Photovoltaic direct current arc fault identification device and identification method
By using a current transformer design with reverse connection method and signal processing technology, combined with artificial intelligence algorithms, the problem of low accuracy in arc fault detection in photovoltaic power generation systems has been solved, achieving efficient identification of arc faults.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHUZHOU CSR TIMES ELECTRIC CO LTD
- Filing Date
- 2024-10-30
- Publication Date
- 2026-05-01
AI Technical Summary
The detection of DC arc faults in photovoltaic power generation systems suffers from problems such as transformer magnetic saturation and interference from interference sources, resulting in low detection accuracy and difficulty in effectively identifying arc faults.
A current transformer design employing the forward and reverse connection method, combined with differential amplifier circuits, filter circuits, and detector circuits, utilizes an arc fault detection model based on a one-dimensional convolutional neural network to improve the accuracy of arc signal extraction and identification through signal processing and artificial intelligence algorithms.
It effectively reduces magnetic saturation and common-mode noise in current transformers, improves the accuracy and sensitivity of arc fault detection, ensures the purity and stability of arc signals, and enhances the reliability of fault identification.
Smart Images
Figure CN121966443A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of photovoltaic power generation technology, specifically to a photovoltaic DC arc fault identification device and identification method. Background Technology
[0002] Solar energy, as a clean energy source, has received widespread attention, with photovoltaic (PV) power generation being its primary development method. PV power generation systems convert solar energy into electrical energy through the photoelectric effect, and have become an indispensable component of the new energy field. However, with the widespread application of PV power generation systems, fire accidents within these systems are gradually increasing, especially those caused by DC arc faults. Arc temperatures can reach thousands of degrees Celsius, easily igniting surrounding combustibles and causing serious safety accidents. Therefore, the detection of DC arcs is essential. The electrical characteristics of a DC arc include current and voltage, but due to the randomness of the arc's location, it is difficult to collect the arc voltage signal using a voltage transformer. This poses a challenge to the acquisition of arc voltage data; therefore, the usual practice is to collect DC arc current data for processing and analysis.
[0003] The DC current in a photovoltaic DC system is a DC signal that does not cross zero or commutate, making it highly susceptible to signal distortion due to transformer magnetic saturation. Avoiding transformer magnetic saturation and reducing interference in the line are key research issues, with the core objective of improving the accuracy of arc fault detection. DC current transformers play a crucial role in monitoring current in photovoltaic systems, but they are prone to saturation when faced with high-amplitude, highly transient DC arc currents, posing a significant challenge to the accuracy of current monitoring.
[0004] First, the magnetic saturation problem of current transformers directly affects the accuracy of arc detection. If a DC component exists in the primary winding of a current transformer, the magnetic flux will continuously increase until it reaches saturation. At this point, all the current in the primary winding will be used for excitation, and the secondary current of the transformer will become zero, causing distortion of the secondary signal. Therefore, if a DC component exists in the primary current of a current transformer, the magnetic flux will continuously increase and eventually reach saturation. The saturation of a DC current transformer can lead to distortion of the arc signal, making accurate detection of fault arcs extremely difficult. Solving the problem of magnetic saturation in DC current transformers will provide a technical foundation for improving the sensitivity and accuracy of arc detection.
[0005] Secondly, various interference sources in photovoltaic DC system lines can easily interfere with the detection of arc signals. These interferences mainly include time-varying photovoltaic array output signals and inverter switching noise, etc. The noise they introduce overlaps with the characteristics of arc signals, making it difficult to effectively distinguish between normal operating signals and arc signals. Therefore, reducing interference in the line and improving signal purity and stability are crucial to ensuring reliable detection of fault arcs. Current diagnostic technologies face challenges such as signal aliasing and low signal-to-noise ratio, which further increases the difficulty of fault arc diagnosis. Summary of the Invention
[0006] To address the technical problems existing in the prior art, this invention provides a photovoltaic DC arc fault identification device and method with accurate fault diagnosis.
[0007] To solve the above-mentioned technical problems, the technical solution proposed by this invention is as follows:
[0008] A photovoltaic DC arc fault identification device includes a sampling unit, a signal processing module, and a processing unit; the sampling unit, the signal processing module, and the processing unit are connected in sequence; the sampling unit includes a current transformer, and the positive busbar from the photovoltaic array passes through the current transformer in a forward and reverse manner.
[0009] Preferably, the signal processing module includes a differential amplifier circuit, a filter circuit, and a detector circuit; the differential amplifier circuit, the filter circuit, and the detector circuit are connected in sequence; the current transformer is used to detect the current output of a photovoltaic string, and after differential processing by the differential amplifier module, the interference signal is filtered out by the filter circuit, and finally sent to the processing unit by the detector circuit.
[0010] Preferably, the processing unit includes an AD conversion circuit and a microcontroller, wherein the AD conversion circuit is connected to the microcontroller.
[0011] Preferably, the processing unit has a built-in arc fault detection model, which is used to receive current data processed by the signal processing module to determine whether there is an arc fault.
[0012] The present invention also discloses an identification method based on the photovoltaic DC arc fault identification device described above, comprising the following steps:
[0013] The positive busbar from the photovoltaic array is passed through the current transformer in both forward and reverse directions.
[0014] The current transformer detects the current signal output by a photovoltaic string;
[0015] The signal processing module analyzes and processes the current signal before sending it to the processing unit;
[0016] The processing unit processes the signal output by the signal processing module to determine whether there is an arc fault.
[0017] Preferably, in the processing unit, the signal output by the signal processing module is first converted by AD, and then the presence or absence of an arc is determined by the fault arc detection algorithm.
[0018] Preferably, the process of determining the presence or absence of an arc using the fault arc detection algorithm is as follows: determine whether the sampling time of the current data has reached the sampling period; if it has reached the sampling period, process it into one-dimensional current data, extract representative features of the current, normalize the feature data, and then send it into the pre-trained arc fault detection model for detection.
[0019] Preferably, the arc fault detection model is an arc detection model based on a one-dimensional convolutional neural network.
[0020] Preferably, after detecting an arc fault, a command is sent to the inverter to disconnect the photovoltaic circuit.
[0021] Preferably, the current signal is analyzed and processed in the signal processing module as follows: first, the current signal is differentially processed, then interference signals are filtered out, and finally, detection processing is performed.
[0022] Compared with the prior art, the advantages of the present invention are as follows:
[0023] This invention employs a current transformer sampling design using a forward and reverse connection method, which can further reduce the possibility of transformer saturation, eliminate common-mode current noise, and effectively extract DC arc signals. The arc detection method of this invention, which integrates the differential amplifier circuit, filter circuit, and logarithmic detector circuit, further amplifies the arc characteristics. Inputting the above-mentioned arc characteristic data into an artificial intelligence-based arc identification model can improve training speed and effectively improve accuracy. Attached Figure Description
[0024] Figure 1 This is a diagram illustrating an embodiment of the photovoltaic DC arc fault identification device of the present invention in a specific application.
[0025] Figure 2 This is a schematic diagram of the positive busbar passing through the current transformer in both forward and reverse directions according to the present invention.
[0026] Figure 3 This is a diagram illustrating an embodiment of the signal processing module of the present invention in a specific application.
[0027] Figure 4 This is a flowchart of an embodiment of the arc fault detection method of the present invention. Detailed Implementation
[0028] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.
[0029] like Figure 1 As shown, the photovoltaic DC arc fault identification device provided in this embodiment of the invention includes a sampling unit, a signal processing module, and a processing unit connected in sequence.
[0030] The sampling unit includes a current transformer. The positive busbar from the photovoltaic array passes through the current transformer in a reverse manner (the wire passes through one side of the circular current transformer, bends, and then passes back from the other side). Specifically, the sampling unit is a crucial component for effectively avoiding interference in the line. Typically, current transformers are used for fault arc detection. A busbar is passed through the current transformer, which senses changes in the line current and collects the current information. In DC systems, the DC current in the line generates excitation, which can easily lead to magnetic saturation of the current transformer. Magnetic saturation is caused by flux saturation in the transformer's magnetic circuit. When the DC component amplitude of the current in the line is high, the transformer flux reaches the saturation limit of the magnetic material. At this point, changes in the primary current do not cause corresponding linear changes in the secondary flux, causing the transformer output signal to no longer increase linearly with the current but instead tends to saturate, losing its ability to accurately measure the current.
[0031] Therefore, the magnetic core of current transformers used on DC buses is usually notched to prevent the DC component of the current from saturating the transformer's magnetic field. However, the notch in the transformer core can lead to magnetic leakage, which can easily cause high-frequency arc signals to leak out, thus reducing its detection accuracy. Furthermore, the noise generated by the inverter during operation can also affect the accuracy of arc fault identification.
[0032] To avoid transformer magnetic saturation, photovoltaic inverter interference noise, and magnetic leakage, this invention uses a differential (reverse-crossing) method to pass the line through the current transformer, such as... Figure 2 As shown, this not only avoids the problem of magnetic saturation in current transformers, but also exhibits better performance in arc signal feature extraction. The specific process of the differential connection method for current transformers is as follows: Figure 2 As shown, the inverter DC bus is first passed through the current transformer magnetic ring, and then it is passed out in reverse so that the current flowing into the current transformer magnetic ring is equal to the current flowing out of the current transformer magnetic ring.
[0033] Reverse current flow involves changing the direction of the current in the transformer windings so that the magnetic fields generated by the current cancel each other out during the reverse flow process. The key to this operation is that the changes in magnetic flux caused by the current flowing through the transformer windings in adjacent positive and negative directions are relatively symmetrical.
[0034] As shown in the following formula, assume that the current flowing into the current transformer is i1, and the magnetic flux it generates is Ф1; the current flowing out of the current transformer is i2, and the magnetic flux it generates is Ф2.
[0035] φ1=k+i1 φ2=k-i2
[0036] The secondary output current ic of the current transformer is:
[0037]
[0038] In the formula, N is the number of turns of the transformer coil, and w is the fundamental angular frequency.
[0039] From Equation 2, we can see that k + -k - The value of i is very small, and when the current frequency is low (i.e., w is very small), i c Approximately 0. When the current is a high-frequency signal, i = 0, i c This refers to the secondary current generated by the high-frequency components of the current signal.
[0040] Furthermore, in photovoltaic systems, when a DC arc fault occurs, the DC current of the positive and negative buses of the photovoltaic power supply simultaneously contains both the fault arc current signal and the inverter common-mode current noise signal. The inverter common-mode current noise is related to the PWM control strategy and typically exhibits periodic repetition over a certain time scale, while the DC arc signal does not. Because the inverter noise signal has a certain regularity, the magnetic field generated in the current transformer can cancel it out, greatly suppressing the inverter noise signal in the current signal induced by the current transformer. However, the DC arc signal is chaotic and disordered; the magnetic field it generates in the current transformer will not cancel it out, and the current signal induced by the current transformer still contains the arc signal. Therefore, the positive and negative connection method can also effectively eliminate the common-mode current noise in the bus, thereby effectively extracting the DC arc fault signal.
[0041] In summary, when current transformers are wired differentially, the magnetic flux generated by the DC component of the current signal cancels out, and the magnetic flux generated by interference signals such as the switching frequency with common-mode characteristics also cancels out. By connecting the DC bus of the current transformer differentially, with the DC bus positioned asymmetrically relative to the input and output lines of the transformer, the magnetic flux induced by high-frequency signals in the transformer core can be asymmetrically distributed, thus retaining only the high-frequency component on the secondary side of the transformer. Unlike traditional methods, this method does not require complex circuitry and signal processing, and can easily detect high-frequency components while filtering out low-frequency components.
[0042] like Figure 3As shown, the signal processing module includes a differential amplifier circuit, a filter circuit, a detector circuit, and an AD conversion circuit connected in sequence. The differential amplifier circuit, filter circuit, and detector circuit work together to improve the accuracy of DC arc detection. Specifically, a current transformer is used to detect the current output of a photovoltaic string. The current signal output by the current transformer is processed differentially by the differential amplifier module to retain all the information sensed by the transformer, and then sent to the filter circuit to filter out unnecessary interference signals. Afterward, it flows into the detector circuit, and the output current data is sent to the processing unit for AD conversion. After being converted into a digital quantity, an algorithm is used to detect the presence or absence of an arc.
[0043] This invention employs a current transformer sampling design using a forward and reverse connection method, which can further reduce the possibility of transformer saturation, eliminate common-mode current noise, and effectively extract DC arc signals. The arc detection method of this invention, which integrates the differential amplifier circuit, filter circuit, and logarithmic detector circuit, further amplifies the arc characteristics. Inputting the above-mentioned arc characteristic data into an artificial intelligence-based arc identification model can improve training speed and effectively improve accuracy.
[0044] This invention also provides an identification method based on the photovoltaic DC arc fault identification device described above, comprising the following steps:
[0045] The positive busbar from the photovoltaic array is passed through the current transformer in both forward and reverse directions.
[0046] The current transformer detects the current signal output by a photovoltaic string;
[0047] The signal processing module analyzes and processes the current signal before sending it to the processing unit;
[0048] The processing unit processes the signal output by the signal processing module to determine whether there is an arc fault.
[0049] like Figure 4 As shown, an artificial intelligence algorithm is used for arc fault detection: First, it is determined whether the sampling time of the data sent to the microcontroller has reached the sampling period. If it has, it is processed into one-dimensional current data, and representative features of the current are extracted. After the feature data is normalized, it is sent to the arc fault detection model for detection. The detection model can be an arc detection model based on a one-dimensional convolutional neural network, or other algorithm models. The model is trained in advance with a large amount of arc data to ensure accuracy before it is put into use. After an arc fault is detected, the microcontroller sends a command to the inverter to disconnect the circuit, thus protecting the circuit.
[0050] In practical applications, the busbar is passed through the current transformer in both directions, and a sampling resistor is connected to the secondary winding of the current transformer to collect the voltage signal across the sampling resistor. The collected signal is processed using a differential module, and a filtering module and a detection module work together. The detection module performs detection processing on the input data, greatly increasing the frequency range of signal measurement to the MHz level. Detection processing is beneficial for highlighting the random changes in the signal and is inherently applicable to situations where the arc current of a series fault is small and the arc characteristics are weak. The processed data is sent to the microcontroller processing unit for AD conversion, and then the presence or absence of an arc is determined by a fault arc detection algorithm.
[0051] In summary, this invention improves the accuracy of arc detection through the following three aspects:
[0052] This study addresses the issues of magnetic saturation in current transformers and interference from sources affecting arc signal detection. A sampling design for current transformers using a reverse-connection method is presented. When the current varies between forward and reverse directions, the reverse connection operation makes the magnetic flux change in the transformer windings more uniform, reducing the possibility of magnetic saturation. The reverse-connection method effectively eliminates common-mode current noise in the busbar, thereby effectively extracting DC arc fault signals.
[0053] Hardware design optimization. An arc detection method integrating differential amplifier circuit, filter circuit, and logarithmic detector circuit. Differential signals are acquired by the differential circuit, bandpass filtered, and then processed by the detection module using logarithmic detection to effectively amplify the arc characteristics.
[0054] Algorithm design optimization. An artificial intelligence-based arc fault detection model is developed. First, it checks if the sampling time of the data sent to the microcontroller has reached the sampling period. If so, it is processed into one-dimensional current data. Representative features of the current are extracted, and the feature data is normalized before being fed into the arc fault detection model for detection. The detection model can be an arc detection model based on a one-dimensional convolutional neural network, or other algorithm models. The model is trained in advance using a large amount of arc data to ensure accuracy before deployment.
[0055] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should be considered within the scope of protection of the present invention.
Claims
1. A photovoltaic DC arc fault identification device, characterized in that, It includes a sampling unit, a signal processing module, and a processing unit; the sampling unit, the signal processing module, and the processing unit are connected in sequence; the sampling unit includes a current transformer, and the positive busbar from the photovoltaic array passes through the current transformer in a forward and reverse manner.
2. The photovoltaic DC arc fault identification device according to claim 1, characterized in that, The signal processing module includes a differential amplifier circuit, a filter circuit, and a detector circuit; the differential amplifier circuit, the filter circuit, and the detector circuit are connected in sequence; the current transformer is used to detect the current output of a photovoltaic string, and after differential processing by the differential amplifier module, the interference signal is filtered out by the filter circuit, and finally sent to the processing unit by the detector circuit.
3. The photovoltaic DC arc fault identification device according to claim 1 or 2, characterized in that, The processing unit includes an AD conversion circuit and a microcontroller, and the AD conversion circuit is connected to the microcontroller.
4. The photovoltaic DC arc fault identification device according to claim 3, characterized in that, The processing unit has a built-in arc fault detection model, which is used to receive current data processed by the signal processing module to determine whether there is an arc fault.
5. A method for identifying photovoltaic DC arc faults based on any one of claims 1-4, characterized in that, Including the following steps: The positive busbar from the photovoltaic array is passed through the current transformer in both forward and reverse directions. The current transformer detects the current signal output by a photovoltaic string; The signal processing module analyzes and processes the current signal before sending it to the processing unit; The processing unit processes the signal output by the signal processing module to determine whether there is an arc fault.
6. The identification method according to claim 5, characterized in that, In the processing unit, the signal output by the signal processing module is first converted by AD, and then the presence or absence of an arc is determined by the fault arc detection algorithm.
7. The identification method according to claim 6, characterized in that, The process of determining the presence or absence of an arc using the fault arc detection algorithm is as follows: determine whether the sampling time of the current data has reached the sampling period. If it has reached the sampling period, process it into one-dimensional current data, extract representative features of the current, normalize the feature data, and then send it into the pre-trained arc fault detection model for detection.
8. The identification method according to claim 7, characterized in that, The arc fault detection model is an arc detection model based on a one-dimensional convolutional neural network.
9. The identification method according to any one of claims 5-8, characterized in that, Upon detecting an arc fault, a command is sent to the inverter to disconnect the photovoltaic circuit.
10. The identification method according to any one of claims 5-8, characterized in that, The signal processing module analyzes and processes the current signal by first performing differential processing on the current signal, then filtering out interference signals, and finally performing detection processing.