An Arc Fault Detection Method Based on Singular Spectrum Analysis

Abstract

The present invention discloses an arc fault detection method based on singular spectrum analysis, comprising: dividing a collected current signal into a plurality of power frequency cycles according to a time sequence, and constructing Hankel matrices based on the data sequences of the divided power frequency cycles respectively, calculating a corresponding covariance matrix based on the constructed Hankel matrix, eigendecomposing the covariance matrix to obtain the singular values ​​of the Hankel matrix, selecting the maximum eigenvalue and the minimum eigenvalue, and calculating the ratio of the maximum eigenvalue to the minimum eigenvalue as a detection amount for judging the arc fault; the present invention proposes an arc fault detection method based on singular spectrum analysis, proposes a detection amount that is effective for all arc fault current signals, and accurately and effectively distinguishes arc fault current signals from normal current signals.

Description

Technical Field

[0001] The invention relates to the technical field of electrical equipment, and in particular to an arc fault detection method based on singular spectrum analysis. Background Art

[0002] The casualties and property losses caused by electrical fires are very prominent, especially electrical fires caused by arc faults, which have unclear detection characteristics and are difficult to detect. Therefore, the detection of arc faults has attracted widespread attention from researchers.

[0003] At present, arc fault detection is mostly completed by arc fault circuit breakers (AFDDs). Effective functional testing is the key to ensure that AFDDs can work properly. Both the American standard UL1699 and the Chinese standard GB / T 31143-2014 have detailed provisions on the rated parameters, performance requirements, and corresponding test contents of arc fault circuit breakers, and provide qualified standards under various test conditions.

[0004] Chinese invention patent application CN114034960A discloses a method for detecting arc faults based on weak arc characteristics. The method converts the collected current signal into time-domain and frequency-domain characteristics and then constructs a Hankel matrix. After extracting the arc characteristics through time-domain and frequency-domain conversion, the method separates the signal from the noise and performs arc fault detection. However, the method cannot directly process and detect the time-domain signal.

[0005] The "Arc Fault Detection Method and Device, Storage Medium and Electronic Equipment" disclosed in Chinese invention patent application CN116338402A also performs noise reduction processing on the collected current signal to improve the accuracy of arc fault detection. The technical solution adopted is to perform Fourier transform on the collected current signal, that is, convert the time domain and frequency domain characteristics, and then construct a Hankel matrix for denoising. It is impossible to directly process the time domain signal, and denoising will also filter out the high-frequency effective part.

[0006] Arc fault detection algorithms used in circuit breakers are the focus of current research. Researchers focus on how to use one or more effective time-frequency domain features to detect arc faults. Due to the diversity of loads and the complexity of actual circuits, the use of time domain features is prone to false alarms, and the characteristic frequency bands of arc faults will change with the change of application objects, resulting in low accuracy of these methods. Summary of the invention

[0007] In order to overcome the shortcomings of the above technologies, the present invention provides an arc fault detection method based on singular spectrum analysis, and innovatively proposes a detection quantity that is effective for all arc fault current signals. Based on the monitoring principle that the ratio of the maximum and minimum eigenvalues ​​of the arc fault current must be smaller than that of the normal current, a Hankel matrix is ​​constructed for the collected current signal, and the singular spectrum analysis in the signal adaptive decomposition is used to adaptively decompose the signal from low frequency to high frequency, and the energy of these frequency bands is quantified by the eigenvalues.

[0008] The technical solution adopted by the present invention to overcome its technical problems is: an arc fault detection method based on singular spectrum analysis proposed by the present invention includes: S1, dividing the collected current signal into a number of power frequency cycles in time sequence, and constructing Hankel matrices based on the data sequences of the divided power frequency cycles; S2, calculating the corresponding covariance matrix based on the constructed Hankel matrix; S3, performing eigendecomposition on the covariance matrix to obtain the singular values ​​of the Hankel matrix, and selecting the maximum eigenvalue and the minimum eigenvalue, and calculating the ratio of the maximum eigenvalue to the minimum eigenvalue as a detection amount for judging the arc fault.

[0009] Based on the monitoring principle that the ratio of the maximum and minimum eigenvalues ​​of arc fault current must be smaller than that of normal current, a detection quantity valid for all arc fault current signals is obtained through singular spectrum analysis in a pioneering way.

[0010] Furthermore, the method further includes S4, comparing the calculated detection value with a preset detection threshold value, and if the detection value of a plurality of consecutive power frequency cycles is less than the preset detection threshold value, it is determined that an arc fault occurs.

[0011] Furthermore, the S1 divides the collected current signal into several power frequency cycles in time sequence, and constructs a Hankel matrix based on the data sequence of the divided power frequency cycles, including: S11, collecting the current signal y of the circuit to be detected, and dividing the collected current signal into M power frequency cycles in time sequence; S12, constructing a corresponding data sequence based on N data points of each power frequency cycle; S13, setting the number of decomposition layers based on the data sequence; S14, constructing a Hankel matrix corresponding to each power frequency cycle based on the number of decomposition layers.

[0012] Furthermore, performing eigendecomposition on the covariance matrix to obtain the singular values ​​of the Hankel matrix specifically includes: performing eigendecomposition on the covariance to obtain eigenvalues ​​corresponding to the decomposition level, wherein the eigenvalue is the square of the singular value of the Hankel matrix.

[0013] Furthermore, the maximum eigenvalue represents the energy of the most fundamental frequency in the power frequency periodic signal, and the minimum eigenvalue represents the energy of the most high frequency in the power frequency periodic signal.

[0014] The beneficial effects of the present invention are:

[0015] 1. The arc fault detection quantity that is effective for all arc fault current signals is proposed in a pioneering way, that is, arc fault detection can be performed solely through the arc fault detection quantity, and the detection quantity can also be used as an important basis for arc detection in the multi-feature detection algorithm;

[0016] 2. Based on the actual collected arc fault current signal, the singular spectrum analysis in adaptive decomposition is used to adaptively decompose the current signal from low frequency to high frequency, effectively distinguishing the arc fault current signal from the normal current signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of a flow chart of an arc fault detection method based on singular spectrum analysis according to an embodiment of the present invention;

[0018] Figure 2 A first current signal collected by an embodiment of the present invention;

[0019] Figure 3 A second current signal collected by an embodiment of the present invention;

[0020] Figure 4 A third current signal collected by an embodiment of the present invention;

[0021] Figure 5 A fourth current signal collected by an embodiment of the present invention;

[0022] Figure 6 A fifth current signal collected by an embodiment of the present invention;

[0023] Figure 7 A sixth current signal collected by an embodiment of the present invention;

[0024] Figure 8 is the ratio of the maximum eigenvalue to the minimum eigenvalue obtained by analyzing the arc fault detection method based on singular spectrum analysis of the first current signal;

[0025] Fig. 9 is the ratio of the maximum eigenvalue to the minimum eigenvalue obtained by analyzing the arc fault detection method based on singular spectrum analysis of the second current signal;

[0026] Fig.10 is the ratio of the maximum eigenvalue to the minimum eigenvalue obtained by analyzing the arc fault detection method based on singular spectrum analysis of the third current signal;

[0027] Fig.11 is a ratio of a maximum eigenvalue to a minimum eigenvalue obtained by analyzing the fourth current signal using an arc fault detection method based on singular spectrum analysis;

[0028] Fig.12 is the ratio of the maximum eigenvalue to the minimum eigenvalue obtained by analyzing the arc fault detection method based on singular spectrum analysis of the fifth current signal;

[0029] Fig.13 It is the ratio of the maximum eigenvalue to the minimum eigenvalue obtained by analyzing the arc fault detection method based on singular spectrum analysis of the sixth current signal. DETAILED DESCRIPTION

[0030] Hankel matrix: A Hankel matrix is ​​a matrix in which the elements on each subdiagonal are equal.

[0031] In order to facilitate those skilled in the art to better understand the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The following is only exemplary and does not limit the protection scope of the present invention.

[0032] like Figure 1 As shown, a flow chart of an arc fault detection method based on singular spectrum analysis described in this embodiment is used to detect and analyze an arc fault current signal under a certain load to determine whether an arc fault occurs.

[0033] In one embodiment of the present invention, an arc fault generating device is connected to a load halogen lamp to generate an arc fault current signal, a data acquisition card and a Rogowski coil are used to record the current signal waveform at a sampling rate of 100K, and an arc fault detection method based on singular spectrum analysis is used to analyze it, which specifically includes the following steps.

[0034] S1, dividing the collected current signal into a number of power frequency periodic signals according to the time sequence, and constructing Hankel matrices based on the data sequences of the divided power frequency periodic signals respectively.

[0035] The arc fault current signal is divided into M power frequency periodic signals including N data points in time sequence, and a Hankel matrix is ​​constructed according to the divided power frequency periodic signals, which specifically includes the following steps.

[0036] S11, collecting the current signal y of the circuit to be detected, and dividing the collected current signal into M power frequency cycles in time sequence.

[0037] In one embodiment of the present invention, a signal of 1 second in length and containing an arc fault is extracted from the collected current signal waveform for analysis, and is divided into 50 power frequency cycles, y1, y2, ..., y 50 .

[0038] S12, construct the corresponding data sequence based on N data points of each power frequency cycle .

[0039] Taking N as 2000 as an example, a data point sequence is constructed for each power frequency cycle of N sampling data points .

[0040] S13, setting the decomposition level L based on the data sequence.

[0041] In one embodiment of the present invention, the number of decomposition levels L = 2. The number of decomposition levels takes different values ​​according to different application scenarios.

[0042] S14, based on the decomposition layer number L, the Hankel matrix Y is constructed as shown in formula (1).

[0043] (1).

[0044] In one embodiment of the present invention, the decomposition level L=2, so the Hankel matrix Y is as shown in formula (2).

[0045] (2).

[0046] S2, calculate the corresponding covariance matrix based on the constructed Hankel matrix.

[0047] According to the matrix , and calculate its covariance matrix as shown in formula (3).

[0048] (3).

[0049] S3, perform eigendecomposition on the covariance matrix to obtain the singular values ​​of the Hankel matrix, select the maximum eigenvalue and the minimum eigenvalue, and calculate the ratio of the maximum eigenvalue to the minimum eigenvalue as the detection amount for judging the arc fault.

[0050] S31, perform eigendecomposition on the covariance to obtain eigenvalues ​​corresponding to the decomposition level L, and select the maximum eigenvalue and the minimum eigenvalue.

[0051] The L eigenvalues ​​obtained by decomposing the covariance matrix are arranged in order from large to small as shown in formula (4).

[0052] (4).

[0053] in, It is also the Hankel matrix The squares of the singular values ​​of .

[0054] In one embodiment of the present invention, the covariance matrix X Performing eigendecomposition, we obtain two eigenvalues, which are arranged from small to large as shown in formula (5).

[0055] (5).

[0056] S32, calculating the ratio of the maximum eigenvalue to the minimum eigenvalue as the detection amount.

[0057] The ratio of the maximum characteristic value to the minimum characteristic value is calculated as formula (6). The amount of detection.

[0058] (6).

[0059] here Represents the energy of the most fundamental frequency part of the signal. Represents the energy of the highest frequency part. The typical feature of an arc fault signal is that the current value is slightly smaller than that of a normal signal, which means that the most fundamental frequency energy of the arc fault signal is smaller than that of a normal signal. Since the random fluctuation of an arc fault signal is generally more intense than that of a normal signal, its highest frequency energy is greater than that of a normal signal. This is the detection principle of the entire method.

[0060] In one embodiment of the present invention, the arc fault detection quantity is as shown in formula (7). Represents the energy of the fundamental frequency part of the signal. Represents the energy of the high frequency part.

[0061] (7).

[0062] S4, constructing a detection quantity sequence based on the calculated detection quantity, and comparing it with a preset detection threshold. If the detection quantity of several consecutive power frequency cycles is less than the preset detection threshold, it is determined that an arc fault occurs.

[0063] In order to reduce the probability of false alarm, the judgment criterion is In other cases, the fault is deemed not to have occurred.

[0064] in It is the detection threshold, that is, a fault can only be identified when the detection value of multiple consecutive power frequency cycles is less than the threshold.

[0065] like Figure 2-Figure 7 The six current signals collected are analyzed by the arc fault detection method based on singular spectrum analysis, so as to obtain the corresponding Figures 8 to 13 The ratio of the maximum and minimum eigenvalues ​​of each current signal shown in FIG. 50 power frequency cycles. Figures 1 to 3 The AFDD, the arc fault protection circuit breaker, breaks the circuit accurately and quickly. Figures 8 to 10 The difference between the fault signal and the normal signal can also be clearly seen. Figures 5 to 7The AFDD fails to disconnect quickly. For this example, if the preset detection threshold is 55000, arc fault detection can be performed quickly and accurately.

[0066] In a series arc fault that is difficult to detect, the fault current is generally slightly smaller than the normal current; although the characteristic frequency band of the arc fault signal is not very stable, it is a recognized fact that the high-frequency component of the arc fault signal is larger than the normal signal. The present invention proposes a detection quantity that is effective for all arc fault current signals. Using the singular spectrum analysis in the signal adaptive decomposition, the signal is adaptively decomposed from low frequency to high frequency, and an eigenvalue is used to quantify the energy of these frequency bands. The ratio of the maximum and minimum eigenvalues ​​of the arc fault current must be smaller than that of the normal current, which is the detection principle of the present invention.

[0067] It should be noted that: in other embodiments, the steps of the corresponding method are not necessarily performed in the order shown and described in this specification. In some other embodiments, the steps included in the method may be more or less than those described in this specification. In addition, a single step described in this specification may be decomposed into multiple steps for description in other embodiments; and multiple steps described in this specification may be combined into a single step for description in other embodiments.

Claims

1. An arc fault detection method based on singular spectrum analysis, characterized in that: include: S1, divide the collected current signal into several power frequency cycles according to the time sequence, and construct the Hankel matrix based on the data sequence of the divided power frequency cycles. S11, collecting a current signal y of the circuit to be detected, and dividing the collected current signal into M power frequency cycles in time sequence; S12, constructing a corresponding data sequence based on N data points of each power frequency cycle; S13, setting the number of decomposition levels based on the data sequence; S14, constructing a Hankel matrix corresponding to each power frequency period based on the number of decomposition levels; S2, calculate the corresponding covariance matrix based on the constructed Hankel matrix; S3, perform eigendecomposition on the covariance matrix to obtain the singular values ​​of the Hankel matrix, select the maximum eigenvalue and the minimum eigenvalue, and calculate the ratio of the maximum eigenvalue to the minimum eigenvalue as a detection quantity for judging the arc fault; S4, comparing the calculated detection value with a preset detection threshold, if the detection value of several consecutive power frequency cycles is less than the preset detection threshold, it is determined that an arc fault occurs.

2. The arc fault detection method based on singular spectrum analysis according to claim 1, characterized in that: The method of performing eigendecomposition on the covariance matrix to obtain the singular values ​​of the Hankel matrix specifically includes: The covariance is decomposed to obtain the eigenvalues ​​of the corresponding decomposition level, where the eigenvalues ​​are the squares of the singular values ​​of the Hankel matrix.

3. The arc fault detection method based on singular spectrum analysis according to claim 1, characterized in that: The maximum eigenvalue represents the energy of the most fundamental frequency in the power frequency periodic signal, and the minimum eigenvalue represents the energy of the most high frequency in the power frequency periodic signal.

Images