signals and series arc fault current signals, and preprocessing the current signals to obtain preprocessed data;

[0059] Extract frequency domain features from the preprocessed data, map the extracted feature values ​​to "nodes", and initialize the positions of the ants;

[0060] Each ant uses the pheromone distribution and heuristic information of the nodes around its current position to Select the next node to move, where It's an ant k An optional set of neighboring nodes, and is the weight coefficient, is the pheromone concentration, indicating that the node i To Node j pheromones; and It is heuristic information; Pheromone concentration, indicating the node To Node pheromones; Represents ants Select the node to move from the current node probability; is an index variable that iterates over the ant The currently selectable set of neighboring nodes ;

[0061] After each iteration, the pheromone concentration on the path is updated according to the effect of the feature combination found by the ants on distinguishing normal and fault currents, and the maximum number of iterations is reached to achieve the feature combination that can best distinguish normal load current from series arc fault current;

[0062] According to the iteration result, the load current signal and the series arc fault current signal are obtained.

[0063] In the embodiment of the present invention, the current data of the system in normal operation and fault state can be collected comprehensively and accurately, and the preprocessing can remove the noise, abnormal values ​​and irrelevant information in the original data, improve the quality of the data, and by extracting the frequency domain features, the behavior pattern of the current signal in the frequency domain can be revealed, which is helpful to capture the unique characteristics of the fault signal, and the characteristic values ​​are mapped to nodes in the algorithm. By using the intelligent search ability of the ant colony algorithm, the feature combination can be efficiently explored and optimized in the complex feature space, thereby improving the accuracy and efficiency of fault detection. By continuously updating the pheromone concentration, the algorithm can adaptively adjust the search direction and gradually focus on those feature combinations that are most effective in distinguishing normal and fault currents; through a sufficient iterative process, it can converge to the optimal or approximately optimal feature combination, and re-analyze the current signal using the optimized feature combination, so that potential fault modes can be more accurately identified.

[0064] In a specific embodiment of the present invention, load current signals and fault current signals are collected under normal working conditions, and features such as spectrum peaks, energy distribution, and harmonic components are extracted from the collected normal load current signals and series arc fault current signals; the extracted eigenvalues ​​are mapped to "nodes", each node represents a eigenvalue or a combination of features, and the basic parameters of the ant colony algorithm are set, such as the initial pheromone concentration, the calculation method of heuristic information, and the position of the ants are initialized; each ant calculates the probability of selecting the next node based on the pheromone distribution and heuristic information of the nodes around the current position, and each ant selects the next node to move based on the probability, that is, selects a eigenvalue or a combination of features. After each iteration, by Calculate the accuracy by Calculate the precision, where is normal current is correctly classified as normal, is the fault current being misclassified as normal, is normal current being misclassified as a fault, The fault current is correctly classified as a fault; the effect of the found feature combination on distinguishing normal and fault currents is evaluated based on the accuracy and precision. The process of ant path selection, feature combination evaluation and pheromone update is repeated until the maximum number of iterations is reached. During the iteration process, the focus is gradually placed on the feature combinations that are most effective in distinguishing normal load current from series arc fault current. Based on the iteration results, the load current signal and series arc fault current signal are obtained.

[0065] In a preferred embodiment of the present invention, the above step 12 may include:

[0066] pass The load current signal and the series arc fault current signal are decomposed to obtain the intrinsic mode components of the original current signal, where: is the total number of different modal components decomposed, It is eigenmode components, is the residual component obtained by decomposition, is the intrinsic mode component of the original current signal;

[0067] According to the intrinsic mode components of the original current signal, Transform to obtain the different frequency eigenmode components of the original current signal, where: is the index of the time domain sampling point, is the frequency, is the length of each short period, It is The eigenmode components are at the sampling point The value at is the index of the eigenmode component, It is The result of Fourier transform of a short period of time is It is the overlap point between short periods; is a complex exponential function, representing the basis function of frequency domain transformation.

[0068] In an embodiment of the present invention, the intrinsic mode components obtained by decomposition can reveal the intrinsic structure and dynamic behavior of the current signal, which is helpful for in-depth understanding of the characteristics of the current signal on different time scales. By transforming and analyzing the frequency characteristics of each mode, the components of different frequency ranges in the current signal can be accurately identified. Series arc faults usually produce specific frequency components in the current signal. By accurately identifying these components, the sensitivity and accuracy of fault detection can be improved; decomposition and frequency characteristic analysis can help remove frequency components that are not related to fault detection, reduce the amount of data processing, and improve the efficiency and real-time performance of signal processing. Through the detailed analysis of the current signal, the operating status of the system can be better understood and abnormal conditions can be discovered in time.

[0069] In a specific embodiment of the present invention, the original current signal is decomposed into a series of simple intrinsic modal components with specific frequency characteristics; in the decomposition process, the total number of different modal components, as well as each intrinsic modal component and the residual component obtained by decomposition are obtained. Each modal component represents an inherent oscillation mode in the original signal. Each intrinsic modal component is divided into several short time periods, the length of each short time period, and these short time periods can overlap, and the number of overlapping points is M For the The transformation results are calculated for each short period. By analyzing the transformation results of each short period, the frequency components of each intrinsic modal component in each period can be determined until the frequency characteristics of all modal components are analyzed. According to the obtained frequency characteristics of each modal component, the frequency range corresponding to different modal components is identified by observing the peak value and energy distribution of the frequency spectrum, and the identified frequency range is compared with the known frequency characteristics of normal load current and series arc fault current to obtain different frequency intrinsic modal components of the original current signal.

[0070] In a preferred embodiment of the present invention, the above step 13 may include:

[0071] pass Calculate the correlation coefficient of each intrinsic mode component of the load current signal and the series arc fault current signal ,in, is the original current signal, It is eigenmode components, is the original current signal sequence length, is the mean of the original signal sequence, It is The mean of the sequence of eigenmode components, is the time index, is the index of the eigenmode component;

[0072] The correlation coefficients of the intrinsic modal components of the load current signal and the series arc fault current signal are calculated. The signals are sorted to obtain the fundamental frequency components of the load current signal and the series arc fault current signal.

[0073] In an embodiment of the present invention, by calculating and sorting the Pearson correlation coefficient, the intrinsic modal component most relevant to the load current signal, that is, the fundamental frequency component, can be accurately identified. After determining the fundamental frequency component, more attention can be paid to this main component for signal analysis, thereby simplifying the analysis process. The fundamental frequency component usually contains the main characteristics of the signal. By analyzing this component, abnormal conditions such as series arc faults can be more effectively detected. By continuously monitoring and analyzing the fundamental frequency component, early warning of potential faults can be achieved. It helps to detect and deal with small faults that may cause bigger problems in advance.

[0074] In a preferred embodiment of the present invention, the above step 14 may include:

[0075] With the fundamental frequency component as the center, the frequency range is expanded to both sides to form a frequency band including the fundamental frequency;

[0076] For each eigenmode component, it is classified into a corresponding frequency band according to its frequency characteristics to obtain a series of divided frequency bands, each of which contains a group of eigenmode components with similar frequency characteristics.

[0077] In the embodiment of the present invention, by setting a threshold value related to the fundamental frequency component, the modal component representing the main characteristics of the signal can be more accurately identified. The division criteria of the high frequency band and the low frequency band are determined based on the fundamental frequency component. By dividing the intrinsic modal component into the high frequency band and the low frequency band, the fault characteristics in the signal can be analyzed more carefully.

[0078] In one embodiment of the present invention, the base frequency is set The frequency range is usually expressed as: and ,in, It is the bandwidth determined according to the specific application requirements. The frequency band includes the fundamental frequency and its surrounding frequency components. If multiple frequency bands are required, the frequency range can be further subdivided into several sub-frequency bands, and each frequency band can be divided linearly or logarithmically. Perform spectral analysis on each intrinsic mode component to find its main frequency, use FFT to calculate the spectrum of each intrinsic mode component, find the frequency corresponding to the maximum amplitude position in the spectrum, and compare the main frequency of the intrinsic mode component with the set frequency band. If the main frequency falls within the set frequency band, the intrinsic mode component is classified into this frequency band. The basis for classification is: if The eigenmode components are classified into the frequency band; after the classification of the eigenmode components is completed, the frequency band to which each eigenmode component belongs is obtained.

[0079] In a preferred embodiment of the present invention, the above step 15 may include:

[0080] According to the divided frequency bands, the eigenmode components corresponding to the fundamental frequency components are found and deleted, and all the eigenmode components classified as the second frequency band components are deleted to obtain the first frequency band components, wherein the first frequency band components contain all high-frequency characteristic information in the current signal except the fundamental frequency and the low frequency band.

[0081] In an embodiment of the present invention, by deleting the baseband component and the second frequency band component, the complexity of the current signal can be simplified, which helps to reduce interference factors in the analysis process and improve the analysis efficiency and accuracy. The first frequency band component usually contains the rapidly changing part of the current signal. By extracting the first frequency band component, the high-frequency characteristics closely related to the fault can be highlighted, making the fault signal easier to identify and distinguish. By focusing on the analysis of the first frequency band component, the detection sensitivity of these specific faults can be improved, and earlier fault warning and more accurate fault location can be achieved. The baseband component and the second frequency band component may contain normal fluctuation information that is not related to the fault. Excluding the component from the analysis can reduce false alarms caused by normal fluctuations, thereby improving the reliability of fault detection. When processing a large amount of current signal data, deleting unnecessary baseband and second frequency band components can significantly reduce the demand for data storage and processing.

[0082] In a preferred embodiment of the present invention, the above step 16 may include:

[0083] Performing standardization on the characteristic components to obtain standardized characteristic components;

[0084] For each pair of normalized feature components, Calculate the similarity value between each pair of standardized feature components, where and are two feature vectors, representing the two data sets whose similarity needs to be calculated. Each vector contains Sample values; Indicates the number of samples; is an index, traversing from 1 to Each sample of and Represents the eigenvectors and the eigenvector Middle The value of samples; and The eigenvectors are and the eigenvector The mean of and The eigenvectors are and the eigenvector The standard deviation of Represents the calculation of eigenvector and the eigenvector The function of the cosine similarity between ;

[0085] Create a size of N × N The matrix of N is the number of eigencomponents;

[0086] Traverse all feature component pairs and fill the similarity values ​​into the corresponding positions of the matrix. Each element of the matrix ( i , j ) indicates the i The feature components and j The similarity between the feature components.

[0087] In an embodiment of the present invention, by analyzing the similarity between the first frequency band components of the normal load current and the series arc fault current, the characteristics of the fault current can be more accurately identified, which helps to accurately detect arc faults in complex current signals and improve the accuracy and reliability of fault identification. The similarity metric matrix provides a comprehensive view of the similarity between the first frequency band components of the normal current and the fault current, which helps to distinguish the fault current from different types of normal currents, thereby optimizing the performance of fault classification. Through the similarity metric values ​​in the matrix, the current characteristics associated with a specific fault mode can be identified. By carefully comparing the first frequency band components of the normal load current and the fault current, the situation of misidentifying the normal current as the fault current (false alarm) or misidentifying the fault current as the normal current (missed alarm) can be reduced. The similarity metric matrix helps to set a more accurate threshold to reduce the occurrence of such errors.

[0088] In one embodiment of the present invention, the number of feature components is determined and a size N × N Similarity matrix, for each eigenvector, calculate its mean and standard deviation, define a method to calculate cosine similarity, measure the similarity between two eigenvectors. Use a double loop to traverse all combinations of feature components. The outer loop selects the first feature component, and the inner loop selects the second feature component. For each pair of feature components: extract the relevant eigenvectors, calculate the mean and standard deviation of the two feature vectors, and calculate the cosine similarity between the two feature vectors using the previously defined method. After the traversal is completed, store the similarity values ​​calculated for each pair of feature components in the corresponding position of the similarity matrix. Each element in the matrix will represent the similarity between two feature components.

[0089] In a preferred embodiment of the present invention, the above step 17 may include:

[0090] According to the similarity measurement matrix, traverse each value in the similarity measurement matrix, according to each high-frequency intrinsic mode component of the normal load current signal, to obtain the similarity with all fault current components;

[0091] The similarities with all fault current components are sorted in descending order and screened to obtain the corresponding key components.

[0092] In an embodiment of the present invention, by screening out the high-frequency intrinsic mode components of the normal load current with the weakest similarity to the fault current components, the unique characteristics of the fault current can be clearly highlighted, which helps to accurately identify the fault signal in a complex signal background and improve the sensitivity of fault detection. In signal analysis, components unrelated to the fault may exist as interference, affecting the accuracy of fault identification. By screening out the normal current components with high similarity to the fault current components, the impact of these interference factors on fault identification can be reduced, and the accuracy of identification can be improved. When performing fault detection and identification, processing a large amount of data will consume a lot of computing resources. By screening out the components with the weakest similarity, the amount of data to be processed subsequently can be reduced, thereby improving computing efficiency and reducing system burden.

[0093] In a specific embodiment of the present invention, starting from the first row of the similarity metric matrix, traversing row by row, each row represents the similarity between a high-frequency modal component of a normal load current and all fault current components. In each row, the similarity values ​​are extracted one by one and stored in the similarity list of the corresponding high-frequency modal component of the normal load current. The similarity list of each high-frequency modal component of the normal load current is sorted using a sorting algorithm. The goal of sorting is to sort by the size of the similarity value, from the smallest (weakest similarity) to the largest (strongest similarity). After sorting, the beginning of each list will contain the similarity values ​​​​that are related to the high-frequency modal component of the normal load current. IMF The similarity values ​​of the fault current components with the weakest similarity are extracted from each sorted similarity list to obtain the corresponding key components.

[0094] In a preferred embodiment of the present invention, the above step 18 may include:

[0095] Setting a modal component similarity threshold, and comparing the value in the similarity measurement matrix with the modal component similarity threshold to obtain a comparison result;

[0096] The results are analyzed and compared to obtain the sensitive fault component signal caused by the series arc fault.

[0097] In an embodiment of the present invention, by setting a similarity threshold, the fault current component that is significantly different from the normal current component can be accurately identified, which helps to accurately extract the sensitive fault component signal caused by the series arc fault from the complex current signal, thereby improving the accuracy of fault identification. The setting of the similarity threshold enables the system to automatically screen out those current components that are closely related to the fault. The components usually contain rich fault feature information, which enhances the system's ability to extract fault features. Reasonable similarity threshold setting can reduce the possibility of misjudgment and missed judgment. Only when the similarity between the fault current component and the normal current component is lower than the set threshold will it be judged as a fault component, which helps to avoid misjudging normal fluctuations as faults or missing minor faults. By automatically comparing the values ​​in the similarity measurement matrix with the threshold, sensitive fault component signals can be quickly identified and the corresponding fault handling mechanism can be triggered.

[0098] In a specific embodiment of the present invention, based on experience or the results of previous experiments, a modal component similarity threshold is set, and the threshold is used to distinguish the similarity between the high-frequency modal component of the normal load current and the fault current component. The elements in the similarity measurement matrix are traversed one by one, and each element in the matrix (i.e., the similarity value) is compared with the set similarity threshold. If the similarity value is less than the threshold, the fault current component corresponding to the position is recorded, indicating that the similarity between the fault current component and the corresponding high-frequency modal component of the normal load current is weak, which may be a manifestation of the fault characteristics. For each fault current high-frequency component, the number of similarity values ​​less than the threshold is counted, and all fault current high-frequency components are sorted according to the number of similarity values ​​less than the threshold, and the fault current high-frequency component with the largest number is selected as the sensitive fault component signal caused by the series arc fault.

[0099] like Figure 2 As shown, an embodiment of the present invention further provides a low-voltage series arc fault component extraction system 20 for a power collection terminal, comprising:

[0100] The acquisition module 21 is used to acquire the load current signal and the series arc fault current signal; perform modal decomposition on the load current signal and the series arc fault current signal to obtain the different frequency intrinsic modal components of the original current signal; calculate the correlation coefficient between the load current signal and the series arc fault current signal and their respective intrinsic modal components according to the different frequency intrinsic modal components, and determine the fundamental frequency components of the load current signal and the series arc fault current signal according to the correlation coefficient;

[0101] The processing module 22 is used to divide the intrinsic mode components of the normal load current signal and the series arc fault current signal into frequency bands according to the fundamental frequency components to obtain divided frequency bands; extract the corresponding characteristic components according to the divided frequency bands; calculate the similarity between the characteristic components to construct a similarity measurement matrix; screen the corresponding key components according to the similarity measurement matrix; compare and analyze the key components with the preset similarity threshold to obtain the sensitive fault component signal caused by the series arc fault.

[0102] It should be noted that the system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0103] The embodiment of the present invention further provides a computing device, comprising: a processor, a memory storing a computer program, wherein when the computer program is executed by the processor, the method described above is executed. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0104] The embodiment of the present invention also provides a computer-readable storage medium storing instructions, which, when executed on a computer, enable the computer to execute the method described above. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0105] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for extracting low-voltage series arc fault components of a power collection terminal, characterized in that: The method comprises: Obtaining load current signal and series arc fault current signal; Perform modal decomposition on the load current signal and the series arc fault current signal to obtain the different frequency intrinsic modal components of the original current signal; According to the different frequency eigenmode components, the correlation coefficients between the load current signal and the series arc fault current signal and their respective eigenmode components are calculated, and the fundamental frequency components of the load current signal and the series arc fault current signal are determined according to the correlation coefficients; According to the fundamental frequency component, each intrinsic mode component of the normal load current signal and the series arc fault current signal is divided into frequency bands to obtain divided frequency bands; According to the divided frequency bands, the corresponding characteristic components are extracted; Calculate the similarity between feature components to construct a similarity measurement matrix; According to the similarity measurement matrix, the corresponding key components are screened; Compare and analyze the key component with the preset similarity threshold to obtain the sensitive fault component signal caused by the series arc fault; Perform modal decomposition on the load current signal and the series arc fault current signal to obtain the different frequency intrinsic modal components of the original current signal, including: pass The load current signal and the series arc fault current signal are decomposed to obtain the intrinsic mode components of the original current signal, where: is the total number of different modal components decomposed, It is The eigenmode components, is the residual component obtained by decomposition, is the intrinsic mode component of the original current signal; According to the intrinsic mode components of the original current signal, Transform to obtain the different frequency eigenmode components of the original current signal, where: is the index of the time domain sampling point, is the frequency, is the length of each short period, It is The eigenmode components are at the sampling point The value at is the index of the eigenmode component, It is The result of Fourier transform of a short period of time is It is the overlap point between short periods; is a complex exponential function, representing the basis function of frequency domain transformation; According to the different frequency eigenmode components, the correlation coefficients between the load current signal and the series arc fault current signal and their respective eigenmode components are calculated, and the fundamental frequency components of the load current signal and the series arc fault current signal are determined according to the correlation coefficients, including: pass Calculate the correlation coefficient of each intrinsic mode component of the load current signal and the series arc fault current signal ,in, is the original current signal, It is The eigenmode components, is the original current signal sequence length, is the mean of the original signal sequence, It is The mean of the sequence of eigenmode components, is the time index, is the index of the eigenmode component; The correlation coefficients of the intrinsic modal components of the load current signal and the series arc fault current signal are calculated. Sorting to obtain the fundamental frequency components of the load current signal and the series arc fault current signal; According to the fundamental frequency component, each intrinsic mode component of the normal load current signal and the series arc fault current signal is divided into frequency bands to obtain the divided frequency bands, including: With the fundamental frequency component as the center, a certain frequency range is extended to both sides to form a frequency band including the fundamental frequency; For each eigenmode component, it is classified into a corresponding frequency band according to its frequency characteristics to obtain a series of divided frequency bands, each of which contains a group of eigenmode components with similar frequency characteristics; Calculate the similarity between feature components to construct a similarity measurement matrix, including: Performing standardization on the characteristic components to obtain standardized characteristic components; For each pair of normalized feature components; pass Calculate the similarity value between each pair of standardized feature components, where and are two feature vectors, representing the two data sets whose similarity needs to be calculated. Each vector contains Sample values; Indicates the number of samples; is an index, traversing from 1 to Each sample of and Represents the eigenvectors and the eigenvector Middle The value of samples; and The eigenvectors are and the eigenvector The mean of and The eigenvectors are and the eigenvector The standard deviation of Represents the calculation of eigenvector and the eigenvector The function of the cosine similarity between ; Create a size of N × N The matrix of N is the number of eigencomponents; Traverse all feature component pairs and fill the similarity values ​​into the corresponding positions of the matrix. Each element of the matrix ( i , j ) indicates the i The feature components and j The similarity between the feature components; According to the similarity measurement matrix, the corresponding key components are screened, including: According to the similarity measurement matrix, traverse each value in the similarity measurement matrix, according to each high-frequency intrinsic mode component of the normal load current signal, to obtain the similarity with all fault current components; The similarities with all fault current components are sorted in descending order and screened to obtain the corresponding key components; The key components are compared and analyzed with the preset similarity threshold to obtain the sensitive fault component signals caused by the series arc fault, including: Setting a modal component similarity threshold, and comparing the value in the similarity measurement matrix with the modal component similarity threshold to obtain a comparison result; Analyze and compare the results to obtain the sensitive fault component signal caused by the series arc fault; Obtain load current signals and series arc fault current signals, including: Collecting normal load current signals and series arc fault current signals, and preprocessing the current signals to obtain preprocessed data; Extract frequency domain features from the preprocessed data, map the extracted feature values ​​to "nodes", and initialize the positions of the ants; Each ant uses the pheromone distribution and heuristic information of the nodes around its current position to Select the next node to move, where It's an ant k An optional set of neighboring nodes, and is the weight coefficient, is the pheromone concentration, indicating that the node i To Node j pheromones; and It is heuristic information; Pheromone concentration, indicating the node To Node pheromones; Represents ants Select the node to move from the current node probability; is an index variable that iterates over the ant The currently selectable set of neighboring nodes ; After each iteration, the pheromone concentration on the path is updated according to the effect of the feature combination found by the ants on distinguishing normal and fault currents, and the maximum number of iterations is reached to achieve the feature combination that can best distinguish normal load current from series arc fault current; According to the iteration result, the load current signal and the series arc fault current signal are obtained.

2. A low-voltage series arc fault component extraction system for a power collection terminal, characterized in that: Applied in the method as claimed in claim 1, comprising: An acquisition module is used to acquire a load current signal and a series arc fault current signal; perform modal decomposition on the load current signal and the series arc fault current signal to acquire different frequency intrinsic modal components of the original current signal; calculate the correlation coefficients between the load current signal and the series arc fault current signal and their respective intrinsic modal components according to the different frequency intrinsic modal components, and determine the fundamental frequency components of the load current signal and the series arc fault current signal according to the correlation coefficients; The processing module is used to divide the intrinsic mode components of the normal load current signal and the series arc fault current signal into frequency bands according to the fundamental frequency components to obtain the divided frequency bands; extract the corresponding characteristic components according to the divided frequency bands; calculate the similarity between the characteristic components to construct a similarity measurement matrix; screen the corresponding key components according to the similarity measurement matrix; compare and analyze the key components with the preset similarity threshold to obtain the sensitive fault component signal caused by the series arc fault.

3. A computing device, characterized in that: include: one or more processors; A storage device, used for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in claim 1.

4. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which implements the method according to claim 1 when executed by a processor.

Images