Arc fault identification method and system

Abstract

The embodiment of the invention discloses an arc fault identification method and system. The arc fault identification method is applied to the arc fault identification system, and the arc fault identification system comprises a plurality of sampling gears. The arc fault identification method comprises the following steps: collecting a current signal under each sampling gear, and determining signal-to-noise ratio data of each current signal; determining a target sampling gear according to the signal-to-noise ratio data of each current signal, and adjusting the sampling gear to be the target sampling gear; collecting a target current signal under the target sampling gear, and determining the current level of the target current signal and the overall signal-to-noise ratio level; and determining an arc analysis model according to the current level of the target current signal and the overall signal-to-noise ratio level, and inputting the target current signal to the arc analysis model to improve the accuracy of arc fault recognition.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of electrical safety protection technology, and in particular to an arc fault identification method and system. Background Art

[0002] In DC systems, the transient high temperatures and electromagnetic interference generated by arcs can cause electrical fires or equipment damage. Traditional arc detection methods often use hardware architectures with fixed sampling rates, analyzing the time-frequency characteristics of current signals (such as high-frequency noise and signal mutations) to identify faults. However, these methods can result in false positives and false negatives, resulting in low accuracy. Summary of the Invention

[0003] Embodiments of the present invention provide an arc fault identification method and system to improve the accuracy of arc fault identification.

[0004] In a first aspect, an embodiment of the present invention provides an arc fault identification method, which is applied to an arc fault identification system, wherein the arc fault identification system includes multiple sampling gears;

[0005] The arc fault identification method comprises:

[0006] Collecting the current signal at each sampling gear, and determining the signal-to-noise ratio data of each current signal;

[0007] Determining a target sampling gear according to the signal-to-noise ratio data of each current signal, and adjusting the sampling gear to the target sampling gear;

[0008] collecting a target current signal at the target sampling gear, and determining a current level and an overall signal-to-noise ratio level of the target current signal;

[0009] An arc analysis model is determined based on the current level and the overall signal-to-noise ratio level of the target current signal, and the target current signal is input into the arc analysis model to identify arc faults.

[0010] Optionally, the signal-to-noise ratio data includes an overall signal-to-noise ratio and an arc-sensitive frequency band signal-to-noise ratio;

[0011] Determining the signal-to-noise ratio data of each current signal includes:

[0012] determining a current level of each of the current signals according to the amplitude of the current signal;

[0013] According to the current level of each current signal, obtaining a sensitive frequency band weight coefficient and a non-sensitive frequency band weight coefficient corresponding to each current signal;

[0014] Calculating the overall signal-to-noise ratio of each current signal according to each current signal and its corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient;

[0015] The arc-sensitive frequency band signal-to-noise ratio of each current signal is calculated according to the arc-sensitive frequency band signal of each current signal.

[0016] Optionally, the step of determining a target sampling gear position includes:

[0017] Screening the current signal corresponding to the largest signal-to-noise ratio in the arc-sensitive frequency band and the current signal whose difference from the largest signal-to-noise ratio in the arc-sensitive frequency band is within a preset range;

[0018] If the number of the screened current signals is 1, the sampling gear corresponding to the current signal is the target sampling gear;

[0019] If the number of the screened current signals is greater than or equal to 2, the target sampling gear is determined according to the overall signal-to-noise ratio of each of the screened current signals.

[0020] Optionally, determining the target sampling gear according to the overall signal-to-noise ratio of each of the screened current signals includes:

[0021] The overall signal-to-noise ratio of each of the screened current signals is compared, and the sampling gear corresponding to the current signal corresponding to the largest overall signal-to-noise ratio is used as the target sampling gear.

[0022] Optionally, the step of determining the current level of the target current signal includes:

[0023] determining the amplitude of the target current signal;

[0024] If the amplitude of the target current signal is less than the first current threshold, the current level is a low current level;

[0025] If the amplitude of the target current signal is greater than or equal to the first current threshold, and the amplitude of the target current signal is less than or equal to the second current threshold, then the current level is a medium current level;

[0026] If the amplitude of the target current signal is greater than the second current threshold, the current level is a high current level.

[0027] Optionally, the step of determining an overall signal-to-noise ratio level of the target current signal includes:

[0028] According to the current level, obtain the corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient;

[0029] Calculating an overall signal-to-noise ratio of the target current signal according to the target current signal and its corresponding sensitive frequency band weight coefficient and the non-sensitive frequency band weight coefficient;

[0030] If the overall signal-to-noise ratio is greater than a first signal-to-noise ratio threshold, the overall signal-to-noise ratio level is a strong signal-to-noise ratio;

[0031] If the overall signal-to-noise ratio is less than or equal to a first signal-to-noise ratio threshold, and the overall signal-to-noise ratio is greater than or equal to a second signal-to-noise ratio threshold, the current level is a medium signal-to-noise ratio;

[0032] If the overall signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the overall signal-to-noise ratio level is a weak signal-to-noise ratio.

[0033] Optionally, the arc analysis model includes a lightweight model and a deep convolutional neural network model;

[0034] The step of determining the arc analysis model includes:

[0035] If the overall signal-to-noise ratio level of the target current signal is the strong signal-to-noise ratio, the arc analysis model is the lightweight model;

[0036] If the overall signal-to-noise ratio level of the target current signal is the weak signal-to-noise ratio, the arc analysis model is the deep convolutional neural network model;

[0037] If the overall signal-to-noise ratio level of the target current signal is the medium signal-to-noise ratio and the current level is the small current level or the medium current level, the arc analysis model is the lightweight model;

[0038] If the overall signal-to-noise ratio level of the target current signal is the medium signal-to-noise ratio and the current level is the high current level, the arc analysis model is the deep convolutional neural network model.

[0039] Optionally, the arc fault identification method further includes:

[0040] collecting the peak and valley values ​​of the current signal in real time;

[0041] If the difference between the peak value and the valley value is greater than a first preset difference, adjusting the lower gear of the current sampling gear to the target sampling gear;

[0042] If the difference between the peak value and the valley value is smaller than a second preset difference, the current sampling gear is adjusted to a higher gear as the target sampling gear.

[0043] Optionally, the arc fault identification method further includes:

[0044] Calculating in real time the overall signal-to-noise ratio fluctuation of the target current signal;

[0045] If the overall signal-to-noise ratio fluctuation value is within the hysteresis band interval, the target sampling gear is not adjusted;

[0046] If the overall signal-to-noise ratio fluctuation value is not within the hysteresis band interval, the process returns to the step of collecting the current signal at each sampling gear and determining the signal-to-noise ratio data of each current signal.

[0047] In a first aspect, an embodiment of the present invention provides an arc fault identification system, which is used in the arc fault identification method proposed in any embodiment of the present invention. The arc fault identification system includes a variable sampling multiple circuit, an amplification circuit, a low-pass circuit, and a control chip;

[0048] The variable sampling multiple circuit is connected to the amplifier circuit, the amplifier circuit is connected to the low-pass circuit, the low-pass circuit is connected to the control chip, and the control chip is connected to the variable sampling multiple circuit;

[0049] Wherein, the variable sampling multiple circuit includes a multi-way switch sampling resistor circuit or a sampling variable gain amplifier.

[0050] The embodiment of the present invention collects current signals at each sampling level and determines the signal-to-noise ratio data of each current signal. Based on the signal-to-noise ratio data of each current signal, a target sampling level corresponding to the current signal that best reflects the detailed characteristics of the original current signal is determined, and the sampling level is adjusted to the target sampling level. A target current signal is collected at the target sampling level, and the current level and overall signal-to-noise ratio level of the target current signal are determined. Based on the current level and overall signal-to-noise ratio level of the target current signal, an appropriate arc analysis model is adapted, and the target current signal is input into the arc analysis model to improve the accuracy of arc fault identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 A multi-way switch sampling resistor circuit provided by an embodiment of the present invention;

[0053] Figure 2 A sampling variable gain amplifier provided by an embodiment of the present invention;

[0054] Figure 3 A schematic diagram of a flow chart of an arc fault identification method provided by an embodiment of the present invention;

[0055] Figure 4 A diagram showing arcing frequency domain data and non-arcing frequency domain data of a current signal with a low current level provided by an embodiment of the present invention;

[0056] Figure 5 Arcing frequency domain data and non-arcing frequency domain data of a current signal with a large current level provided by an embodiment of the present invention;

[0057] Figure 6 A schematic flow chart of steps for determining signal-to-noise ratio data of each current signal provided by an embodiment of the present invention;

[0058] Figure 7 A time domain waveform of a relatively small sampling resistor provided by an embodiment of the present invention;

[0059] Figure 8 The time domain waveform of a larger sampling resistor provided by an embodiment of the present invention;

[0060] Figure 9 A signal-to-noise ratio curve diagram at different current levels under different sampling strategies provided by an embodiment of the present invention;

[0061] Figure 10 A line graph of false alarm rate and missed alarm rate using fixed resistance value sampling provided by an embodiment of the present invention;

[0062] Figure 11 A line graph showing the accuracy of sampling using a fixed resistance value provided by an embodiment of the present invention;

[0063] Figure 12 A line graph of false alarm rate and missed alarm rate using variable resistor sampling provided by an embodiment of the present invention;

[0064] Figure 13 A line graph showing the accuracy of sampling using a variable resistor provided by an embodiment of the present invention.

[0065] Figure 14 A schematic structural diagram of an arc fault identification system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0066] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0067] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0068] An embodiment of the present invention provides a flow chart of an arc fault identification method. The arc fault identification method is applied to an arc fault identification system. The arc fault identification system includes multiple sampling gears.

[0069] Figure 1 A multi-way switch sampling resistor circuit is provided in an embodiment of the present invention. Figure 2 A sampling variable gain amplifier is provided in an embodiment of the present invention. Figure 1 As shown, the multi-way switch sampling resistor circuit switches different resistors to achieve different sampling gear switching. Figure 2 As shown, the sampling variable gain amplifier switches between different sampling gears by changing the sampling amplification gain. Therefore, this solution can amplify the resolution of arcing current and non-arcing current in the original current signal by configuring different sampling gears, and prevent sampling circuit saturation caused by fixed sampling gears (i.e., the collected current exceeds the range that the sampling circuit can handle, resulting in the collected current signal not accurately reflecting the original input current signal).

[0070] Figure 3 A flow chart of an arc fault identification method provided by an embodiment of the present invention is shown as follows: Figure 3 As shown, the arc fault identification method specifically includes:

[0071] S110 , collecting current signals at each sampling level, and determining signal-to-noise ratio data of each current signal.

[0072] Among them, the signal-to-noise ratio data of the current signal is a measure of the relative strength of the effective information and noise in the current signal. The signal-to-noise ratio data of the current signal includes the overall signal-to-noise ratio and the arc-sensitive frequency band signal-to-noise ratio. The overall signal-to-noise ratio is the relative strength of the effective information and noise in the entire current signal frequency band. The arc-sensitive frequency band signal-to-noise ratio is the relative strength of the effective information and noise in the arc-sensitive frequency band of the current signal. Different sampling gears have different amplification factors for the original current signal. The current signals with different amplification factors can reflect the detailed features of the original current signal to different degrees. The degree to which the current signal can reflect the detailed features of the original current signal can be determined by the signal-to-noise ratio data of the collected current signal. Therefore, the current signal under each sampling gear is collected, and the signal-to-noise ratio data of each current signal is determined by performing time domain analysis on the current signal under each gear, so that the appropriate sampling gear for collecting the original current signal can be determined by the signal-to-noise ratio data of each current signal.

[0073] S120 : Determine a target sampling gear according to the signal-to-noise ratio data of each current signal, and adjust the sampling gear to the target sampling gear.

[0074] The sampling gear corresponding to the current signal with the best signal-to-noise ratio is the target sampling gear. The target sampling gear determined based on the signal-to-noise ratio of each current signal is the optimal sampling gain for collecting the original current signal, which can enhance the distinction between arcing and non-arcing signals.

[0075] S130 : Acquire a target current signal at a target sampling level, and determine a current level and an overall signal-to-noise ratio level of the target current signal.

[0076] For example, Figure 4 The embodiment of the present invention provides a diagram of arcing frequency domain data and non-arcing frequency domain data of a current signal with a low current level, such as Figure 4 As shown in FIG, the distinction between the arcing frequency domain data and the non-arcing frequency domain data of the current signal of the small current level is more obvious. Figure 5 The embodiment of the present invention provides arcing frequency domain data and non-arcing frequency domain data of a current signal with a large current level, such as Figure 5As shown, the current signal of the large current level has a high background noise content in the current signal, so that the frequency domain component of the background noise in the current signal of the large current level masks the distinction between arcing and non-arcing, making the distinction between the arcing frequency domain data and the non-arcing frequency domain data of the current signal of the large current level unclear, thereby causing errors in arc fault identification and the risk of missed reports and false alarms. In addition, the overall signal-to-noise ratio of the current signal of the small current level is relatively large, and the overall signal-to-noise ratio of the current signal of the large current level is relatively small. Therefore, it is necessary to determine the current level and the overall signal-to-noise ratio level of the target current signal, so as to subsequently accurately adapt different arc analysis models according to the current level and the overall signal-to-noise ratio level of the target current signal to perform arc analysis on the current signal, so as to accurately identify arc faults.

[0077] S140 : Determine an arc analysis model according to the current level and the overall signal-to-noise ratio level of the target current signal, and input the target current signal into the arc analysis model to identify arc faults.

[0078] The arc analysis model consists of a lightweight model and a deep convolutional neural network model. The lightweight model is used to analyze arc faults on target current signals with a relatively high overall signal-to-noise ratio, while the deep convolutional neural network model is used for target current signals with a relatively low overall signal-to-noise ratio. For signals with a medium overall signal-to-noise ratio, further analysis based on the current level of the target current signal is required to accurately match the arc analysis model.

[0079] Specifically, the lightweight model can leverage the significant arcing characteristics in high signal-to-noise ratio environments to quickly identify arc faults. The lightweight model's reduced parameter count can further improve identification speed without compromising accuracy. Given a relatively low overall signal-to-noise ratio, where arcing and non-arcing characteristics are not significantly different, a deep convolutional neural network model is employed to further improve arc fault identification accuracy by enhancing its ability to resolve high-frequency details.

[0080] The embodiments of the present invention collect current signals at each sampling level and determine the signal-to-noise ratio data of each current signal. Based on the signal-to-noise ratio data of each current signal, a target sampling level corresponding to the current signal that best reflects the detailed characteristics of the original current signal is determined, and the sampling level is adjusted to the target sampling level. A target current signal is collected at the target sampling level, and the current level and overall signal-to-noise ratio level of the target current signal are determined. Based on the current level and overall signal-to-noise ratio level of the target current signal, an appropriate arc analysis model is adapted, and the target current signal is input into the arc analysis model to improve the accuracy of arc fault identification.

[0081] Based on the above embodiment, optionally, the signal-to-noise ratio data includes an overall signal-to-noise ratio and a signal-to-noise ratio in an arc-sensitive frequency band.

[0082] The overall signal-to-noise ratio is the relative strength of effective information and noise in the entire current signal frequency band. The arc-sensitive frequency band signal-to-noise ratio is the relative strength of effective information and noise in the arc-sensitive frequency band of the current signal.

[0083] Based on the above embodiment, optionally, Figure 6 A flow chart of the steps for determining the signal-to-noise ratio data of each current signal provided by an embodiment of the present invention. Figure 6 As shown in FIG, the steps for determining the signal-to-noise ratio data of each current signal are described:

[0084] S210 : Determine the current level of each current signal according to the amplitude of the current signal.

[0085] The magnitude of the current signal can be determined based on the amplitude of the current signal, and the current level of the current signal can be determined based on the magnitude of the current signal. The current level of the current signal includes a low current level and a high current level. When the amplitude of the current signal is less than a preset value (e.g., 10A), the current level of the current signal can be determined to be a low current level. When the amplitude of the current signal is greater than a preset value (e.g., 50A), the current level of the current signal can be determined to be a high current level.

[0086] S220 . Obtain, according to the current level of each current signal, a sensitive frequency band weight coefficient and a non-sensitive frequency band weight coefficient corresponding to each current signal.

[0087] Among them, the weight coefficient of the sensitive frequency band corresponding to each current signal is greater than the weight coefficient of the non-sensitive frequency band. When the current signal is at a low current level, the distinction of its arc sensitive frequency band is more obvious. At this time, the weight coefficient of the sensitive frequency band corresponding to the current signal is smaller (for example, 60%, which is smaller than the weight coefficient of the sensitive frequency band of the current signal at a high current level), and the weight coefficient of the non-sensitive frequency band is larger (for example, 20%, which is larger than the weight coefficient of the non-sensitive frequency band of the current signal at a high current level). As a result, the signal-to-noise ratio of the arc sensitive frequency band and the overall signal-to-noise ratio can be relatively high.

[0088] When the current signal is at a high current level, the distinction between the arc sensitive frequency bands is not obvious. At this time, the weight coefficient of the sensitive frequency band corresponding to the current signal is larger (for example, 80%, which is larger than the weight coefficient of the sensitive frequency band of the current signal at a small current level), and the weight coefficient of the non-sensitive frequency band is smaller (for example, 10%, which is smaller than the weight coefficient of the non-sensitive frequency band of the current signal at a small current level). This is to highlight the weight of the arc sensitive frequency band signal-to-noise ratio in the overall signal-to-noise ratio, better distinguish arcing signals from non-arcing signals, and reduce the missed alarm rate and false alarm rate at high current levels.

[0089] In addition, the sensitive frequency band weight coefficients and the non-sensitive frequency band weight coefficients corresponding to current signals of different current levels are preset and can be set according to actual conditions.

[0090] S230 , calculating the overall signal-to-noise ratio of each current signal according to each current signal and its corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient.

[0091] Among them, the overall signal-to-noise ratio = sensitive frequency band weight coefficient * arc sensitive frequency band signal-to-noise ratio + non-sensitive frequency band weight coefficient * non-arc sensitive frequency band signal-to-noise ratio.

[0092] For example, according to Figure 4 It can be seen that the frequency domain differentiation between arcing current and non-arcing current is mainly concentrated in the frequency range of 5-30kHz. Therefore, this frequency range is set as the arc-sensitive frequency range, and the sensitive frequency band weight coefficient of this frequency range is set to a1. The remaining frequency ranges are set as non-arcing sensitive frequency ranges (0-5kHz, 30kHz-100kHz), and the non-sensitive frequency band weight coefficient of this frequency range is set to a2. The overall signal-to-noise ratio of the current signal = a1 * arc-sensitive frequency band signal-to-noise ratio + non-sensitive frequency band signal-to-noise ratio a2 * non-arcing sensitive frequency band signal-to-noise ratio.

[0093] S240 , calculating the arc-sensitive frequency band signal-to-noise ratio of each current signal according to the arc-sensitive frequency band signal of each current signal.

[0094] The signal-to-noise ratio of the arc-sensitive frequency band = the signal power of the arc-sensitive frequency band / the noise power of the arc-sensitive frequency band. Similarly, the signal-to-noise ratio of the non-arc-sensitive frequency band = the signal power of the non-arc-sensitive frequency band / the noise power of the non-arc-sensitive frequency band.

[0095] Based on the above embodiment, optionally, the steps of determining the target sampling gear are described:

[0096] The current signal corresponding to the maximum arc-sensitive frequency band signal-to-noise ratio and the current signal whose difference with the maximum arc-sensitive frequency band signal-to-noise ratio is within a preset range are screened.

[0097] If the number of the screened current signals is 1, the sampling gear corresponding to the current signal is the target sampling gear.

[0098] If the number of the screened current signals is greater than or equal to 2, the target sampling gear is determined according to the overall signal-to-noise ratio of each screened current signal.

[0099] Specifically, the overall signal-to-noise ratio of each screened current signal is compared, and the sampling gear corresponding to the current signal corresponding to the largest overall signal-to-noise ratio is the target sampling gear.

[0100] Based on the above embodiment, optionally, the step of determining the current level of the target current signal is described:

[0101] Determine the amplitude of the target current signal.

[0102] The amplitude of the target current signal can reflect the size of the target current signal, and the size of the target current signal can be determined by the size of the target current.

[0103] If the amplitude of the target current signal is less than the first current threshold, the current level is a low current level.

[0104] The low current level may indicate that the current range of the target current signal is low current, and the first current threshold is a pre-set value. For example, the first current threshold is 10A. If the amplitude of the target current signal is less than 10A, the current level may be determined to be low current level, i.e., the target current signal is low current.

[0105] If the amplitude of the target current signal is greater than or equal to the first current threshold, and the amplitude of the target current signal is less than or equal to the second current threshold, the current level is a medium current level.

[0106] The medium current level can indicate that the current range of the target current signal is medium current, and the second current threshold is a pre-set value. For example, the second current threshold is 50A. If the amplitude of the target current signal is greater than 10A and less than 50A, the current level can be determined to be medium current level, that is, the target current signal is medium current.

[0107] If the amplitude of the target current signal is greater than the second current threshold, the current level is a high current level.

[0108] The high current level may indicate that the current range of the target current signal is high current, and the second current threshold is a pre-set value. For example, the second current threshold is 50A. If the amplitude of the target current signal is greater than 50A, the current level may be determined to be high current level, i.e., the target current signal is high current.

[0109] In the above embodiment, optionally, the step of determining the overall signal-to-noise ratio level of the target current signal includes:

[0110] According to the current level, the corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient are obtained.

[0111] Among them, the sensitive frequency band weight coefficients and the non-sensitive frequency band weight coefficients corresponding to the current signals of different current levels are preset and can be set according to actual conditions, which will not be described in detail here.

[0112] The overall signal-to-noise ratio of the target current signal is calculated based on the target current signal and its corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient.

[0113] Among them, the overall signal-to-noise ratio = sensitive frequency band weight coefficient * arc sensitive frequency band signal-to-noise ratio + non-sensitive frequency band weight coefficient * non-arc sensitive frequency band signal-to-noise ratio.

[0114] If the overall signal-to-noise ratio is greater than a first signal-to-noise ratio threshold, the overall signal-to-noise ratio level is a strong signal-to-noise ratio;

[0115] If the overall signal-to-noise ratio is less than or equal to a first signal-to-noise ratio threshold, and the overall signal-to-noise ratio is greater than or equal to a second signal-to-noise ratio threshold, then the current level is a medium signal-to-noise ratio;

[0116] If the overall signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the overall signal-to-noise ratio level is a weak signal-to-noise ratio.

[0117] The first signal-to-noise ratio threshold and the second signal-to-noise ratio threshold are preset according to actual conditions.

[0118] Based on the above embodiment, optionally, the arc analysis model includes a lightweight model and a deep convolutional neural network model.

[0119] The lightweight model can leverage the significant differentiation of arcing characteristics in high signal-to-noise ratio environments to quickly identify arc faults. The lightweight model, with its low parameter count, can further improve identification speed without compromising accuracy. High-current target current signals have relatively low signal-to-noise ratios, and the differences between arcing and non-arcing characteristics are not obvious. Using a deep convolutional neural network model, by enhancing the ability to analyze high-frequency details, can further improve the accuracy of arc fault identification.

[0120] Based on the above embodiment, optionally, the steps of determining the arc analysis model are described:

[0121] If the overall signal-to-noise ratio level of the target current signal is a strong signal-to-noise ratio, the arc analysis model is a lightweight model.

[0122] Among them, the overall signal-to-noise ratio level of the target current signal is a strong signal-to-noise ratio, indicating that the overall signal-to-noise ratio of the target current signal is relatively large, and the distinction between the arcing frequency domain data and the non-arcing frequency domain data of the current signal is relatively obvious. Therefore, the lightweight model can be used to quickly identify arc faults.

[0123] If the overall signal-to-noise ratio level of the target current signal is a weak signal-to-noise ratio, the arc analysis model is a deep convolutional neural network model.

[0124] Among them, the overall signal-to-noise ratio level of the target current signal is a weak signal-to-noise ratio, indicating that the overall signal-to-noise ratio of the target current signal is relatively small, and the distinction between the arcing frequency domain data and the non-arcing frequency domain data of the target current signal is not obvious. A deep convolutional neural network model is used to accurately identify arc faults by enhancing the high-frequency detail analysis capability.

[0125] In addition, the overall signal-to-noise ratio level of the target current signal is medium, and it is impossible to determine whether the current target current signal's arcing frequency domain data and non-arcing frequency domain data are clearly distinguishable. Therefore, it is necessary to further determine whether the target current signal's arcing frequency domain data and non-arcing frequency domain data are clearly distinguishable through the current level, and then adapt an appropriate arc analysis model. When the overall signal-to-noise ratio level of the target current signal is medium, the steps for determining the arc analysis model specifically include:

[0126] If the overall signal-to-noise ratio level of the target current signal is a medium signal-to-noise ratio and the current level is a small current level or a medium current level, the arc analysis model is a lightweight model.

[0127] Among them, when the overall signal-to-noise ratio level of the target current signal is medium, the current level is small current level or medium current level, indicating that the distinction between the arcing frequency domain data and the non-arcing frequency domain data of the current signal is relatively obvious. At this time, the lightweight model can be used to quickly identify arc faults.

[0128] If the overall signal-to-noise ratio level of the target current signal is a medium signal-to-noise ratio and the current level is a high current level, the arc analysis model is a deep convolutional neural network model.

[0129] Among them, when the overall signal-to-noise ratio level of the target current signal is medium, the current level is a large current level, indicating that the distinction between the arcing frequency domain data and the non-arcing frequency domain data of the target current signal is not obvious. At this time, a deep convolutional neural network model is needed to accurately identify arc faults by enhancing the high-frequency detail analysis capability.

[0130] Based on the above embodiment, optionally, the arc fault identification method further includes:

[0131] Real-time acquisition of peak and valley values ​​of current signals;

[0132] If the difference between the peak value and the valley value is greater than the first preset difference, adjust the current sampling gear to a lower gear as the target sampling gear;

[0133] If the difference between the peak value and the valley value is smaller than the second preset difference, the current sampling gear is adjusted to a higher gear as the target sampling gear.

[0134] The above-mentioned method of adjusting the target sampling gear is a hysteresis switching logic, which can prevent frequent switching of the sampling gear during the sampling process of the current signal. The sampling range boundary is set. The upper limit of the difference between the peak and valley values ​​of the sampled current signal can be set to 0.8 times the analog-to-digital converter range (a first preset difference), and the lower limit can be set to 0.3 times the analog-to-digital converter range (a second preset difference). Therefore, when the difference between the peak and valley values ​​of the current signal reaches 0.9 times the analog-to-digital converter range, the lower sampling gear is forced to be selected. When the difference between the peak and valley values ​​of the current signal is less than 0.2 times the analog-to-digital converter range, the higher sampling gear is selected, ensuring that the collected current signal does not overshoot and can have a large gain.

[0135] Based on the above embodiment, optionally, the arc fault identification method further includes:

[0136] Calculate the overall signal-to-noise ratio fluctuation of the target current signal in real time;

[0137] If the overall signal-to-noise ratio fluctuation value is within the hysteresis band, the target sampling gear will not be adjusted;

[0138] If the overall signal-to-noise ratio fluctuation value is not within the hysteresis band range, the process returns to the step of collecting current signals at each sampling gear and determining the signal-to-noise ratio data of each current signal.

[0139] The above-mentioned signal-to-noise ratio hysteresis control sets the signal-to-noise ratio fluctuation range, which can achieve switching of the sampling gear only when the overall signal-to-noise ratio of the current signal drops significantly, thereby improving the robustness of the system.

[0140] The variable sampling position in this embodiment uses a hardware architecture consisting of a multi-way switch and multiple precision resistors. For example, it consists of eight precision resistors of different resistance values ​​and a multi-way switch. This solution achieves optimized signal acquisition over a wide range by dynamically switching sampling resistors (sampling positions) of different resistance values. In different current acquisition scenarios, it can automatically match the optimal sampling resistor value (sampling position), ensuring that the current signal can be effectively sampled and its signal-to-noise ratio is optimized. Figure 7 The time domain waveform of a smaller sampling resistor provided by an embodiment of the present invention is: Figure 8 This is the time domain waveform of a larger sampling resistor provided by an embodiment of the present invention. Figure 7 and Figure 8 It can be seen that when a larger sampling resistor is selected, the time-domain waveform amplitude of the arcing signal is significantly improved, effectively expanding the system's dynamic measurement range while maintaining signal integrity. In contrast, when a fixed small resistor is used, the signal amplitude is compressed to less than 20% of the full scale, resulting in the loss of key characteristic information.

[0141] Figure 9 The embodiment of the present invention provides a signal-to-noise ratio curve diagram at different current levels under different sampling strategies, such as Figure 9 As shown, in the fixed resistor sampling mode, as the current level increases from 0A to 30A, the signal-to-noise ratio shows a decreasing trend (from 40dB to 27dB). However, after adopting the variable resistor sampling method of the present invention, the signal-to-noise ratio at all current levels remains stable above 35dB. In particular, in the high current range of 15A-30A, the signal-to-noise ratio is improved by more than 8dB, which improves the frequency domain differentiation between arcing and non-arcing signals by 30%, providing a clearer feature space for subsequent pattern recognition.

[0142] Figure 10 A line graph of false alarm rate and missed alarm rate using fixed resistance value sampling provided by an embodiment of the present invention is provided. Figure 11 A line graph showing the accuracy of sampling using a fixed resistance value provided by an embodiment of the present invention. Figure 12 A line graph of false alarm rate and missed alarm rate using variable resistor sampling provided by an embodiment of the present invention, Figure 13 A line graph showing the accuracy of sampling using a variable resistor provided by an embodiment of the present invention. Figure 10 and Figure 11 The statistical data further confirms the limitations of fixed resistor sampling. When the current exceeds 12A, the false alarm rate rises sharply from 0.1% to 1.2%, and the missed alarm rate increases to 1%, resulting in a significant decrease in overall accuracy. Figure 12 and Figure 13 Statistical data further verified that the use of variable resistor sampling maintained excellent performance across the full range. The false alarm rate and missed alarm rate at each current level were strictly controlled within 0.1%, and the accuracy rate was stably maintained at above 99.85%.

[0143] An embodiment of the present invention further provides an arc fault identification system, which is used to execute the arc fault identification method provided by any embodiment of the present invention. Figure 14 A schematic diagram of the structure of an arc fault identification system provided by an embodiment of the present invention is shown in FIG. Figure 14 As shown, the arc fault identification system includes a variable sampling multiple circuit 10, an amplifying circuit 20, a low-pass circuit 30 and a control chip 40;

[0144] The variable sampling multiple circuit 10 is connected to the amplifier circuit 20, the amplifier circuit 20 is connected to the low-pass circuit 30, the low-pass circuit 30 is connected to the control chip 40, and the control chip 40 is connected to the variable sampling multiple circuit 10;

[0145] The variable sampling multiple circuit 10 includes a multi-way switch sampling resistor circuit or a sampling variable gain amplifier.

[0146] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0147] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for identifying an arc fault, characterized in that: Applied to an arc fault identification system, the arc fault identification system includes multiple sampling gears; The arc fault identification method comprises: Collecting the current signal at each sampling gear, and determining the signal-to-noise ratio data of each current signal; Determining a target sampling gear according to the signal-to-noise ratio data of each current signal, and adjusting the sampling gear to the target sampling gear; collecting a target current signal at the target sampling gear, and determining a current level and an overall signal-to-noise ratio level of the target current signal; An arc analysis model is determined based on the current level and the overall signal-to-noise ratio level of the target current signal, and the target current signal is input into the arc analysis model to identify arc faults.

2. The arc fault identification method according to claim 1, characterized in that: The signal-to-noise ratio data includes the overall signal-to-noise ratio and the signal-to-noise ratio of the arc-sensitive frequency band; Determining the signal-to-noise ratio data of each current signal includes: determining a current level of each of the current signals according to the amplitude of the current signal; According to the current level of each current signal, obtaining a sensitive frequency band weight coefficient and a non-sensitive frequency band weight coefficient corresponding to each current signal; Calculating the overall signal-to-noise ratio of each current signal according to each current signal and its corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient; The arc-sensitive frequency band signal-to-noise ratio of each current signal is calculated according to the arc-sensitive frequency band signal of each current signal.

3. The arc fault identification method according to claim 2, characterized in that: The step of determining the target sampling gear position includes: Screening the current signal corresponding to the largest signal-to-noise ratio in the arc-sensitive frequency band and the current signal whose difference from the largest signal-to-noise ratio in the arc-sensitive frequency band is within a preset range; If the number of the screened current signals is 1, the sampling gear corresponding to the current signal is the target sampling gear; If the number of the screened current signals is greater than or equal to 2, the target sampling gear is determined according to the overall signal-to-noise ratio of each of the screened current signals.

4. The arc fault identification method according to claim 3, characterized in that: Determining the target sampling gear according to the overall signal-to-noise ratio of each of the screened current signals includes: The overall signal-to-noise ratio of each of the screened current signals is compared, and the sampling gear corresponding to the current signal corresponding to the largest overall signal-to-noise ratio is used as the target sampling gear.

5. The arc fault identification method according to claim 1, characterized in that: The step of determining the current level of the target current signal includes: determining the amplitude of the target current signal; If the amplitude of the target current signal is less than the first current threshold, the current level is a low current level; If the amplitude of the target current signal is greater than or equal to the first current threshold, and the amplitude of the target current signal is less than or equal to the second current threshold, then the current level is a medium current level; If the amplitude of the target current signal is greater than the second current threshold, the current level is a high current level.

6. The arc fault identification method according to claim 5, characterized in that: The step of determining the overall signal-to-noise ratio level of the target current signal comprises: According to the current level, obtain the corresponding sensitive frequency band weight coefficient and non-sensitive frequency band weight coefficient; Calculating an overall signal-to-noise ratio of the target current signal according to the target current signal and its corresponding sensitive frequency band weight coefficient and the non-sensitive frequency band weight coefficient; If the overall signal-to-noise ratio is greater than a first signal-to-noise ratio threshold, the overall signal-to-noise ratio level is a strong signal-to-noise ratio; If the overall signal-to-noise ratio is less than or equal to a first signal-to-noise ratio threshold, and the overall signal-to-noise ratio is greater than or equal to a second signal-to-noise ratio threshold, the current level is a medium signal-to-noise ratio; If the overall signal-to-noise ratio is less than the second signal-to-noise ratio threshold, the overall signal-to-noise ratio level is a weak signal-to-noise ratio.

7. The arc fault identification method according to claim 6, characterized in that: The arc analysis model includes a lightweight model and a deep convolutional neural network model; The step of determining the arc analysis model includes: If the overall signal-to-noise ratio level of the target current signal is the strong signal-to-noise ratio, the arc analysis model is the lightweight model; If the overall signal-to-noise ratio level of the target current signal is the weak signal-to-noise ratio, the arc analysis model is the deep convolutional neural network model; If the overall signal-to-noise ratio level of the target current signal is the medium signal-to-noise ratio and the current level is the small current level or the medium current level, the arc analysis model is the lightweight model; If the overall signal-to-noise ratio level of the target current signal is the medium signal-to-noise ratio and the current level is the high current level, the arc analysis model is the deep convolutional neural network model.

8. The arc fault identification method according to claim 1, characterized in that: Also includes: collecting the peak and valley values ​​of the current signal in real time; If the difference between the peak value and the valley value is greater than a first preset difference, adjusting the lower gear of the current sampling gear to the target sampling gear; If the difference between the peak value and the valley value is smaller than a second preset difference, the current sampling gear is adjusted to a higher gear as the target sampling gear.

9. The arc fault identification method according to claim 1, characterized in that: Also includes: Calculating in real time the overall signal-to-noise ratio fluctuation of the target current signal; If the overall signal-to-noise ratio fluctuation value is within the hysteresis band interval, the target sampling gear is not adjusted; If the overall signal-to-noise ratio fluctuation value is not within the hysteresis band interval, the process returns to the step of collecting the current signal at each sampling gear and determining the signal-to-noise ratio data of each current signal.

10. An arc fault identification system, using the arc fault identification method according to any one of claims 1 to 9, characterized in that: It includes a variable sampling multiple circuit, an amplifier circuit, a low-pass circuit and a control chip; The variable sampling multiple circuit is connected to the amplifier circuit, the amplifier circuit is connected to the low-pass circuit, the low-pass circuit is connected to the control chip, and the control chip is connected to the variable sampling multiple circuit; Wherein, the variable sampling multiple circuit includes a multi-way switch sampling resistor circuit or a sampling variable gain amplifier.

Images