AC arc detection method and related device based on current and harmonic fluctuations

Abstract

The present application discloses an AC arc detection method and related device based on current and harmonic fluctuations. The method comprises: collecting first electrical signal data at a preset location in a target power grid and extracting harmonic features therefrom; concurrently, collecting second electrical signal data at the preset location and extracting current waveform features therefrom; determining a binary feature dataset and a fluctuation feature dataset based on the harmonic features and current waveform features; training a preset model based on the binary feature dataset and the fluctuation feature dataset to obtain a binary model and a fluctuation model; and performing arc fault detection on the electrical signal of a preset second time period at the preset location using the binary model and the fluctuation model to determine a target arc fault detection result at the preset location. By employing the present application, arc fault detection can be performed using a combined binary classification model and a fluctuation model, thereby improving the accuracy of arc fault detection.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to an AC arc detection method and related devices based on current and harmonic fluctuations. Background Art

[0002] With the rapid improvement of global electrification levels, social electricity demand is increasing day by day, but the proportion of electrical fires is also increasing. Most of the causes of electrical fires are arc faults. Therefore, arc detection is necessary during actual power operation.

[0003] Existing arc detection technologies typically use current threshold detection, which determines faults by setting a fixed current threshold, but cannot distinguish between normal startup and shutdown events and arc faults. Alternatively, low-frequency harmonic detection utilizes the low-frequency harmonic characteristics generated by arcs, but contains less information and has poor differentiation from many interfering electrical device signatures. Alternatively, waveform slope detection analyzes the rate of change of current to determine if an arc fault has occurred, but lacks sensitivity for low-power loads. Existing arc detection technologies suffer from high false alarm rates, slow response speeds, and poor anti-interference capabilities. Therefore, improving the accuracy of arc fault detection is a pressing issue. Summary of the Invention

[0004] The embodiments of the present application provide an AC arc detection method and related devices based on current and harmonic fluctuations, which improve the detection accuracy of fault arcs by jointly performing arc fault detection through a binary classification model and a fluctuation model.

[0005] In a first aspect, an embodiment of the present application provides an AC arc detection method based on current and harmonic fluctuations, which is applied to a control device of an arc detection system, wherein the arc detection system further includes: a first data acquisition device and a second data acquisition device; the method includes:

[0006] collecting first electrical signal data at a preset location in a target power grid using the first data acquisition device at a first sampling frequency within a preset first time period, and extracting harmonic features from the first electrical signal data; and simultaneously collecting second electrical signal data at the preset location at a second sampling frequency within the preset first time period using the second data acquisition device at a second sampling frequency, and extracting current waveform features from the second electrical signal data; the first sampling frequency being greater than the second sampling frequency;

[0007] Determine a binary feature data set according to the harmonic features and the current waveform features;

[0008] determining a fluctuation characteristic data set according to the harmonic characteristics and the current waveform characteristics;

[0009] The preset models are trained based on the binary classification feature data set and the fluctuation feature data set to obtain a binary classification model and a fluctuation model respectively;

[0010] Performing arc fault detection on the electrical signal of the preset second time period at the preset position using the binary classification model and the fluctuation model respectively, to obtain a first detection result and a second detection result;

[0011] A target arc fault detection result at the preset position is determined according to the first detection result and the second detection result.

[0012] In a second aspect, an embodiment of the present application provides an AC arc detection device based on current and harmonic fluctuations, which is applied to a control device of an arc detection system. The arc detection system further includes: a first data acquisition device and a second data acquisition device; the AC arc detection device based on current and harmonic fluctuations includes: a data acquisition module, a first data set processing module, a second data set processing module, a model training module, an arc fault detection module, and a detection and determination module, wherein:

[0013] The data acquisition module is configured to acquire first electrical signal data at a preset location in a target power grid using the first data acquisition device at a first sampling frequency within a preset first time period, and extract harmonic characteristics from the first electrical signal data; and concurrently, acquire second electrical signal data at the preset location using the second data acquisition device at a second sampling frequency within the preset first time period, and extract current waveform characteristics from the second electrical signal data; the first sampling frequency being greater than the second sampling frequency;

[0014] The first data set processing module is used to determine a binary feature data set according to the harmonic characteristics and the current waveform characteristics;

[0015] The second data set processing module is used to determine a fluctuation feature data set according to the harmonic feature and the current waveform feature;

[0016] The model training module is used to train the preset model based on the binary classification feature data set and the fluctuation feature data set to obtain a binary classification model and a fluctuation model;

[0017] The arc fault detection module is used to perform arc fault detection on the electrical signal of the preset second time period at the preset position by using the binary classification model and the fluctuation model respectively, to obtain a first detection result and a second detection result;

[0018] The detection and determination module is used to determine a target arc fault detection result at the preset position according to the first detection result and the second detection result.

[0019] In a third aspect, an embodiment of the present application provides an electronic device comprising: a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the program includes instructions for executing the steps in the first aspect of the embodiment of the present application.

[0020] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the above-mentioned computer-readable storage medium stores a computer program for electronic data exchange, wherein the above-mentioned computer program enables a computer to execute some or all of the steps described in the first aspect of the embodiment of the present application.

[0021] In a fifth aspect, embodiments of the present application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, wherein the computer program is operable to cause a computer to perform some or all of the steps described in the first aspect of the embodiments of the present application. The computer program product may be a software installation package.

[0022] It can be seen that the embodiments of the present application have the following beneficial effects:

[0023] By implementing the embodiments of the present application, first electrical signal data at a preset location in a target power grid is collected by the first data acquisition device using a first sampling frequency within a preset first time period, and harmonic features are extracted from the first electrical signal data. Simultaneously, second electrical signal data at the preset location is collected by the second data acquisition device using a second sampling frequency within the preset first time period, and current waveform features are extracted from the second electrical signal data. A binary feature dataset is determined based on the harmonic features and the current waveform features. A fluctuation feature dataset is determined based on the harmonic features and the current waveform features. A preset model is trained based on the binary feature dataset and the fluctuation feature dataset to obtain a binary model and a fluctuation model. Arc fault detection is performed on the electrical signal of the preset location in a preset second time period using the binary model and the fluctuation model, respectively, to obtain a first detection result and a second detection result. A target arc fault detection result for the preset location is determined based on the first detection result and the second detection result. It can be seen that arc fault detection is performed by combining the binary classification model and the fluctuation model to improve the detection accuracy of arc faults. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the drawings required for use in the embodiments of the present application or the background technology will be described below.

[0025] Figure 1This is a flow chart of an AC arc detection method based on current and harmonic fluctuations provided in an embodiment of the present application;

[0026] Figure 2 This is a schematic diagram of a harmonic feature data set belonging to a normal signal provided in an embodiment of the present application;

[0027] Figure 3 This is a schematic diagram of a harmonic characteristic data set belonging to an arc signal provided in an embodiment of the present application;

[0028] Figure 4 This is a schematic diagram of fluctuations of harmonic characteristics of a normal signal provided by an embodiment of the present application;

[0029] Figure 5 This is a schematic diagram of fluctuations of harmonic characteristics of an arc signal provided by an embodiment of the present application;

[0030] Figure 6 This is a schematic diagram of a current fluctuation characteristic data set belonging to a normal signal provided in an embodiment of the present application;

[0031] Figure 7 This is a schematic diagram of a current fluctuation characteristic data set belonging to an arc signal provided in an embodiment of the present application;

[0032] Figure 8 This is a schematic diagram of characteristic fluctuations of a current waveform belonging to a normal signal provided by an embodiment of the present application;

[0033] Figure 9 This is a schematic diagram of characteristic fluctuations of a current waveform belonging to an arc signal provided in an embodiment of the present application;

[0034] Figure 10 This is an application scenario diagram of an AC arc detection method based on current and harmonic fluctuations provided in an embodiment of the present application;

[0035] Figure 11 1 is a schematic structural diagram of an AC arc detection device based on current and harmonic fluctuations provided in an embodiment of the present application;

[0036] Figure 12 This is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments 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 creative work are within the scope of protection of this application.

[0038] The terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0039] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0040] The following describes the relevant contents, concepts, meanings, technical issues, technical solutions, beneficial effects, etc. involved in the embodiments of this application.

[0041] See Figure 1 , Figure 1 This is a flow chart of an AC arc detection method based on current and harmonic fluctuations provided by an embodiment of the present application. The method is applied to a control device of an arc detection system. The arc detection system further includes: a first data acquisition device and a second data acquisition device. The method includes but is not limited to the following steps:

[0042] S101. Collect first electrical signal data at a preset position in a target power grid using a first data acquisition device at a first sampling frequency within a preset first time period, and extract harmonic features from the first electrical signal data. Simultaneously, collect second electrical signal data at a preset position using a second data acquisition device at a second sampling frequency within the preset first time period, and extract current waveform features from the second electrical signal data.

[0043] The arc detection system of the embodiment of the present application can be used for arc fault detection in AC power grids, for example, low-voltage distribution systems, industrial electrical equipment and other scenarios. It achieves high-precision and high-reliability detection of arc faults by analyzing the fluctuation characteristics of current and harmonics.

[0044] In an embodiment of the present application, an arc detection system includes a first data acquisition device and a second data acquisition device. Both the first and second data acquisition devices can be a combination of a momentary-action coil (e.g., a Rogowski coil) and an acquisition card (a high-speed ADC acquisition card). The data acquisition devices can sense current signals at a preset location on a target power grid in real time using the momentary-action coil. The ADC acquisition card converts the analog signals into digital signals, generating raw high-frequency current data, thereby implementing data sampling.

[0045] The first data acquisition device can collect high-frequency current signals at a preset location in the target power grid at a first sampling frequency (e.g., 10Msps) for harmonic feature extraction. This high-frequency sampling mechanism can capture sudden harmonic changes in different frequency bands caused by arc faults. The spectral distribution of different frequency channels is further analyzed using a short-time Fourier transform to obtain harmonic features (e.g., harmonic components and phase mutations in the 1kHz-10kHz frequency band). The second data acquisition device can collect current signals at the same preset location at a second sampling frequency (20ksps). The current signals collected at this sampling frequency preserve the time-domain profile of the current waveform and are used to extract current waveform features (e.g., RMS fluctuation, waveform distortion, and zero-crossing offset). This sampling mechanism allows the collection of data points aligned with the time dimension of the harmonic features, avoiding feature misalignment caused by timing deviations and facilitating subsequent feature fusion analysis. It can also effectively reflect the current amplitude variation trend and the degree of non-sinusoidal waveform distortion.

[0046] In the embodiment of the present application, the preset first time period is a preset time period, for example, the past month, etc. Feature extraction is performed on the electrical signal data in the preset first time period to train a binary classification model and a fluctuation model that can identify arcs.

[0047] In an embodiment of the present application, the target power grid can be a low-voltage distribution network or industrial power network with an AC power frequency of 50Hz / 60Hz, which includes a single-phase / three-phase power supply system, specifically including but not limited to low-voltage distribution lines for residential homes and commercial buildings, and industrial power equipment connection lines in factories and workshops. Various types of loads can be connected to the power grid, such as resistive loads, inductive loads, and nonlinear loads, and support power grid scenarios with different line lengths and different topologies. Among them, the preset location can be a specific line node or load input terminal on the target power grid.

[0048] In a specific embodiment, within a preset first time period, first electrical signal data at a preset position in the target power grid is collected by a first data acquisition device using a first sampling frequency, and then the harmonic characteristics in the first electrical signal data can be extracted. At the same time, within the preset first time period, second electrical signal data at a preset position is collected by a second data acquisition device using a second sampling frequency, and then the current waveform characteristics in the second electrical signal data can be extracted.

[0049] The first sampling frequency is greater than the second sampling frequency. Since the first electrical signal data is used to extract harmonic features, and harmonics, as high-frequency components, have a frequency range significantly higher than the power frequency of 50 Hz, according to the Nyquist sampling theorem, a higher sampling frequency (such as 10Msps) is required to avoid spectral aliasing and accurately capture high-frequency harmonic components to ensure that high-frequency details in the original signal (such as 5kHz harmonics) can be fully sampled. Under this circumstance, the current waveform features extracted from the second electrical signal data primarily reflect the overall change trend of the current in the time domain (such as effective value fluctuations, waveform distortion, etc.). These features typically contain more low-frequency components, and there is no need to retain high-frequency details. Therefore, the first sampling frequency is greater than the second sampling frequency.

[0050] It's important to note that the first and second data acquisition devices must use the grid voltage zero-crossing point as the synchronization trigger reference to ensure that the collected high-frequency and low-frequency signals are perfectly aligned in time. For example, when inspecting an industrial motor line, the first data acquisition device captured a surge in 5kHz harmonics (amplitude change rate >30%) at the moment of an arc fault, while the second data acquisition device simultaneously detected that the RMS current fluctuation exceeded the normal range of ±15%. The simultaneous acquisition of these two types of features ensures subsequent joint identification.

[0051] Optionally, the above step of extracting harmonic features from the first electrical signal data may include the following steps:

[0052] A101. Determine the power frequency cycle corresponding to the target power grid;

[0053] A102. Determine a target window according to the power frequency cycle;

[0054] A103, extracting n frequency channels from the first electrical signal data using the target window based on a preset short-time fast Fourier transform, where n is a positive integer;

[0055] A104. Determine a first classification label corresponding to the first electrical signal data; the first classification label includes one of the following: a normal label and an arc label;

[0056] A105: Construct the harmonic feature according to the n frequency channels and the first classification label.

[0057] The preset short-time fast Fourier transform is a pre-set algorithm for time-frequency analysis of time-domain signals. It can segment a non-stationary signal into multiple short-time stationary segments using a sliding window, and then perform a fast Fourier transform on each segment to obtain frequency domain features. In the embodiment of the present application, the window function of the preset short-time fast Fourier transform can use a Hanning window, and the window size can be set to 64 sampling points, which can extract 20 frequency channels.

[0058] In a specific embodiment, the power frequency period corresponding to the target power grid is first determined, and then the target window is determined based on the power frequency period. For example, if the target power grid in the embodiment of the present application has an AC power frequency of 50Hz and a power frequency period of 20ms, the target window can be half of the power frequency period, that is, 10ms, which is the time period from one voltage zero crossing to the next zero crossing. Arc faults may cause half-cycle waveform distortion, and the target window covers the entire half-cycle signal, facilitating analysis of the symmetry of the positive and negative half cycles.

[0059] Based on the preset short-time fast Fourier transform, n frequency channels are extracted from the first electrical signal data with a target window, where n is a positive integer. According to the typical harmonic frequency range of arc faults (1kHz~10kHz), 20 frequency channels can be selected. Each channel corresponds to the energy distribution in a specific frequency range, and each frequency channel can include 200 sampling points.

[0060] Determine a first classification label corresponding to the first electrical signal data, wherein the first classification label includes one of the following: a normal label and an arc label. It should be noted that the first electrical signal data and the second electrical signal data collected within the preset first time period both include normal signals or arc signals. For normal signals, they are collected in arc-free scenarios, such as normal start and stop of equipment, steady-state operation, and the harmonic components in the signal are within the normal range. For arc signals, they are collected in arc fault scenarios, such as artificial arcing by an arc generator or line breakdown, and abnormal high-frequency harmonics exist in the signal. Generated through manual labeling or physical experiments, for normal signals, they can be set to normal labels (marked as 0), and for arc signals, they can be set to arc labels (marked as 1).

[0061] Finally, the harmonic feature is constructed based on the n frequency channels and the first classification label. Specifically, the frequency domain data of the 20 frequency channels, that is, the 20×200 matrix, is associated with the classification label (0 or 1) to form a labeled harmonic feature sample.

[0062] See Figure 2 , Figure 2This is a schematic diagram of a harmonic feature dataset for a normal signal, provided in an embodiment of the present application. As shown in the figure, it illustrates the harmonic feature dataset of a normal signal in the time domain. The horizontal axis represents time (in milliseconds), covering the time span from 0ms to 100ms, while the vertical axis represents the amplitude of the time-frequency feature, reflecting the strength of the harmonic feature. Since the target window is 10ms, the figure shows the harmonic features of 10 normal signals.

[0063] As shown in the figure, while certain harmonic components exist during normal grid operation, there are no drastic harmonic fluctuations, as would be expected from an abnormal situation such as an arc fault. The fluctuations in the figure represent the combined effects of the harmonics generated by various loads under normal operating conditions. Their amplitudes fluctuate within a certain range, with no significant surges or sudden changes. This indicates relatively stable grid operation, without the abnormal increase in harmonics at specific frequencies characteristic of an arc fault.

[0064] See Figure 3 , Figure 3 This figure shows a harmonic signature dataset for arc signals, as provided in an embodiment of the present application. It illustrates the time domain representation of the harmonic signature dataset for arc signals. The horizontal axis represents time (in milliseconds), spanning the time range from 0 to 100 milliseconds. The vertical axis represents the amplitude of the time-frequency signature, reflecting the strength of the harmonic signature. Because the target window is 10 milliseconds, the figure displays the harmonic signatures of ten arc signals.

[0065] The curve in the figure fluctuates significantly, with multiple large spikes. This is because arc faults generate a large number of high-frequency harmonics, disrupting the relatively stable harmonic distribution during normal grid operation. These spikes correspond to the appearance of abnormal high-frequency harmonics at specific moments, indicating that arc faults cause the harmonic components to fluctuate dramatically at certain moments, with energy concentrated in these high-frequency harmonics. Compared with the harmonic feature dataset of normal signals, the time-frequency characteristics of arc signals are significantly larger in amplitude and fluctuate more dramatically, resulting in significant differences in characteristics.

[0066] See Figure 4 , Figure 4 This is a schematic diagram of the fluctuations of the harmonic characteristics of a normal signal, provided in an embodiment of the present application. As shown in the figure, the fluctuations of the harmonic characteristics of a normal signal are shown. The horizontal axis represents time (in milliseconds), ranging from 0 ms to 100 ms, and the vertical axis represents the amplitude of the time-frequency characteristics, reflecting the strength of the harmonic characteristics. The figure shows the fluctuations of the harmonic characteristics from a normal signal to a normal signal within 100 ms.

[0067] The curve in the figure is labeled "Normal to Normal," indicating that the grid operated normally throughout the entire period. As can be seen, the curve exhibits dense and relatively stable fluctuations, without significant spikes or sudden changes. This fluctuation is the result of the combined effects of harmonics generated by various loads during normal grid operation. The amplitude is controlled within a certain range, and there is no sharp increase in energy at any particular frequency.

[0068] See Figure 5 , Figure 5 This diagram illustrates the fluctuations of the harmonic characteristics of an arc signal, as provided in an embodiment of the present application. The diagram shows the fluctuations of the harmonic characteristics of an arc signal. The horizontal axis represents time (in milliseconds) from 0 to 100 milliseconds, while the vertical axis represents the amplitude of the time-frequency characteristics, reflecting the strength of the harmonic characteristics. The diagram shows the fluctuations of the harmonic characteristics from a normal signal to an arc signal within 100 milliseconds.

[0069] In the first half of the timeline (e.g., from 0ms to 50ms), the time-frequency characteristic amplitude is low and fluctuates gently, indicating normal grid operation. Harmonic components are stable, with no abnormal high-frequency energy concentration. However, in the second half (after 50ms), the time-frequency characteristic amplitude increases sharply, with numerous large spikes. This is due to the arc fault generating abundant high-frequency harmonics, resulting in a sudden surge in energy at specific frequencies. This significant difference in the characteristics before and after clearly demonstrates the impact of the arc fault on the grid's harmonic characteristics, demonstrating the transition from normal to faulty grid conditions.

[0070] Optionally, the above step of extracting the current waveform feature from the second electrical signal data may include the following steps:

[0071] B101. Extracting a waveform matrix from the second electrical signal data based on the target window;

[0072] B102. Determine a second classification label corresponding to the second electrical signal data; the second classification label includes one of the following: a normal label and an arc label;

[0073] B103. Determine the current waveform characteristics according to the waveform matrix and the second classification label.

[0074] In a specific embodiment, a waveform matrix is ​​extracted from the second electrical signal data based on a target window, which is consistent with the target window used in harmonic feature extraction, ensuring strict temporal alignment between the current waveform and the harmonic features. The first electrical signal data is acquired at a first sampling frequency (10Msps), and the number of harmonic feature data points within the target window for extraction is 20×200. The second electrical signal data is acquired at a second sampling frequency (20ksps). Within the same target window, 200 sampling points can be acquired. Therefore, the resulting waveform matrix is ​​a 1×200 matrix, representing the instantaneous value of the current at each moment, thereby preserving the time domain profile of the current waveform.

[0075] A second classification label corresponding to the second electrical signal data is determined, wherein the second classification label includes one of the following: a normal label and an arc label. A current waveform feature can be determined based on the waveform matrix and the second classification label.

[0076] See Figure 6 , Figure 6 This is a schematic diagram of a current fluctuation characteristic data set belonging to a normal signal provided by an embodiment of the present application. As shown in the figure, the horizontal axis is time (in ms), covering the time span from 0ms to 100ms, and the vertical axis is the current waveform amplitude, reflecting the instantaneous value of the current. Since the target window is 10ms, the figure shows the current fluctuation characteristics of 10 normal signals. The curve showing a regular and smooth fluctuation shape in the figure indicates that under normal operating conditions, the grid current is in a stable state, conforming to the law of change of the power frequency current, without sudden mutations or abnormal distortion. This regular fluctuation indicates that no abnormal events such as arc faults have occurred in the grid at this time, the current waveform is not disturbed by the fault, and maintains normal electrical characteristics.

[0077] See Figure 7 , Figure 7 This is a schematic diagram of a current fluctuation characteristic data set belonging to an arc signal provided by an embodiment of the present application. As shown in the figure, the horizontal axis is time (in ms), covering the time span from 0ms to 100ms, and the vertical axis is the current waveform amplitude, reflecting the instantaneous value of the current. Since the target window is 10ms, the figure shows the current fluctuation characteristics of 10 arc signals. The figure shows a curve of the fluctuation form of the power frequency current, but on the rising or falling edge of each half cycle, there are obvious small spikes or mutations superimposed. This is due to the instability of the arc fault, resulting in additional high-frequency oscillations or instantaneous changes on the basis of the normal power frequency current fluctuations. These distortion characteristics are in sharp contrast to the smooth and regular waveform of the normal signal, indicating that the arc interference makes the current waveform no longer stable.

[0078] See Figure 8 , Figure 8This is a schematic diagram of the current waveform characteristic fluctuations of a normal signal, provided in an embodiment of the present application. As shown in the figure, this diagram illustrates the fluctuations of the current waveform characteristic of a normal signal. The horizontal axis represents time (in milliseconds), ranging from 0 ms to 100 ms, and the vertical axis represents the current waveform amplitude, reflecting the instantaneous value of the current. This figure illustrates the fluctuations of the current waveform characteristic from a normal signal to a normal signal within 100 ms.

[0079] The curve in the figure is labeled "Normal to Normal," indicating that the power grid remained in normal operation throughout the entire observation period of 0 to 100ms. The waveform shows regular, smooth, and periodically repeating fluctuations, resembling a standard sine wave. Within each cycle, the rising and falling edges of the current waveform are symmetrical and smooth, with no abnormalities such as sudden changes, spikes, or distortion. This demonstrates the stability of the current under normal operating conditions, following the power frequency variation pattern. This regular fluctuation indicates that the power grid was not disturbed by abnormal factors such as arc faults, and the current maintained normal electrical characteristics.

[0080] See Figure 9 , Figure 9 This diagram illustrates the fluctuations in the current waveform characteristics of an arc signal, as provided in an embodiment of the present application. The diagram shows the fluctuations in the current waveform characteristics of an arc signal. The horizontal axis represents time (in milliseconds) ranging from 0 to 100 milliseconds, while the vertical axis represents the current waveform amplitude, reflecting the instantaneous value of the current. The diagram illustrates the fluctuations in the current waveform characteristics from a normal signal to an arc signal over a 100-ms period.

[0081] In the first half of the timeline (e.g., from 0ms to 50ms), the current waveform resembles a regular sine wave, with relatively symmetrical rising and falling edges, reflecting the current characteristics of the power grid during normal operation. However, in the second half (after 50ms), the waveform becomes significantly distorted, with several small sudden changes or spikes superimposed on the original sinusoidal form. This is because the instability of an arc fault disrupts the normal flow of current, disrupting the smoothness and symmetry of the waveform and causing localized transient variations. This change in waveform characteristics from regular to distorted clearly demonstrates the transition from a normal power grid state to an arc fault state.

[0082] S102: Determine a binary feature data set based on harmonic features and current waveform features.

[0083] In a specific embodiment, a binary feature data set can be determined based on the harmonic features and the current waveform features. Specifically, the harmonic features and the current waveform features can be spliced, and the spliced ​​features can be classified. For example, it can be determined whether the spliced ​​features belong to normal signals or arc signals, thereby obtaining binary features and determining a binary feature data set.

[0084] Optionally, the above step of determining a binary feature data set based on harmonic features and current waveform features may include the following steps:

[0085] A201. Combining the harmonic feature and the current waveform feature to obtain a target combined feature;

[0086] A202. Normalize the target combined feature to obtain a first combined feature;

[0087] A203: Determine a target category corresponding to the first combination feature according to the first classification label and the second classification label; the target category includes one of the following: normal category and arc category;

[0088] A204. Determine the binary classification feature data set according to the first combined feature and the target category.

[0089] In this specific embodiment, the harmonic signature is a 20×200 matrix representing 20 frequency channels, each with 200 points, reflecting the frequency domain energy distribution. The current waveform signature is a 1×200 matrix representing the instantaneous current value after downsampling, reflecting the time domain waveform profile. Combining the harmonic signature and the current waveform signature yields a target combined signature, a 21×200 matrix. This target combined signature ensures feature alignment within the same time window (10ms) while preserving the independence of the frequency and time domain features.

[0090] Next, each element in the 21×200 matrix is ​​divided by the effective current value within the current 10ms. This represents the equivalent DC current value of the alternating current within the current 10ms time interval. Harmonic features and current waveform features have different dimensions, so normalization is used to avoid large numerical features. Furthermore, current amplitudes vary significantly under different loads. Normalization ensures that features reflect only relative changes, improving the adaptability of subsequently trained models to different operating conditions.

[0091] The target category corresponding to the first combined feature is determined based on the first and second classification labels. The target category can be one of the following: normal or arc. For example, an arc fault is determined only when both the harmonic and current waveform features simultaneously detect an arc. This simultaneous determination can filter out false positives from a single feature, such as a motor startup with a distorted waveform but normal harmonics, improving reliability.

[0092] Based on the first combined feature and the target category, a binary feature dataset can be determined. The frequency domain and time domain features are combined to cover the multi-dimensional manifestations of arc faults, such as high-frequency harmonic surges and time domain waveform distortion. Simultaneous dual-label judgment can reduce misjudgments.

[0093] S103: Determine a fluctuation characteristic data set according to the harmonic characteristics and the current waveform characteristics.

[0094] In a specific embodiment, a fluctuation feature dataset can be determined based on harmonic features and current waveform features. This fluctuation feature dataset is used to characterize the dynamic changes in features. Specifically, by analyzing the changes in harmonic features and current waveform features within multiple consecutive target windows, this change information is integrated. For example, if the target window is 10ms, a fluctuation feature can be constructed by integrating 10 harmonic features and current waveform features. This fluctuation feature can then characterize feature changes within 100ms.

[0095] Optionally, the harmonic feature includes multiple harmonic features, and the current waveform feature includes multiple current waveform features; the above step of determining the fluctuation feature data set based on the harmonic feature and the current waveform feature may include the following steps:

[0096] A301. Determine a plurality of first combination features according to the plurality of harmonic features and the plurality of current waveform features;

[0097] A302. Combining the plurality of first combination features into a second combination feature;

[0098] A303: Determine the category corresponding to the second combination feature according to the number of first combination features belonging to the normal category in the second combination feature;

[0099] A304. Determine the fluctuation feature dataset according to the second combined feature and the category corresponding to the second combined feature.

[0100] Among them, the harmonic feature may include multiple harmonic features, and the current waveform feature may include multiple current waveform features. For example, in the construction of a binary feature data set, the harmonic feature and the current waveform feature are both based on a target window as the standard, and the fluctuation feature data set is used to characterize the dynamic changes of the feature and can be constructed based on multiple target windows as the standard. At this time, the harmonic feature extracted from the first electrical signal data may include multiple harmonic features, and the current waveform feature extracted from the second electrical signal data may include multiple current waveform features.

[0101] In a specific embodiment, multiple first combination features can be determined based on multiple harmonic features and multiple current waveform features. Each first combination feature corresponds to a target category, where the target category includes one of the following: normal category and arc category. Multiple first combination features are stacked in chronological order to form a second combination feature, which represents the changes in time-frequency features within multiple time windows.

[0102] The category corresponding to the second combination feature is determined according to the number of first combination features belonging to the normal category in the second combination feature. Specifically, for each first combination feature, according to its target category, the number of first combination features belonging to the normal category in multiple target categories is counted to determine the category corresponding to the second combination feature. For example, if the number of first combination features belonging to the normal category exceeds a preset threshold, the category corresponding to the second combination feature is the normal category, otherwise the category corresponding to the second combination feature is the arc category, wherein the preset threshold is a pre-set or system default value, and its value can be 7.

[0103] The fluctuation feature dataset can be determined based on the second combined feature and its corresponding category. A binary feature dataset focuses only on the steady-state characteristics of a single time window, while a fluctuation feature dataset can characterize how features change over time. By combining multiple windows for judgment, false alarms caused by non-fault factors such as load starts and stops and short pulses can be reduced.

[0104] S104 , training the preset models based on the binary classification feature data set and the fluctuation feature data set respectively to obtain a binary classification model and a fluctuation model.

[0105] In an embodiment of the present application, the preset model can be a one-dimensional convolutional neural network. The one-dimensional convolutional neural network has advantages in processing sequence data. The harmonic characteristics and current waveform characteristics involved in arc fault detection, whether data in a binary feature data set or a fluctuation feature data set, all belong to the category of sequence data. For a binary feature data set, a one-dimensional convolutional neural network can quickly identify the characteristic differences between normal and arc states within a single time window. For a fluctuation feature data set, a one-dimensional convolutional neural network can capture the dynamic change trend of features in multiple continuous time windows, thereby accurately judging the occurrence process of the arc fault.

[0106] In a specific embodiment, a binary feature dataset represents the fusion of harmonic and current waveform features within a single time window. This binary feature dataset is used to train a pre-defined model. The model learns the differences in static features between normal and fault conditions. After optimizing parameters, a binary classification model is generated. This model can directly determine whether an arc fault exists in the power grid at the current moment based on the single-window features.

[0107] The fluctuation feature dataset characterizes the dynamic changes in features. Its samples are composed of stacked first-order combined features from multiple consecutive time windows, reflecting the evolution of features over time. When training a preset model using the fluctuation feature dataset, the model focuses on learning the characteristics of feature fluctuations, capturing the dynamic patterns during the occurrence and development of arc faults. Ultimately, a fluctuation model is trained that can identify arc faults based on feature change trends.

[0108] By training the binary classification model and the fluctuation model separately, the dual utilization of static and dynamic features is achieved, thereby improving the accuracy and reliability of arc fault detection.

[0109] S105 , performing arc fault detection on the electrical signal of the preset second time period at the preset position using a binary classification model and a wave model respectively, to obtain a first detection result and a second detection result.

[0110] In the embodiment of the present application, the preset second time period refers to a preset continuous duration for arc fault detection. For example, the preset second time period may be 2000 ms.

[0111] In a specific embodiment, arc fault detection is performed on electrical signals at a preset location and within a preset second time period using a binary classification model. The electrical signals at the preset location are collected and input into the trained binary classification model for analysis, and the binary classification model can output a first detection result. Arc fault detection is performed on electrical signals at the preset location and within a preset second time period using a fluctuation model. The electrical signals at the preset location are collected and input into the trained fluctuation model for analysis, and the fluctuation model can output a second detection result.

[0112] The binary classification model can provide high-precision single-point detection capabilities, and the fluctuation model reduces misjudgments caused by short-term interference through time series correlation. The combination of the two can improve the accuracy of arc fault detection.

[0113] Optionally, the above step of performing arc fault detection on the electrical signal of the preset second time period at the preset position using a binary classification model and a fluctuation model respectively to obtain a first detection result and a second detection result may include the following steps:

[0114] A501. Collect third electrical signal data at the preset position using the first data acquisition device and the first sampling frequency within the preset second time period;

[0115] A502. Detect the third electrical signal data based on the first detection frequency using the binary classification model to obtain the first detection result;

[0116] A503. Collect fourth electrical signal data at the preset position using the second data acquisition device and the second sampling frequency within the preset second time period;

[0117] A504. Detect the fourth electrical signal data based on a second detection frequency using the fluctuation model to obtain the second detection result; the second detection frequency is less than the first detection frequency.

[0118] In a specific embodiment, the third electrical signal data of a preset position is collected by a first data acquisition device using a first sampling frequency within a preset second time period, and then the third electrical signal data is detected based on the first detection frequency using a binary classification model to obtain a first detection result.

[0119] The fourth electrical signal data of the preset position is collected by the second data collection device using the second sampling frequency within the preset second time period, and then the fourth electrical signal data is detected based on the second detection frequency using the fluctuation model to obtain a second detection result.

[0120] The second detection frequency is less than the first detection frequency. For example, the first detection frequency may be once every 10 ms, and the second detection frequency may be once every 100 ms. If the second time period is preset to 2000 ms, it can be determined that the first detection result includes 200 data items, and the second detection result includes 20 data items.

[0121] Through differentiated sampling strategies and detection frequencies, the binary classification model can capture the characteristics of sudden faults in real time and trigger a rapid response, while the fluctuation model confirms the authenticity of the fault through time series analysis and reduces false alarms. The collaborative mechanism of high-frequency early warning and trend verification formed by this dual model can improve the detection accuracy of arc faults.

[0122] S106 . Determine a target arc fault detection result at a preset location according to the first detection result and the second detection result.

[0123] In a specific embodiment, the first detection result is the output of a binary classification model, which is used to capture arc characteristics within a single time window in real time. For example, the binary classification model analyzes the time-frequency feature matrix of each 10ms window to quickly identify whether there is a sudden change in harmonic energy or current waveform distortion, thereby providing high-frequency single-point fault warnings. The second detection result is the output of a fluctuation model, which is used to analyze the dynamic evolution trend of features in continuous time windows. For example, the fluctuation model integrates the feature tensors of multiple windows to determine whether harmonic energy is increasing or whether current waveform distortion is continuously increasing, thereby filtering out false alarms caused by short-term interference.

[0124] The target arc fault detection result at the preset location can be determined based on the first detection result and the second detection result. An arc fault is determined to have occurred only when the binary classification model and the fluctuation model simultaneously identify the arc, thereby improving the accuracy and reliability of the detection.

[0125] Optionally, the first detection result includes: p first categories, where p is a positive integer; the second detection result includes: q second categories, where q is a positive integer less than p; the category includes one of the following: a normal category and an arc category; the above step of determining a target arc fault detection result at a preset location based on the first detection result and the second detection result may include the following steps:

[0126] A601: Filter out q first categories from the p first categories that correspond to the q second categories at the time point of the detection output result; the q first categories correspond one-to-one to the q second categories;

[0127] A602. Perform a target determination step on the q first categories and the q second categories to obtain q third categories.

[0128] The target determination steps are as follows:

[0129] Obtaining a reference first category and a reference second category corresponding to the reference first category among the q second categories; the reference first category is any first category among the q first categories;

[0130] If both the reference first category and the reference second category are the arc category, then the reference third category is determined to be the arc category; otherwise, the reference third category is determined to be the normal category; the reference third category is the third category among the q third categories corresponding to the reference first category and the reference second category;

[0131] A603. If at least k third categories among the q third categories are the arc categories, it is determined that an arc fault exists at the preset position; otherwise, it is determined that no arc fault exists at the preset position; k is a positive integer less than q.

[0132] The first detection result includes p first categories, where p is a positive integer. The second detection result includes q second categories, where q is a positive integer less than p. For example, if the second time period is preset to 2000 ms, the first detection frequency is once every 10 ms using the binary classification model, and the second detection frequency is once every 100 ms using the fluctuation model, then p is 200 and q is 20. A category can be one of the following: normal category or arc category.

[0133] In a specific embodiment, q first categories corresponding to the q second categories at the time of the detection output are selected from the p first categories, where the q first categories correspond one-to-one with the q second categories. Specifically, the fluctuation model outputs a detection result once every 100ms, while the binary classification model outputs ten detection results at the 100th ms. At this time, the tenth output detection result of the binary classification model is matched with the detection result output by the fluctuation model for joint detection.

[0134] By performing the target determination step on the q first categories and the q second categories, q third categories can be obtained. Specifically, when the first category and the second category are both determined to be arc categories, the third category is the arc category; otherwise, the third category is the normal category.

[0135] If at least k of the q third categories are arc categories, an arc fault is determined to exist at the preset location; otherwise, an arc fault is determined not to exist at the preset location, where k is a positive integer less than q. For example, if the second time period is preset to 2000 ms and twenty joint determinations are performed, and at least ten of the third categories are arc categories, an arc fault is determined to exist at the preset location, representing a one-half ratio.

[0136] See Figure 10 , Figure 10 This is an application scenario diagram of an AC arc detection method based on current and harmonic fluctuations provided by an embodiment of the present application. As shown in the figure, the binary classification model performs detection in a 10ms time window and outputs a series of detection results (0 for normal, 1 for arc fault) within 2000ms. This demonstrates its ability to quickly determine the electrical signal characteristics within each short time window and promptly capture possible arc fault characteristics. The fluctuation model performs analysis in a 100ms time window and also outputs corresponding results (0 for normal, 1 for arc fault) within the same 2000ms period. This reflects the dynamic change trend of electrical signal characteristics within a longer time range (100ms), filters short-term interference, and avoids possible false alarms caused by the binary classification model.

[0137] The final detection result combines the outputs of the two models. As shown in the figure, if both models output 1, an arc fault is determined to be present; otherwise, the system is considered normal. This approach leverages the rapid detection capabilities of the binary classification model and the trend analysis capabilities of the fluctuation model, complementing each other to improve the accuracy and reliability of AC arc fault detection and more effectively identify arc faults in actual electrical systems.

[0138] In summary, by implementing the embodiments of the present application, first electrical signal data at a preset location in a target power grid is collected by the first data acquisition device using a first sampling frequency within a preset first time period, and harmonic features are extracted from the first electrical signal data. Simultaneously, second electrical signal data at the preset location is collected by the second data acquisition device using a second sampling frequency within the preset first time period, and current waveform features are extracted from the second electrical signal data. A binary feature dataset is determined based on the harmonic features and the current waveform features. A fluctuation feature dataset is determined based on the harmonic features and the current waveform features. A preset model is trained based on the binary feature dataset and the fluctuation feature dataset to obtain a binary model and a fluctuation model. Arc fault detection is performed on the electrical signal of the preset location in a preset second time period using the binary model and the fluctuation model, respectively, to obtain a first detection result and a second detection result. A target arc fault detection result for the preset location is determined based on the first detection result and the second detection result. It can be seen that arc fault detection is performed by combining the binary classification model and the fluctuation model to improve the detection accuracy of arc faults.

[0139] See Figure 11 , Figure 11 : This is a structural diagram of an AC arc detection device based on current and harmonic fluctuations provided in an embodiment of the present application. The AC arc detection device based on current and harmonic fluctuations 200 includes: a data acquisition module 201, a first data set processing module 202, a second data set processing module 203, a model training module 204, an arc fault detection module 205, and a detection and determination module 206, wherein:

[0140] The data acquisition module 201 is configured to acquire first electrical signal data at a preset location in a target power grid using a first data acquisition device at a first sampling frequency within a preset first time period, and extract harmonic features from the first electrical signal data; and concurrently, acquire second electrical signal data at the preset location using a second data acquisition device at a second sampling frequency within the preset first time period, and extract current waveform features from the second electrical signal data; the first sampling frequency being greater than the second sampling frequency;

[0141] The first data set processing module 202 is used to determine a binary feature data set according to the harmonic features and the current waveform features;

[0142] The second data set processing module 203 is used to determine a fluctuation feature data set according to the harmonic feature and the current waveform feature;

[0143] The model training module 204 is used to train the preset model based on the binary classification feature data set and the fluctuation feature data set to obtain a binary classification model and a fluctuation model;

[0144] The arc fault detection module 205 is configured to perform arc fault detection on the electrical signal of the preset second time period at the preset position using the binary classification model and the fluctuation model, respectively, to obtain a first detection result and a second detection result;

[0145] The detection and determination module 206 is configured to determine a target arc fault detection result at the preset location according to the first detection result and the second detection result.

[0146] Optionally, in terms of extracting harmonic features from the first electrical signal data, the data acquisition module 201 is further specifically configured to:

[0147] Determining a power frequency cycle corresponding to the target power grid;

[0148] Determining a target window according to the power frequency cycle;

[0149] Extracting n frequency channels from the first electrical signal data using the target window based on a preset short-time fast Fourier transform; n is a positive integer;

[0150] Determine a first classification label corresponding to the first electrical signal data; the first classification label includes one of the following: a normal label and an arc label;

[0151] The harmonic feature is constructed according to the n frequency channels and the first classification label.

[0152] Optionally, in terms of extracting current waveform features from the second electrical signal data, the data acquisition module 201 is further specifically configured to:

[0153] extracting a waveform matrix from the second electrical signal data based on the target window;

[0154] Determine a second classification label corresponding to the second electrical signal data; the second classification label includes one of the following: a normal label and an arc label;

[0155] The current waveform feature is determined according to the waveform matrix and the second classification label.

[0156] Optionally, in determining a binary feature dataset based on the harmonic features and the current waveform features, the first dataset processing module 202 is further specifically configured to:

[0157] Combining the harmonic feature and the current waveform feature to obtain a target combined feature;

[0158] Normalizing the target combined feature to obtain a first combined feature;

[0159] Determine the target category corresponding to the first combination feature according to the first classification label and the second classification label; the target category includes one of the following: normal category and arc category;

[0160] The binary feature dataset is determined according to the first combined feature and the target category.

[0161] Optionally, the harmonic feature includes multiple harmonic features, and the current waveform feature includes multiple current waveform features; in determining the fluctuation feature dataset based on the harmonic feature and the current waveform feature, the second dataset processing module 203 is further specifically configured to:

[0162] determining a plurality of first combination features according to the plurality of harmonic features and the plurality of current waveform features;

[0163] Combining the plurality of first combination features into a second combination feature;

[0164] determining the category corresponding to the second combined feature according to the number of first combined features belonging to the normal category in the second combined feature;

[0165] The fluctuation feature data set is determined according to the second combined feature and the category corresponding to the second combined feature.

[0166] Optionally, in terms of performing arc fault detection on the electrical signal of the preset second time period at the preset position by using the binary classification model and the fluctuation model respectively to obtain a first detection result and a second detection result, the arc fault detection module 205 is further specifically configured to:

[0167] Collecting third electrical signal data at the preset position by the first data acquisition device using the first sampling frequency within the preset second time period;

[0168] Detecting the third electrical signal data based on the first detection frequency using the binary classification model to obtain the first detection result;

[0169] Collecting fourth electrical signal data at the preset position by the second data acquisition device using the second sampling frequency within the preset second time period;

[0170] The fourth electrical signal data is detected by the fluctuation model based on a second detection frequency to obtain the second detection result; the second detection frequency is less than the first detection frequency.

[0171] Optionally, the first detection result includes: p first categories, where p is a positive integer; the second detection result includes: q second categories, where q is a positive integer less than p; the category includes one of the following: a normal category and an arc category; in determining the target arc fault detection result of the preset location based on the first detection result and the second detection result, the detection determination module 206 is further specifically configured to:

[0172] Filtering out q first categories corresponding to the q second categories at the time point of the detection output result from the p first categories; the q first categories correspond one-to-one to the q second categories;

[0173] Performing a target determination step on the q first categories and the q second categories to obtain q third categories;

[0174] The target determination steps are as follows:

[0175] Obtaining a reference first category and a reference second category corresponding to the reference first category among the q second categories; the reference first category is any first category among the q first categories;

[0176] If both the reference first category and the reference second category are the arc category, then the reference third category is determined to be the arc category; otherwise, the reference third category is determined to be the normal category; the reference third category is the third category among the q third categories corresponding to the reference first category and the reference second category;

[0177] If at least k of the q third categories are the arc category, it is determined that an arc fault exists at the preset position; otherwise, it is determined that no arc fault exists at the preset position; k is a positive integer less than q.

[0178] The AC arc detection device 200 based on current and harmonic fluctuations described in this application can collect first electrical signal data at a preset location in a target power grid using a first data acquisition device at a first sampling frequency within a preset first time period, extracting harmonic features from the first electrical signal data. Simultaneously, within the preset first time period, collect second electrical signal data at the preset location using a second data acquisition device at a second sampling frequency, extracting current waveform features from the second electrical signal data. A binary feature dataset is determined based on the harmonic features and the current waveform features. A fluctuation feature dataset is determined based on the harmonic features and the current waveform features. Preset models are trained based on the binary feature dataset and the fluctuation feature dataset to obtain a binary model and a fluctuation model. Arc fault detection is performed on the electrical signal at the preset location during a preset second time period using the binary model and the fluctuation model, respectively, to obtain a first detection result and a second detection result. A target arc fault detection result for the preset location is determined based on the first detection result and the second detection result. It can be seen that arc fault detection is performed by combining the binary model and the fluctuation model to improve the detection accuracy of arc faults.

[0179] See Figure 12 , Figure 12 : is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include a processor, a memory, a communication interface, and one or more programs. The processor, memory, and communication interface may be interconnected via a bus. The one or more programs are stored in the memory and configured to be executed by the processor. In the embodiment of the present application, the program includes instructions for performing the following steps:

[0180] collecting first electrical signal data at a preset location in a target power grid using a first data acquisition device at a first sampling frequency within a preset first time period, and extracting harmonic features from the first electrical signal data; and simultaneously collecting second electrical signal data at the preset location at a second sampling frequency within the preset first time period using a second data acquisition device at a second sampling frequency, and extracting current waveform features from the second electrical signal data; the first sampling frequency being greater than the second sampling frequency;

[0181] Determine a binary feature data set according to the harmonic features and the current waveform features;

[0182] determining a fluctuation characteristic data set according to the harmonic characteristics and the current waveform characteristics;

[0183] The preset models are trained based on the binary classification feature data set and the fluctuation feature data set to obtain a binary classification model and a fluctuation model respectively;

[0184] Performing arc fault detection on the electrical signal of the preset second time period at the preset position using the binary classification model and the fluctuation model respectively, to obtain a first detection result and a second detection result;

[0185] A target arc fault detection result at the preset position is determined according to the first detection result and the second detection result.

[0186] The electronic device described in this application can collect first electrical signal data at a preset location in a target power grid using a first data acquisition device at a first sampling frequency within a preset first time period, extract harmonic features from the first electrical signal data, and simultaneously collect second electrical signal data at the preset location using a second data acquisition device at a second sampling frequency within the preset first time period, extract current waveform features from the second electrical signal data; determine a binary feature dataset based on the harmonic features and the current waveform features; determine a fluctuation feature dataset based on the harmonic features and the current waveform features; train a preset model based on the binary feature dataset and the fluctuation feature dataset to obtain a binary model and a fluctuation model; perform arc fault detection on the electrical signal at the preset location in a preset second time period using the binary model and the fluctuation model, respectively, to obtain a first detection result and a second detection result; and determine a target arc fault detection result for the preset location based on the first detection result and the second detection result. It can be seen that arc fault detection is performed by combining the binary classification model and the fluctuation model to improve the detection accuracy of arc faults.

[0187] An embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, and the computer program enables a computer to execute part or all of the steps of any method described in the above method embodiments, and the above computer includes an electronic device.

[0188] The present application also provides a computer program product comprising a non-transitory computer-readable storage medium storing a computer program, wherein the computer program is operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments. The computer program product may be a software installation package, and the computer may comprise an electronic device.

[0189] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0190] The steps of the method or algorithm described in the embodiments of the present application can be implemented in hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, which can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (EEPROM), registers, hard disk, removable hard disk, CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and storage medium can be located in an ASIC. In addition, the ASIC can be located in a terminal device or a management device. Of course, the processor and storage medium can also exist as discrete components in the terminal device or the management device.

[0191] Those skilled in the art will appreciate that in one or more of the above examples, the functions described in the embodiments of the present application can be implemented in whole or in part via software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented in whole or in part in the form of a computer program product. This computer program product comprises one or more computer instructions. When these computer program instructions are loaded and executed on a computer, they fully or partially produce the processes or functions described in the embodiments of the present application. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0192] The modules / units included in the various devices and products described in the above embodiments may be software modules / units, hardware modules / units, or partly software modules / units and partly hardware modules / units. For example, for various devices and products applied to or integrated into a chip, the modules / units included therein may all be implemented in the form of hardware such as circuits, or at least part of the modules / units may be implemented in the form of software programs, which run on a processor integrated inside the chip, and the remaining (if any) modules / units may be implemented in the form of hardware such as circuits; for various devices and products applied to or integrated into a chip module, the modules / units included therein may all be implemented in the form of hardware such as circuits, and different modules / units may be located in the same component (such as a chip, circuit module, etc.) or different components of the chip module, or at least part of the modules / units may be It is implemented in the form of a software program, which runs on the processor integrated inside the chip module, and the remaining (if any) modules / units can be implemented in the form of hardware such as circuits; for various devices and products applied to or integrated in the terminal equipment, the various modules / units contained therein can be implemented in the form of hardware such as circuits, and different modules / units can be located in the same component (for example, chip, circuit module, etc.) or different components in the terminal equipment, or at least some modules / units can be implemented in the form of a software program, which runs on the processor integrated inside the terminal equipment, and the remaining (if any) modules / units can be implemented in the form of hardware such as circuits.

[0193] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above description is only a specific implementation method of the embodiments of the present application and is not intended to limit the scope of protection of the embodiments of the present application. Any modifications, equivalent replacements, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the scope of protection of the embodiments of the present application.

Claims

1. An AC arc detection method based on current and harmonic fluctuations, characterized in that: A control device applied to an arc detection system, wherein the arc detection system further comprises: a first data acquisition device and a second data acquisition device; and the method comprises: collecting first electrical signal data at a preset location in a target power grid using the first data acquisition device at a first sampling frequency within a preset first time period, and extracting harmonic features from the first electrical signal data; and simultaneously collecting second electrical signal data at the preset location at a second sampling frequency within the preset first time period using the second data acquisition device at a second sampling frequency, and extracting current waveform features from the second electrical signal data; the first sampling frequency being greater than the second sampling frequency; Determine a binary feature data set according to the harmonic features and the current waveform features; determining a fluctuation characteristic data set according to the harmonic characteristics and the current waveform characteristics; The preset models are trained based on the binary classification feature data set and the fluctuation feature data set to obtain a binary classification model and a fluctuation model respectively; Performing arc fault detection on the electrical signal of the preset second time period at the preset position using the binary classification model and the fluctuation model respectively, to obtain a first detection result and a second detection result; Determining a target arc fault detection result at the preset position according to the first detection result and the second detection result; The arc fault detection is performed on the electrical signal of the preset second time period at the preset position by using the binary classification model and the fluctuation model respectively to obtain a first detection result and a second detection result, including: Collecting third electrical signal data at the preset position by the first data acquisition device using the first sampling frequency within the preset second time period; Detecting the third electrical signal data based on the first detection frequency using the binary classification model to obtain the first detection result; Collecting fourth electrical signal data at the preset position by the second data acquisition device using the second sampling frequency within the preset second time period; detecting the fourth electrical signal data based on a second detection frequency using the fluctuation model to obtain a second detection result; wherein the second detection frequency is less than the first detection frequency; The first detection result includes p first categories, where p is a positive integer; the second detection result includes q second categories, where q is a positive integer less than p; the category includes one of the following: normal category and arc category; the target arc fault detection result of the preset location is determined based on the first detection result and the second detection result, including: Filtering out q first categories corresponding to the q second categories at the time point of the detection output result from the p first categories; the q first categories correspond one-to-one to the q second categories; Performing a target determination step on the q first categories and the q second categories to obtain q third categories; The target determination steps are as follows: Obtaining a reference first category and a reference second category corresponding to the reference first category among the q second categories; the reference first category is any first category among the q first categories; If both the reference first category and the reference second category are the arc category, then the reference third category is determined to be the arc category; otherwise, the reference third category is determined to be the normal category; the reference third category is the third category among the q third categories corresponding to the reference first category and the reference second category; If at least k of the q third categories are the arc category, it is determined that an arc fault exists at the preset position; otherwise, it is determined that no arc fault exists at the preset position; k is a positive integer less than q.

2. The method according to claim 1, wherein The extracting harmonic features from the first electrical signal data includes: Determining a power frequency cycle corresponding to the target power grid; Determining a target window according to the power frequency cycle; Extracting n frequency channels from the first electrical signal data using the target window based on a preset short-time fast Fourier transform; n is a positive integer; Determine a first classification label corresponding to the first electrical signal data; the first classification label includes one of the following: a normal label and an arc label; The harmonic feature is constructed according to the n frequency channels and the first classification label.

3. The method according to claim 2, wherein The extracting the current waveform feature from the second electrical signal data includes: extracting a waveform matrix from the second electrical signal data based on the target window; Determine a second classification label corresponding to the second electrical signal data; the second classification label includes one of the following: a normal label and an arc label; The current waveform feature is determined according to the waveform matrix and the second classification label.

4. The method according to claim 3, wherein The determining of a binary feature data set according to the harmonic feature and the current waveform feature includes: Combining the harmonic feature and the current waveform feature to obtain a target combined feature; Normalizing the target combined feature to obtain a first combined feature; Determine the target category corresponding to the first combination feature according to the first classification label and the second classification label; the target category includes one of the following: normal category and arc category; The binary feature dataset is determined according to the first combined feature and the target category.

5. The method according to claim 4, wherein The harmonic feature includes a plurality of harmonic features, and the current waveform feature includes a plurality of current waveform features; The determining of the fluctuation characteristic data set according to the harmonic characteristic and the current waveform characteristic includes: determining a plurality of first combination features according to the plurality of harmonic features and the plurality of current waveform features; Combining the plurality of first combination features into a second combination feature; determining the category corresponding to the second combined feature according to the number of first combined features belonging to the normal category in the second combined feature; The fluctuation feature data set is determined according to the second combined feature and the category corresponding to the second combined feature.

6. An AC arc detection device based on current and harmonic fluctuations, characterized in that: A control device applied to an arc detection system, wherein the arc detection system further comprises: a first data acquisition device and a second data acquisition device; the AC arc detection device based on current and harmonic fluctuations comprises: a data acquisition module, a first data set processing module, a second data set processing module, a model training module, an arc fault detection module, and a detection and determination module, wherein: The data acquisition module is configured to acquire first electrical signal data at a preset location in a target power grid using the first data acquisition device at a first sampling frequency within a preset first time period, and extract harmonic characteristics from the first electrical signal data; and concurrently, acquire second electrical signal data at the preset location using the second data acquisition device at a second sampling frequency within the preset first time period, and extract current waveform characteristics from the second electrical signal data; the first sampling frequency being greater than the second sampling frequency; The first data set processing module is used to determine a binary feature data set according to the harmonic characteristics and the current waveform characteristics; The second data set processing module is used to determine a fluctuation feature data set according to the harmonic feature and the current waveform feature; The model training module is used to train the preset model based on the binary classification feature data set and the fluctuation feature data set to obtain a binary classification model and a fluctuation model; The arc fault detection module is used to perform arc fault detection on the electrical signal of the preset second time period at the preset position by using the binary classification model and the fluctuation model respectively, to obtain a first detection result and a second detection result; The detection and determination module is used to determine the target arc fault detection result of the preset position according to the first detection result and the second detection result; The arc fault detection is performed on the electrical signal of the preset second time period at the preset position by using the binary classification model and the fluctuation model respectively to obtain a first detection result and a second detection result, including: Collecting third electrical signal data at the preset position by the first data acquisition device using the first sampling frequency within the preset second time period; Detecting the third electrical signal data based on the first detection frequency using the binary classification model to obtain the first detection result; Collecting fourth electrical signal data at the preset position by the second data acquisition device using the second sampling frequency within the preset second time period; detecting the fourth electrical signal data based on a second detection frequency using the fluctuation model to obtain a second detection result; wherein the second detection frequency is less than the first detection frequency; The first detection result includes p first categories, where p is a positive integer; the second detection result includes q second categories, where q is a positive integer less than p; the category includes one of the following: normal category and arc category; the target arc fault detection result of the preset location is determined based on the first detection result and the second detection result, including: Filtering out q first categories corresponding to the q second categories at the time point of the detection output result from the p first categories; the q first categories correspond one-to-one to the q second categories; Performing a target determination step on the q first categories and the q second categories to obtain q third categories; The target determination steps are as follows: Obtaining a reference first category and a reference second category corresponding to the reference first category among the q second categories; the reference first category is any first category among the q first categories; If both the reference first category and the reference second category are the arc category, then the reference third category is determined to be the arc category; otherwise, the reference third category is determined to be the normal category; the reference third category is the third category among the q third categories corresponding to the reference first category and the reference second category; If at least k of the q third categories are the arc category, it is determined that an arc fault exists at the preset position; otherwise, it is determined that no arc fault exists at the preset position; k is a positive integer less than q.

7. An electronic device, characterized in that: include: a processor, a memory, a communication interface, and one or more programs; The one or more programs are stored in the memory and configured to be executed by the processor, wherein the programs include instructions for executing the steps of the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions. When the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 5.

Images