Electric arc detection method and device for direct current combiner box

Abstract

The invention relates to the technical field of combiner box arc detection, and discloses an arc detection method and device for a direct-current combiner box, and the method comprises the steps: sampling the current of a direct-current bus in the combiner box, obtaining a sampling current, sequentially carrying out the amplification and filtering of the sampling current, and obtaining a target current; performing analog-to-digital conversion on the target current to obtain a digital signal corresponding to the target current; respectively carrying out time domain analysis and frequency domain analysis on the digital signal, and extracting a time domain characteristic value and a frequency domain characteristic value; judging the time domain characteristic value and the frequency domain characteristic value so as to detect whether an electric arc appears or not; in the use process, compared with the mode that electric arc detection is conducted directly through the current threshold value, electric arc detection is conducted through the time domain characteristic value and the frequency domain characteristic value; and the accuracy and the real-time performance of arc identification can be further improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric arc detection of a combiner box, and particularly relates to an electric arc detection method and device for a direct current combiner box. BACKGROUND

[0002] In the field of photovoltaic power generation, the core role of a photovoltaic combiner box is to collect direct currents of multiple photovoltaic modules and uniformly deliver the direct currents to an inverter, while providing protection functions such as overcurrent and lightning protection.

[0003] During use of the photovoltaic combiner box, due to long-term exposure to outdoor environments, problems such as cable aging, poor contact, and insulation damage may occur inside the combiner box, which may further cause series arc or parallel arc failure problems in the combiner box. For the electric arc failure, when the electric arc occurs, the electric arc temperature is extremely high, up to 3000-10000℃, which can instantly ignite the insulation materials and cable sheaths in the combiner box, and then spread to the entire photovoltaic array or machine room, causing large-scale fires. In addition, the high-temperature electric arc can burn the core components such as the wiring terminal, fuse, and lightning protection module, causing an increase in contact resistance, device performance failure, and even direct burning of the combiner box, resulting in system shutdown and economic losses.

[0004] In order to ensure safe use of the photovoltaic combiner box, it is necessary to detect the electric arc in the photovoltaic combiner box and timely disconnect the power switch in the power supply branch when the electric arc occurs. However, current detection methods mostly detect whether the electric arc occurs in the photovoltaic combiner box by judging a current threshold, and such detection methods are easily disturbed by noise of the photovoltaic system, have a high misjudgment rate, and have poor use stability. SUMMARY

[0005] In view of the deficiencies in the background art, the present application provides an electric arc detection method and device for a direct current combiner box, and the technical problem to be solved is that current detection methods mostly detect whether the electric arc occurs in the photovoltaic combiner box by judging a current threshold, and such detection methods are easily disturbed by noise of the photovoltaic system, have a high misjudgment rate, and have poor use stability.

[0006] To solve the above technical problems, in a first aspect, the present application provides the following technical solution: an electric arc detection method for a direct current combiner box, comprising the following steps: S1: sampling a current of a direct current bus in the combiner box to obtain a sampled current; S2: sequentially performing amplification and filtering processing on the sampled current to obtain a target current; S3: performing analog-to-digital conversion on the target current to obtain a digital signal corresponding to the target current; S4: respectively performing time domain analysis and frequency domain analysis on the digital signal to extract a time domain characteristic value and a frequency domain characteristic value of the digital signal; S5: judging the time domain characteristic value and the frequency domain characteristic value to detect whether the arc appears.

[0007] In some embodiments of the first aspect, the time domain characteristic value comprises at least one of a current mean value, a current standard deviation and a current abrupt change point, and the frequency domain characteristic value comprises at least one of a high frequency energy, a spectrum entropy and a wavelet coefficient variance.

[0008] In some embodiments of the first aspect, the current mean value is calculated according to the following formula: ; wherein N is the total number of sampling points in a detection window, i(n) is the current value of the nth sampling point, and n is the sequence index of the sampling point; The current standard deviation is calculated according to the following formula: ; The current abrupt change point is calculated according to the following formula: ; wherein i(n-1) is the current value of the previous sampling point of the nth sampling point; The high frequency energy is calculated according to the following formula: ; ; wherein I(fk) is the transformed complex spectrum, including the amplitude and phase information of the frequency fk, i(n) is the nth sampling point of the time domain current signal, fk is the kth frequency point, fk=k*Fs / N, Fs is the sampling frequency, N is the total number of sampling points for FFT calculation, n is the index of the time domain sampling point, n is greater than or equal to 0 and less than N, k is the index of the time domain frequency point, k is greater than or equal to 0 and less than N; The wavelet coefficient variance is calculated according to the following formula: ; ; wherein is the calculation result, representing the variance of the jth layer of detail coefficients d j , is the arithmetic mean of all detail coefficients d j (m) of the jth layer, d j (m) is the mth detail coefficient of the jth layer of wavelet decomposition, j represents the number of layers of wavelet decomposition, m is the time index of the detail coefficient, and M is the total number of detail coefficients dj of the jth layer; The spectrum entropy is calculated according to the following formula: ; ; wherein pk is the proportion or probability of the energy of the kth frequency point in the total energy or I(fk) = pk k is the complex spectrum value obtained at frequency fk after fast Fourier transform of the original time-domain current signal; k is the total energy of the signal in the entire analysis frequency band.

[0009] In some embodiments of the first aspect, in step S5, the process of judging the time-domain characteristic value and the frequency-domain characteristic value is as follows: whether the corresponding time-domain characteristic value and the frequency-domain characteristic value are both greater than the corresponding set threshold value, and if the time-domain characteristic value and the frequency-domain characteristic value are both greater than the corresponding set threshold value, it is considered that an arc occurs.

[0010] In a second aspect, the present application further provides an arc detection device for a direct current busbar, comprising a current detection unit, a signal preprocessing unit, an analog-to-digital conversion unit, a control unit, a driving unit and a direct current switch. The current detection unit is used to detect the current size of the direct current busbar in the busbar and send the sampling current to the signal preprocessing unit. The signal preprocessing unit generates a target current after preprocessing the sampling current and inputs the target current to the analog-to-digital conversion unit. The analog-to-digital conversion unit is used to convert the input target current into a digital signal. The control unit is electrically connected with the analog-to-digital conversion unit, is used to perform time-domain analysis and frequency-domain analysis on the digital signal, extract the time-domain characteristic value and the frequency-domain characteristic value of the digital signal, and judge whether an arc occurs through the time-domain characteristic value and the frequency-domain characteristic value. The direct current switch is arranged on the direct current busbar, and the control unit controls the direct current switch to be turned off through the driving unit when it is judged that an arc occurs.

[0011] In some embodiments of the second aspect, the current detection unit is a Rogowski coil, and the Rogowski coil is sleeved on the direct current busbar.

[0012] In some embodiments of the second aspect, the signal preprocessing unit comprises an amplification circuit and a band-pass filter, the amplification circuit is used to amplify the sampling current, the band-pass filter is electrically connected with the amplification circuit and is used to filter the amplified sampling current to generate a target current, and the frequency band of the target current is between 100 kHz and 1 MHz.

[0013] ​In some embodiments of the second aspect, the voltage conversion unit is configured to convert an output voltage of the photovoltaic array into a working voltage, the working voltage is input to an input terminal of the drive switch, the control unit is electrically connected to a control terminal of the drive switch, and an output terminal of the drive switch is electrically connected to a control coil of the DC circuit breaker. The voltage conversion unit is configured to convert an output voltage of the photovoltaic array into a working voltage, the working voltage is input to an input terminal of the drive switch, the control unit is electrically connected to a control terminal of the drive switch, and an output terminal of the drive switch is electrically connected to a control coil of the DC circuit breaker.

[0014] In some embodiments of the second aspect, the control unit is further electrically connected to a communication unit, the control unit further generates a detection log, and sends the detection log to a monitoring platform through the communication unit.

[0015] In some embodiments of the second aspect, the control unit stores the detected data, generates a historical database, the historical database includes normal sample data, arc sample data and interference sample data, the control unit learns the sample data in the historical database to generate a detection model, and after the detection model is generated, the detection model is trained for credibility, after the credibility training is completed, the detection model is trained for accuracy, and finally a final model is obtained, the final model is used to adjust the size of the time domain feature value and the frequency domain feature value according to the running environment of the DC bus box.

[0016] Compared with the prior art, the present application has the following beneficial effects: In use, the present application obtains a target current by pre-processing the collected current signal through amplification and filtering, converts the target current into a digital signal, and obtains time domain feature values and frequency domain feature values by simultaneously performing time domain analysis and frequency domain analysis on the digital signal through the control unit, and finally performs arc detection through the time domain feature values and the frequency domain feature values. Compared with directly performing arc detection through a current threshold, the accuracy and real-time performance of arc recognition can be further improved. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The flowchart of the detection method in embodiment one is shown in the figure. Figure 2 The structural schematic diagram of the detection device in embodiment two is shown in the figure. DETAILED DESCRIPTION

[0018] The illustrative embodiments of the present application include, but are not limited to, an arc detection method and device for a DC bus box.

[0019] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The description of the exemplary embodiments is intended to apply to all alternative embodiments, as would be understood by persons skilled in the art. To the extent that they do not particularize to the application, the following exemplary embodiments are not intended to be limiting of the present application, but rather a device and method in accordance with some aspects of the present application, as detailed in the appended claims.

[0020] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items. "Includes," "including," and like terms means encompassing all the elements listed and any and all equivalent thereof, as well as additional elements. "Coupled" or "connected" or like terms are not restricted to direct or physical connections or associations.

[0021] Embodiment One As shown in the figure, the embodiment provides an arc detection method for a DC busbar, comprising the following steps: Figure 1 S1: sampling the current of a DC busbar in the busbar, and obtaining a sampling current; wherein the busbar collects the currents of a plurality of photovoltaic arrays and transmits the currents through a DC busbar; In actual implementation, a Rogowski coil can be sleeved on the DC busbar to sample the current and obtain the sampling current, and the frequency response range of the Rogowski coil is wide, which is suitable for high-frequency arc detection. S2: sequentially amplifying and filtering the sampling current to obtain a target current;

[0022] In actual implementation, a band-pass filter can be used for filtering to eliminate power frequency interference and inverter switching noise, and only signals in an arc characteristic frequency band are retained, wherein the arc characteristic frequency band is between 100 kHz and 1 MHz. S3: analog-to-digital converting the target current to obtain a digital signal corresponding to the target current;

[0023] S4: respectively performing time domain analysis and frequency domain analysis on the digital signal to extract a time domain characteristic value and a frequency domain characteristic value of the digital signal; S5: judging the time domain characteristic value and the frequency domain characteristic value to detect whether an arc occurs. ​

[0024] In use, the present application obtains a target current by pre-processing the collected current signal through amplification and filtering, converts the target current into a digital signal, and obtains time domain characteristic values and frequency domain characteristic values by simultaneously performing time domain analysis and frequency domain analysis on the digital signal through a control unit, and finally performs arc detection through the time domain characteristic values and the frequency domain characteristic values. Compared with directly performing arc detection through a current threshold, the present application can further improve the accuracy and real-time performance of arc recognition.

[0025] Specifically, in the present embodiment, the time domain characteristic values include at least one of a current mean value, a current standard deviation, and a current mutation point. When an arc occurs, the arc causes the effective value of the loop current to decrease, and the average value also decreases accordingly, so the arc can be detected through the current average value. The current standard deviation reflects the fluctuation degree of the current. When an arc occurs, the current mutation intensifies, and the standard deviation usually increases, so the arc can be detected through the current standard deviation. The sharp mutation of the signal is a typical feature of an arc, so the difference between the peak-to-peak values of the current in adjacent time windows can be compared to capture the current mutation point and achieve arc detection.

[0026] Specifically, in the present embodiment, the frequency domain characteristic values include at least one of a high-frequency energy, a spectral entropy, and a wavelet coefficient variance. When an arc occurs, a large amount of noise is generated in the high frequency (such as above 10 kHz), causing the energy in this frequency band to significantly increase, which is the most core frequency domain feature, so the arc can be detected through the high-frequency energy parameter. After wavelet decomposition of the signal, high-frequency coefficients can be obtained. When an arc occurs, the high-frequency coefficients fluctuate, so the wavelet coefficient equation is greater than the normal state, and the arc can be detected through the wavelet coefficient variance. When an arc occurs, the spectrum is broadened and chaotic, and the spectral entropy value increases, so the arc can be checked through the spectral entropy.

[0027] More specifically, in the present embodiment, the calculation formula of the current average value is as follows: ; Where N is the total number of samples in a detection window, i(n) is the current size of the nth sample point, and n is the sequence index of the sample point. The calculation formula of the current standard deviation is as follows: ; In actual implementation, the larger the value of the standard deviation, the more intense the surface current fluctuation. The calculation formula of the current mutation point is as follows: ; where i(n-1) is the current value of the previous sample point of the nth sample point, and the size of the current sudden change point d(n) reflects the instantaneous change amount and direction of the current from n 1 to n moment; The calculation formula of high-frequency energy is as follows: ; ; where I(fk) is the complex spectrum after transformation, including the amplitude and phase information of frequency fk, i(n) is the nth sample point of the time-domain current signal, fk is the kth frequency point, fk=k*Fs / N, Fs is the sampling frequency, N is the total number of sample points for FFT calculation, which is usually an integer power of 2, such as 512 or 1024, n is the index of the time-domain sample point, n is greater than or equal to 0 and less than N, k is the index of the time-domain frequency point, k is greater than or equal to 0 and less than N; The calculation formula of wavelet coefficient variance is as follows: ; ; where is the calculation result, representing the variance of the jth layer of detail coefficients d j , measuring the degree of dispersion of the wavelet coefficients in the jth layer around their mean value. The greater the variance, the more intense the fluctuations of the signal in the frequency band; is the arithmetic mean of all detail coefficients d j (m) of the jth layer, d j (m) is the mth detail coefficient of the jth layer of wavelet decomposition, corresponding to the time point of the original signal, reflecting the signal component intensity at j frequency band and m moment, j represents the number of layers of wavelet decomposition, the smaller the value of j, the higher the center frequency of the corresponding frequency band, m is the time index of the detail coefficient, and M is the total number of detail coefficients dj of the jth layer; The calculation formula of spectral entropy is as follows: ; ; where pk is the proportion or probability of the energy of the kth frequency point in the total energy I(f k ) is the complex spectrum value obtained at frequency f k by performing fast Fourier transform on the original time-domain current signal; is the total energy of the signal in the entire analysis frequency band.

[0028] Specifically, in step S5, the process of judging the time-domain feature value and the frequency-domain feature value is as follows: Determine whether the corresponding time-domain feature value and frequency-domain feature value are both greater than the corresponding set threshold. If both the time-domain feature value and frequency-domain feature value are greater than the corresponding set threshold, then an electric arc is considered to have occurred.

[0029] Example 2 like Figure 2 As shown, this embodiment provides an arc detection device for a DC combiner box, including a current detection unit 1, a signal preprocessing unit 2, an analog-to-digital conversion unit 3, a control unit 4, a drive unit 5, and a DC switch 6; The current detection unit 1 is used to detect the magnitude of the current in the DC bus in the combiner box and send the sampled current to the signal preprocessing unit 2; Signal preprocessing unit 2 preprocesses the sampled current to generate the target current, and then inputs the target current into analog-to-digital conversion unit 3; Analog-to-digital converter 3 is used to convert the input target current into a digital signal; The control unit 4 is electrically connected to the analog-to-digital conversion unit 3 and is used to perform time-domain and frequency-domain analysis on digital signals, extract time-domain and frequency-domain feature values ​​of digital signals, and determine whether an electric arc occurs based on the time-domain and frequency-domain feature values. DC switch 5 is installed on the DC bus. When the control unit 4 detects the presence of an electric arc, it controls DC switch 6 to turn off through drive unit 5.

[0030] In use, this invention uses a signal preprocessing unit 2 to amplify and filter the current signal collected by the current detection unit 1 to obtain the target current, and converts the target current into a digital signal. The control unit 4 then performs time-domain and frequency-domain analysis on the digital signal to obtain time-domain and frequency-domain feature values. Finally, the time-domain and frequency-domain feature values ​​are used for arc detection. Compared with directly using the current threshold for arc detection, this invention can further improve the accuracy and real-time performance of arc identification.

[0031] Specifically, in this embodiment, the current detection unit 1 is a Rogowski coil, which is wound around the DC bus. In use, the Rogowski coil has a wide frequency response range, making it suitable for high-frequency arc detection.

[0032] Specifically, in this embodiment, in Figure 2 In the signal preprocessing unit 2, there are an amplifier circuit 20 and a bandpass filter 21. The amplifier circuit 20 is used to amplify the sampled current. The bandpass filter 21 is electrically connected to the amplifier circuit 20 and is used to filter the amplified sampled current to generate a target current. The frequency band of the target current is between 100kHz and 1MHz.

[0033] Specifically, in this embodiment, in Figure 2In the embodiment, the voltage conversion unit 7 is used to convert the output voltage of the photovoltaic array into a working voltage, the working voltage is input to the input end of the driving switch, the control unit 4 is electrically connected with the control end of the driving switch, and the output end of the driving switch is electrically connected with the control coil of the DC circuit breaker. The voltage conversion unit 7 is used to convert the output voltage of the photovoltaic array into a working voltage, the working voltage is input to the input end of the driving switch, the control unit 4 is electrically connected with the control end of the driving switch, and the output end of the driving switch is electrically connected with the control coil of the DC circuit breaker.

[0034] In actual use, when the control unit 4 controls the driving switch to be turned on, the working voltage is input to the control coil of the DC circuit breaker, so that the DC circuit breaker is turned on; when the arc appears, the control unit 4 controls the driving switch to be turned off, so that the DC circuit breaker is turned off, thereby realizing arc protection.

[0035] Further, in the embodiment, the driving switch can be a thyristor or a circuit breaker, and the control unit 4 can be an MCU, and the model of the MCU is selected according to actual requirements. In addition, the voltage conversion unit 7 is also in communication connection with the control unit 4, and relevant power parameters are sent to the control unit 4.

[0036] Specifically, in the embodiment, the control unit 4 is also electrically connected with a communication unit 8, the control unit 4 also generates a detection log, and the detection log is sent to a monitoring platform through the communication unit 8, so that personnel can remotely view the arc state of the busbar cabinet.

[0037] Further, the monitoring platform can generate a log report according to the sent detection log, and the log report is used to display the normal and abnormal conditions of the arc of the busbar cabinet every day, and the monitoring platform is also provided with a search window, and the search window is used to input arc search information of the busbar cabinet, and the arc search information is date information.

[0038] Specifically, in the embodiment, the control unit 4 stores the detected data to generate a historical database, the historical database includes normal sample data, arc sample data and interference sample data, the control unit 4 learns the sample data in the historical database to generate a detection model, and after the detection model is generated, the detection model is trained in credibility, and after the credibility training is completed, the detection model is trained in accuracy, and finally a final model is obtained, and the final model is used to adjust the size of the time domain characteristic value and the frequency domain characteristic value according to the running environment of the DC busbar; so that the present application has better system adaptability, can automatically adjust the detection threshold according to the use of the DC busbar, and can further improve the checking accuracy.

[0039] According to the above description, relevant personnel can make various changes and modifications without deviating from the technical idea of the present application. The technical scope of the present application is not limited to the content in the specification, and must be determined according to the scope of claims.

Claims

1. An arc detection method for a DC combiner box, characterized in that, The process includes the following steps: S1: Sample the current of the DC bus in the combiner box to obtain the sampled current; S2: Amplify and filter the sampled current sequentially to obtain the target current; S3: Perform analog-to-digital conversion on the target current to obtain a digital signal corresponding to the target current; S4: Perform time-domain analysis and frequency-domain analysis on the digital signal respectively, and extract the time-domain feature value and frequency-domain feature value of the digital signal; S5: The time-domain feature value and frequency-domain feature value are judged to detect whether an electric arc occurs.

2. The arc detection method for a DC combiner box according to claim 1, characterized in that, The time-domain feature values ​​include at least one of the current mean, current standard deviation, and current abrupt change point, and the frequency-domain feature values ​​include at least one of the high-frequency energy, spectral entropy, and wavelet coefficient variance.

3. The arc detection method for a DC combiner box according to claim 2, characterized in that, The formula for calculating the average current is as follows: ； Where N is the total number of samples within a detection window, i(n) is the current magnitude of the nth sampling point, and n is the sequence number of the sampling points; The formula for calculating the standard deviation of the current is as follows: ； The formula for calculating the current abrupt change point is as follows: ； Where i(n-1) is the current value of the previous sampling point of the nth sampling point; The formula for calculating the high-frequency energy is as follows: ； ； Where I(fk) is the transformed complex spectrum, including the amplitude and phase information of frequency fk, i(n) is the nth sampling point of the time-domain current signal, fk is the kth frequency point, fk=k*Fs / N, Fs is the sampling frequency, N is the total number of sampling points used for FFT calculation, n is the index of the time-domain sampling point, n is greater than or equal to 0 and less than N, and k is the index of the time-domain frequency point, k is greater than or equal to 0 and less than N; The formula for calculating the variance of the wavelet coefficients is as follows: ； ； in The result represents the detail coefficient d of the j-th layer. j variance For all detail coefficients d in the j-th layer j The arithmetic mean of (m), d j (m) represents the m-th detail coefficient of the j-th wavelet decomposition, j represents the wavelet decomposition level, m is the time index of the detail coefficient, and M is the total number of detail coefficients dj of the j-th level. The formula for calculating the spectral entropy is as follows: ； ； Where pk is the proportion or probability I(f) of the energy at the k-th frequency point in the total energy. k The original time-domain current signal, after undergoing a Fast Fourier Transform, is at a frequency f. k The complex spectrum value obtained at; This represents the total energy of the signal across the entire analysis frequency band.

4. The arc detection method for a DC combiner box according to any one of claims 1-3, characterized in that, In step S5, the process of judging the time-domain feature value and the frequency-domain feature value is as follows: Determine whether the corresponding time-domain feature value and frequency-domain feature value are both greater than the corresponding set threshold. If both the time-domain feature value and frequency-domain feature value are greater than the corresponding set threshold, then an electric arc is considered to have occurred.

5. An arc detection device for a DC combiner box, characterized in that, It includes a current detection unit, a signal preprocessing unit, an analog-to-digital conversion unit, a control unit, a drive unit, and a DC switch; The current detection unit is used to detect the magnitude of the current in the DC bus in the combiner box and send the sampled current to the signal preprocessing unit. The signal preprocessing unit preprocesses the sampled current to generate a target current, and then inputs the target current into the analog-to-digital conversion unit. The analog-to-digital converter unit is used to convert the input target current into a digital signal; The control unit is electrically connected to the analog-to-digital conversion unit and is used to perform time-domain analysis and frequency-domain analysis on the digital signal, extract the time-domain feature value and frequency-domain feature value of the digital signal, and determine whether an electric arc occurs based on the time-domain feature value and frequency-domain feature value. The DC switch is installed on the DC bus, and when the control unit determines that an electric arc has occurred, it controls the DC switch to turn off through the drive unit.

6. An arc detection device for a DC combiner box according to claim 5, characterized in that, The current detection unit is a Rogowski coil, which is wound around the DC bus.

7. An arc detection device for a DC combiner box according to claim 5, characterized in that, The signal preprocessing unit includes an amplifier circuit and a bandpass filter. The amplifier circuit amplifies the sampled current, and the bandpass filter is electrically connected to the amplifier circuit to filter the amplified sampled current to generate a target current. The frequency band of the target current is between 100kHz and 1MHz.

8. An arc detection device for a DC combiner box according to claim 5, characterized in that, It also includes a voltage conversion unit, and the drive unit includes a drive switch, wherein the DC switch is a DC circuit breaker; The voltage conversion unit is used to convert the output voltage of the photovoltaic array into the working voltage. The working voltage is input to the input terminal of the drive switch. The control unit is electrically connected to the control terminal of the drive switch. The output terminal of the drive switch is electrically connected to the control coil of the DC circuit breaker.

9. An arc detection device for a DC combiner box according to claim 5, characterized in that, The control unit is also electrically connected to a communication unit, and the control unit also generates detection logs and sends the detection logs to the monitoring platform through the communication unit.

10. An arc detection device for a DC combiner box according to claim 5, characterized in that, The control unit stores the detected data and generates a historical database, which includes normal sample data, arcing sample data, and interference sample data. The control unit learns from the sample data in the historical database to generate a detection model, and after the detection model is generated, it performs credibility training on the detection model. After the feasibility training is completed, it performs accuracy training to obtain the final model. The final model is used to adjust the corresponding set threshold values ​​of the time domain feature values ​​and frequency domain feature values ​​according to the operating environment of the DC combiner box.

Images