Photovoltaic system and arc fault detection method therefor

Abstract

A photovoltaic system and an arc fault detection method therefor, relating to the technical field of photovoltaics. The arc fault detection method of the photovoltaic system comprises: acquiring a current operation point of the photovoltaic system, and determining an operation area to which the current operation point belongs, wherein a full operation point range of the photovoltaic system is divided into at least two operation areas, at least some of the operation areas have corresponding arc fault detection models, and the confidence levels of arc fault detection results corresponding to different operation areas are different (S110); determining a target operation point on the basis of the current operation point and the operation area to which the current operation point belongs, wherein the confidence level of an arc fault detection result of an operation area to which the target operation point belongs is higher than or equal to that of the operation area to which the current operation point belongs (S120); and testing the target operation point on the basis of an arc fault detection model corresponding to the operation area to which the target operation point belongs, so as to determine whether an arc fault occurs in the photovoltaic system (S130).

Description

Photovoltaic systems and their arc fault detection methods

[0001] This application claims priority to Chinese Patent Application No. 202410480172.7, filed with the Chinese Patent Office on April 22, 2024, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of photovoltaic technology, for example to a photovoltaic system and a method for detecting arc faults therein. Background Technology

[0003] In photovoltaic systems, arcing faults can cause electrical fires and system failures. Current detection methods have limitations in sensitivity and specificity, posing a risk of false alarms or missed detections. For example, judging arcing solely by comparing the decibel value of a high-frequency signal to a threshold may lead to oversensitivity or insensitivity within certain current ranges, resulting in false alarms. Another method is to control the photovoltaic panel's output current to zero and then determine if an arc has occurred based on whether the operating state at the operating point can be restored; however, this method may miss faults if the arc is not completely melted. Still other methods use a single arcing model to evaluate all data, but this may lead to a significant decrease in model accuracy or oversensitivity within certain voltage and current ranges, resulting in missed or false alarms.

[0004] Summary of the Invention

[0005] This application provides a photovoltaic system and an arc fault detection method thereof to improve the accuracy and reliability of arc fault detection results in the photovoltaic system.

[0006] In a first aspect, embodiments of this application provide a method for detecting arc faults in a photovoltaic system, comprising:

[0007] The current operating point of the photovoltaic system is obtained, and the operating area to which the current operating point belongs is determined; wherein, the entire operating point range of the photovoltaic system is divided into at least two operating areas, and at least some operating areas have corresponding arc fault detection models; the confidence level of the arc fault detection results corresponding to different operating areas is different;

[0008] The target operation point is determined based on the current operation point and the operation area to which the current operation point belongs; wherein, the operation area to which the target operation point belongs has a corresponding arc fault detection model, and the confidence level of the arc fault detection result of the operation area to which the target operation point belongs is higher than or equal to the confidence level of the arc fault detection result of the operation area to which the current operation point belongs.

[0009] The target operating point is detected based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, and the photovoltaic system is determined to have an arc fault based on the detection results.

[0010] Secondly, embodiments of this application also provide a photovoltaic system, including:

[0011] Photovoltaic array;

[0012] A photovoltaic converter, connected to the photovoltaic array, is configured as an operating point to control the photovoltaic system;

[0013] A signal acquisition module, connected to the photovoltaic array, is configured to acquire the output voltage and output current of the photovoltaic array to determine the operating point of the photovoltaic system.

[0014] The control module is connected to the photovoltaic converter and the signal acquisition module respectively, and is configured to execute the arc fault detection method of the photovoltaic system provided in any embodiment of this application. Attached Figure Description

[0015] Figure 1 is a flowchart illustrating an arc fault detection method for a photovoltaic system provided in an embodiment of this application;

[0016] Figure 2 is a schematic diagram of the process of dividing the operating area and training an arc fault detection model according to an embodiment of this application;

[0017] Figure 3 is a flowchart illustrating another method for detecting arc faults in a photovoltaic system provided in an embodiment of this application;

[0018] Figure 4 is a flowchart illustrating another method for detecting arc faults in a photovoltaic system provided in an embodiment of this application;

[0019] Figure 5 is a diagram showing the division of an operating area according to an embodiment of this application;

[0020] Figure 6 is a diagram showing another division result of the operating area provided in an embodiment of this application;

[0021] Figure 7 is a diagram showing another division result of the operating area provided in an embodiment of this application;

[0022] Figure 8 is a schematic diagram of a photovoltaic system provided in an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of this application.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0025] This application provides a method for detecting arc faults in a photovoltaic system, which can accurately identify whether an arc fault has occurred on the DC side of the photovoltaic system. This method can be executed by the control module in the photovoltaic system. Figure 1 is a schematic flowchart of an arc fault detection method for a photovoltaic system provided in this application. As shown in Figure 1, the method includes the following steps.

[0026] S110. Obtain the current operating point of the photovoltaic system and determine the operating area to which the current operating point belongs; wherein, the entire operating point range of the photovoltaic system is divided into at least two operating areas, and at least some operating areas have corresponding arc fault detection models; the confidence level of the arc fault detection results corresponding to different operating areas is different.

[0027] A photovoltaic (PV) system may include a photovoltaic array and a photovoltaic converter. The photovoltaic array may include multiple photovoltaic modules, each of which may include at least one photovoltaic panel, configured to convert solar energy into electrical energy. The electrical signals generated by the multiple photovoltaic modules are aggregated and output from the output terminal of the photovoltaic array. The photovoltaic converter is configured to control the operating point of the photovoltaic system; the photovoltaic converter may include a solar controller and a photovoltaic inverter, such as a Maximum Power Point Tracking (MPPT) solar controller. The operating point of the photovoltaic system may include the output information of the photovoltaic array, such as the current signal (PV current or Ipv) and voltage signal (PV voltage or Vpv) output by the photovoltaic array. In other words, the current operating point of the photovoltaic array can be obtained by acquiring the current and voltage signals currently output by the photovoltaic array. Correspondingly, the full operating point range may be an operating point range based on the full voltage (the voltage range between the minimum and maximum operating voltages of the photovoltaic array) and the full current (the current range between the minimum and maximum operating currents of the photovoltaic array), covering all possible operating points under normal operating conditions of the photovoltaic system. An operating area includes at least one operating point.

[0028] Furthermore, the photovoltaic system may also include a signal acquisition module and a control module. The signal acquisition module is configured to collect the electrical parameters output by the photovoltaic array to determine the operating point of the photovoltaic system, such as acquiring the voltage and current of the photovoltaic array according to a preset sampling frequency. The control module is connected to both the signal acquisition module and the photovoltaic converter, and is configured to obtain the operating point of the photovoltaic system through the signal acquisition module, adjust the operating point of the photovoltaic system through the photovoltaic converter, and execute the arc fault detection method.

[0029] For example, the confidence level of the arc fault detection result corresponding to an operating area can be used to characterize the ease or difficulty of detecting whether an arc fault has occurred at the operating point in that operating area. For instance, a higher confidence level indicates that arc fault detection at the operating point in that operating area is simpler and the detection result is more reliable. For an operating area with a corresponding arc fault detection model, the confidence level of the arc fault detection result corresponding to that operating area can be understood as the confidence level of the detection result output by the arc fault detection model corresponding to that operating area. For an operating area without a corresponding arc fault detection model, the confidence level of the arc fault detection result corresponding to that operating area can be considered as 0. For example, for each operating area with an arc fault detection model, there is a one-to-one correspondence between the operating area and the arc fault detection model, and the confidence levels of the detection results output by the arc fault detection models corresponding to different operating areas are different. The confidence level of the arc fault detection result corresponding to an operating area can be represented by the statistical value of the confidence level of the arc fault detection results corresponding to the operating point in that operating area, such as the minimum or average value.

[0030] Arc fault detection models can be trained and constructed using methods such as machine learning and neural networks. For example, normal data and arcing data during DC arcing of a photovoltaic system across the entire operating range can be obtained in advance through simulation or actual measurement. The spectral characteristics under normal and arcing conditions can be analyzed in the frequency domain. By comparing the signal strength of different frequency bands, feature values ​​for neural network models can be extracted. These feature values ​​can be used to establish a sample library and to construct arc fault detection models.

[0031] S120. Determine the target operating point based on the current operating point and the operating area to which the current operating point belongs; wherein, the operating area to which the target operating point belongs has a corresponding arc fault detection model, and the confidence level of the arc fault detection result of the operating area to which the target operating point belongs is higher than or equal to the confidence level of the arc fault detection result of the operating area to which the current operating point belongs.

[0032] For ease of explanation, the operating area to which the current operating point belongs will be referred to as the current operating area, and the operating area to which the target operating point belongs will be referred to as the target operating area.

[0033] In this step, because a corresponding arc fault detection model exists in the target operating area, the current state of the photovoltaic system and whether arcing has occurred can be detected and identified, preventing missed detections. For example, if there is no corresponding arc fault detection model in the current operating area, switching the operating point to the target operating point ensures that whether arcing has occurred on the DC side of the photovoltaic system can be detected. Furthermore, by setting the confidence level of the arc fault detection results in the target operating area to be no lower than that in the current operating area, the reliability and credibility of the arc fault detection results can be guaranteed.

[0034] S130. Based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, the target operating point is detected, and the photovoltaic system is determined to have an arc fault based on the detection results.

[0035] This step can be as follows: extract relevant features from the target operation point and input them into the arc fault detection model corresponding to the target operation area. The model output is the detection result, which can be used to determine whether an arc fault has occurred in the photovoltaic system.

[0036] The arc fault detection method for photovoltaic systems provided in this application divides the entire operating point range of the photovoltaic system into at least two operating regions. The confidence level of the arc fault detection result corresponding to the operating region is used to characterize the difficulty of arc fault detection in that operating region. A target operating point selection step is set to ensure that the confidence level of the arc fault detection result of the operating region to which the target operating point belongs is not lower than the confidence level of the arc fault detection result of the operating region to which the current operating point belongs. Arc fault identification and detection are performed using the target operating point and its corresponding arc fault detection model. This method can ensure that whether an arc is currently occurring in the photovoltaic system can be detected, and it can also ensure the reliability of the detection results, minimizing missed detections and false detections, and improving the accuracy and reliability of arc fault detection results in the photovoltaic system.

[0037] Based on the above-mentioned multiple implementation methods, optionally, the target operating point can be detected based on the arc fault detection model corresponding to the operating area to which the target operating point belongs. This can be achieved by: bringing the detection signal in decibel format corresponding to the target operating point into the arc fault detection model corresponding to the operating area to which the target operating point belongs for detection.

[0038] The process of acquiring the decibel-formatted detection signal includes: extracting the current signal from the target operating point; amplifying the signal corresponding to the preset frequency range in the current signal to obtain the characteristic signal; performing analog-to-digital conversion on the characteristic signal to obtain the characteristic digital signal; and performing a Fast Fourier Transform (FFT) on the characteristic digital signal to obtain the decibel-formatted detection signal. When an arc fault occurs on the DC side of a photovoltaic system, the current signal of the photovoltaic array changes significantly in the frequency domain, exhibiting a certain amount of high-frequency components. The detection signal obtained based on FFT processing can intuitively provide signal strength information (in decibels (dB)) within the preset frequency range. Arc fault detection based on the spectral signal can reliably guarantee the accuracy of the detection results. Correspondingly, each arc fault detection model is also constructed based on the decibel-formatted signal obtained after processing the original data through the same steps described above. For example, the preset frequency range can be a frequency band with obvious arc characteristics, which can be obtained through experience or prior experiments. Amplifying the signal corresponding to the preset frequency range in the current signal can be divided into the following steps: filtering the current signal to retain the signal within the preset frequency range, and amplifying the filtered signal to obtain the characteristic signal.

[0039] In this embodiment, arc fault detection is performed by combining time-domain and frequency-domain information, which can improve the reliability of the detection results. The operating area to which the current operating point belongs is determined by the time-domain information, and the target operating point is determined. Then, the frequency-domain information of the target operating point is obtained and used into the arc fault detection model of the target operating area for detection, taking into account both the time-domain and frequency-domain characteristics of the arc.

[0040] Based on the above-described embodiments, optionally, when the target operating point is different from the current operating point, before detecting the target operating point based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, the method further includes: switching the operating point of the photovoltaic system to the target operating point. For example, the operating point can be adjusted by adjusting the current signal and / or voltage signal of the photovoltaic array.

[0041] Based on the above-described embodiments, optionally, the target operating point can be the operating point in the operating area with the highest confidence level of the arc fault detection result, so as to maximize the reliability of the detection result.

[0042] And / or, the target operating point can be an operating point located on the maximum power point tracking curve of the photovoltaic system to ensure the output power of the photovoltaic array. This setting ensures accurate detection of arc faults while minimizing power loss, balancing the safety and efficiency of the photovoltaic system operation.

[0043] Based on the above-described multiple implementation methods, there are alternative ways to divide the operating area and construct the arc fault detection model. Several of these methods will be described below, but they are not intended to limit this application.

[0044] In one implementation, optionally, the operating area can be divided according to the difficulty of arc identification, and then arc fault detection models can be constructed for each of the multiple operating areas. This may include:

[0045] 1) Take multiple points within the entire operating point range, collect normal data and arc data from multiple points, and divide the entire operating point range into at least two operating areas based on the difference between the normal data and the arc data.

[0046] The sampling points in this step can cover as many operating points as possible to establish an arc fault sample library across the entire voltage and current range. Regions can be divided based on the frequency domain difference between normal and arc fault data. For example, the maximum and minimum values ​​of the difference between normal and arc fault data can be extracted to form a difference interval, which can then be divided into multiple segments. Operating points where normal and arc fault data fall within the same difference interval are grouped into the same operating region. Since larger difference intervals are easier to detect arc faults and yield more reliable results, a higher confidence level for arc fault detection results can be directly assigned to intervals with larger difference values.

[0047] 2) For any operating area, train an arc fault detection model for that operating area based on the relevant data of the operating points within that operating area.

[0048] For example, model training can be performed based on arcing data from multiple operating points within the same operating area; the training method is not limited here. The training methods for different arc fault detection models can be the same or different. For example, when the confidence level of the output result of an arc fault detection model in an operating area is higher than a preset threshold, the model training is considered complete. The preset threshold can be set according to actual needs, and the preset thresholds for different operating areas can be the same or different. For example, the preset threshold can be set to be higher for operating areas with higher confidence levels in arc fault detection results. When the confidence level of an arc fault detection model in an operating area consistently fails to reach the corresponding preset threshold, it can be determined that the operating area has no corresponding arc fault detection model; when multiple operating areas have no corresponding arc fault detection models, the multiple operating areas without corresponding models can be merged into one operating area.

[0049] The following example, using the division of the entire operating point range into three operating regions, illustrates the model construction process. For ease of explanation, the three operating regions are referred to as the first operating region, the second operating region, and the third operating region, respectively, and the corresponding arc fault detection models are referred to as the first arc fault detection model, the second arc fault detection model, and the third arc fault detection model, respectively. Figure 2 is a schematic flowchart of an operating region division and arc fault detection model training provided in an embodiment of this application. Referring to Figure 2, the model construction process includes the following steps.

[0050] S201, Obtain the operation point.

[0051] This step is equivalent to the input phase: the system takes current and voltage signals from the photovoltaic array as input data, which directly reflects the array's electrical performance. This is the starting point of the entire arc fault detection process. For example, this step can acquire relevant data from one operating point, or it can acquire relevant data from multiple operating points simultaneously. The relevant data from multiple operating points can be used to execute subsequent steps simultaneously or at different times.

[0052] S202, Extract the current signal from the operating point.

[0053] This step is equivalent to current processing: the current signal is first processed by a current sensing circuit. This circuit not only accurately detects the current flowing through the photovoltaic system, but also serves as the primary link in the signal processing chain, providing the foundation for subsequent steps.

[0054] S203. Filter and amplify the current signal to obtain the characteristic signal.

[0055] This step is equivalent to signal filtering and amplification: the output of the current sensing circuit is fed into a bandpass filter and a signal amplifier. At this stage, the system filters out unwanted noise and amplifies signals corresponding to a specific frequency range (preset frequency range) that corresponds to the characteristics of the arc. This step is crucial for accurately identifying key signal characteristics associated with arc faults.

[0056] S204. Perform analog-to-digital conversion on the characteristic signal to obtain the characteristic digital signal.

[0057] In this step, the amplified analog signal is converted into digital format by an analog-to-digital converter circuit, preparing it for subsequent processing by the control module (e.g., a microcontroller).

[0058] S205. Perform a fast Fourier transform on the characteristic digital signal to obtain a signal in decibel format.

[0059] This step can be performed in a microcontroller, which performs a fast Fourier transform on the digital signal to convert it into decibel (dB) format, thereby describing the signal's relative strength to a reference level.

[0060] S206. Determine if the operation point belongs to the first operation area. If the operation point belongs to the first operation area, execute S207; if the operation point does not belong to the first operation area, execute S208.

[0061] Whether the operating point belongs to the first operating region can be determined based on the values ​​of the current signal and the voltage signal. The same applies to S208 and S210, which will not be elaborated further.

[0062] S207. Construct the first arc fault detection model using a signal in decibel format.

[0063] S208. Determine whether the operation point belongs to the second operation area. If the operation point belongs to the second operation area, then execute S209; if the operation point does not belong to the second operation area, then execute S210.

[0064] S209. Construct a second arc fault detection model using a decibel-formatted signal. If the operation point does not belong to the third operating area, proceed to S211; if the operation point does not belong to the third operating area, proceed to S212.

[0065] S211. A third arc fault detection model is constructed using a signal in decibel format.

[0066] S212, Error in marking data.

[0067] S206-S212 outline the model generation decision process: It determines whether the operating point conforms to different predefined 'scaling' ranges (i.e., operating regions) based on time-domain parameters (current and voltage values). If the operating point falls within the first operating region, a first arc fault detection model is constructed, and subsequent identification is specifically performed for signal characteristics within this range. Similarly, if the operating point falls within the second operating region, a second arc fault detection model is constructed, and so on. If the time-domain parameters of the operating point do not conform to any operating region, it is marked as "data error," indicating that the signal does not meet the expected parameters for arc fault detection, and further analysis or other measures may be required.

[0068] This embodiment, through steps S201-S210, demonstrates how to train multiple arc fault detection models based on the characteristics of the current and voltage signals of a photovoltaic array. This process includes data acquisition, filtering, and algorithm analysis of the high-frequency signals superimposed on the DC arc during the arcing process. Model feature values ​​and samples are extracted through time-domain and frequency-domain analysis, followed by neural network training and machine learning to ultimately form multiple arc fault detection models. The trained models are then configured to analyze real-time data from the photovoltaic system for effective arc fault detection.

[0069] In another implementation, optionally, multiple points can be selected across the entire operating point range to collect normal data and arcing data from these points. Multiple preset models are then trained based on all the data. The operating region is divided according to the training results, and arc fault detection models corresponding to multiple operating regions are simultaneously obtained. For example, multiple confidence levels can be pre-defined and sequentially denoted as the first confidence level, the second confidence level, ..., the nth confidence level; this serves as the standard for region division. After training, when the confidence level of a model for arc identification results at a subset of operating points reaches at least the nth confidence level, these operating points can be classified into the same operating region, and the confidence level of the arc fault detection results for that operating region is recorded as the nth confidence level value. For the remaining data, after training, when the confidence level of a model for arc identification results at a subset of operating points reaches at least the (n-1)th confidence level, these operating points can be classified into the same operating region, and the confidence level of the arc fault detection results for that operating region is recorded as the (n-1)th confidence level value, and so on. If the data of some operation points cannot meet the first confidence level regardless of the model used or how it is trained, the remaining operation points can be classified into the same operation area, and the confidence level of the arc fault detection result of the operation area can be recorded as 0.

[0070] Based on the above-described implementation methods, optionally, the time domain and spectrum of DC arcing can vary significantly under different weather conditions, arc location, voltage and current ranges, and power converter noise levels. For any photovoltaic system, after its components and installation environment are determined, a database and model can be established to facilitate arc detection during subsequent operation.

[0071] Based on the above-described embodiments, optionally, the arc fault detection method further includes updating the arc fault detection model corresponding to each operating area when the operating conditions of the photovoltaic system meet the model update conditions, or re-dividing the operating areas of the photovoltaic system and determining the arc fault detection model corresponding to each re-divided operating area. This configuration allows the arc fault detection model to be updated in a timely manner following changes in operating conditions, ensuring the accuracy of detection results throughout the system's entire lifecycle.

[0072] For example, the model update conditions may be: the system has been running for a preset time since the last model update, or an event that may affect the accuracy of the test results has occurred, such as the replacement of components in the photovoltaic system, so as to fully consider the impact of component aging and replacement in the photovoltaic system during operation.

[0073] The above-described embodiments exemplify the construction of multiple arc fault detection modules. The process of determining the target operating point is described below.

[0074] Optionally, based on the above-described embodiments, S120 includes:

[0075] When the operating area to which the current operating point belongs is the operating area with the highest confidence level of the arc fault detection result, the current operating point is taken as the target operating point.

[0076] When the operating area to which the current operating point belongs is not the operating area with the highest confidence level of the arc fault detection result:

[0077] If there is no corresponding arc fault detection model in the operating area to which the current operating point belongs, then the operating point in the operating area with the highest confidence of the arc fault detection result is selected as the target operating point.

[0078] If a corresponding arc fault detection model exists in the operating area to which the current operating point belongs, then the target operating point is determined based on the operating point holding conditions corresponding to the operating area to which the current operating point belongs.

[0079] The target operation point is determined based on the operation point hold conditions corresponding to the operation area to which the current operation point belongs, including:

[0080] Determine whether the current operation point meets the operation point hold condition corresponding to the operation area to which the current operation point belongs;

[0081] If the current operation point meets the operation point hold condition corresponding to the operation area to which the current operation point belongs, then the current operation point will be used as the target operation point.

[0082] If the current operation point does not meet the operation point holding condition corresponding to the operation area to which the current operation point belongs, then the operation point in the operation area corresponding to the arc fault detection model with the highest confidence of the arc fault detection result is selected as the target operation point.

[0083] The operating point holding condition for any operating region can be: (DB-Threshold) / Threshold>P; where DB is the intensity of the current signal in the current operating point at a preset frequency, and the preset frequency is within a preset frequency range; Threshold is the preset intensity threshold corresponding to the operating region to which the current operating point belongs at the preset frequency, and P is the preset percentage corresponding to the operating region to which the current operating point belongs at the preset frequency.

[0084] When the percentage of the difference between the actual intensity at the current operating point and the preset intensity threshold at the preset frequency is higher than the preset intensity threshold, it can be considered that the data differs significantly from the normal data at that operating point. The reliability of identification based on the arc fault detection model corresponding to this operating area is high, and the model can be directly used for identification without switching operating points to avoid impacting system operation. Conversely, when the percentage of the difference between the actual intensity at the current operating point and the preset intensity threshold at the preset frequency is lower than the preset intensity threshold, it can be considered that the data differs only slightly from the normal data at that operating point. In this case, the operating point needs to be switched to an operating area with higher confidence in the arc fault detection results to ensure the reliability of the detection results.

[0085] Based on the above-described embodiments, optionally, determining the operating area to which the current operating point belongs may include:

[0086] According to the preset operation area order, it is determined whether the current operation point belongs to the current operation area. The preset operation area order is the order of confidence of the arc fault detection results corresponding to the operation area from low to high. The confidence of the arc fault detection results of the operation area without a corresponding arc fault detection model can be recorded as 0.

[0087] Based on the above-described embodiments, optionally, before determining the operating area of ​​the current operating point, the method further includes: determining whether the current operating point meets the arc fault judgment triggering condition; if the current operating point meets the arc fault judgment triggering condition, then the step of determining the operating area of ​​the current operating point is executed; if the current operating point does not meet the arc fault judgment triggering condition, then the operating point of the photovoltaic system is obtained according to a preset sampling frequency. For example, the arc fault judgment triggering condition is equivalent to the initial judgment process, and its judgment condition is more lenient than the judgment condition for operating point adjustment. Meeting this condition before proceeding to the steps of operating area judgment and operating point adjustment can effectively avoid the impact of frequent operating point changes on the operational stability of the photovoltaic system. For instance, the arc fault judgment triggering condition can adopt the same judgment form as the operating point holding condition, the difference being that the preset percentage in the arc fault judgment triggering condition can be lower than the preset percentage in the multiple operating point holding conditions. When the preset percentage in the arc fault judgment triggering condition is exceeded, it can be considered that an arc fault has occurred in the system, and the steps of operating area judgment, operating point adjustment, and arc fault identification are initiated.

[0088] The following example illustrates the fault detection process by dividing the entire operating point range into three operating areas, with the confidence level of the arc fault detection results in the first, second, and third operating areas increasing sequentially.

[0089] Figure 3 is a flowchart illustrating another method for detecting arc faults in a photovoltaic system provided in an embodiment of this application. Referring to Figure 3, in one embodiment, optionally, multiple operating regions each have corresponding arc fault detection models. The operating point holding condition corresponding to the nth operating region is denoted as the nth operating point holding condition, and the arc fault detection model corresponding to the nth operating region is denoted as the nth arc fault detection model. The method includes the following steps.

[0090] S310, Get the current operation point.

[0091] S320. Determine if the current operation point belongs to the first operation area. If the current operation point belongs to the first operation area, execute S330; if the current operation point does not belong to the first operation area, execute S370.

[0092] S330. Determine whether the current operation point meets the first operation point hold condition. If the current operation point meets the first operation point hold condition, then execute S340; if the current operation point does not meet the first operation point hold condition, then execute S350.

[0093] S320, in combination with S330, essentially provides decision points related to the first operating area, including: (a) Data check: First, check the real-time data to determine whether the voltage and current signals of the operating point meet the standards of the first operating area range. (b) Threshold comparison: Compare the data with the set first operating point holding conditions and determine whether to adjust the operating point.

[0094] S340. The first arc fault detection model is used for fault detection.

[0095] S350, Adjust the target operation point to the third operation area.

[0096] S360 uses a third arc fault detection model for fault detection.

[0097] S370. Determine if the current operation point belongs to the second operation area. If the current operation point belongs to the second operation area, execute S380; if the current operation point does not belong to the second operation area, execute S360.

[0098] When the operation point does not conform to either the first operation area or the second operation area, the operation point is defaulted to the third operation area, and the third arc fault detection model is used for fault judgment. The third operation area is no longer judged, which simplifies the detection process.

[0099] S380. Determine whether the current operation point meets the second operation point hold condition. If the current operation point meets the second operation point hold condition, then execute S390; if the current operation point does not meet the second operation point hold condition, then execute S350.

[0100] The combination of S370 and S380 essentially provides decision points related to the second operating area. The principle is similar to that of decision points related to the first operating area, so it will not be elaborated further.

[0101] S390. The second arc fault detection model is used for fault detection.

[0102] The embodiments of this application explain the decision logic used in real-time fault detection through S310-S390, focusing on how to select the most suitable arc model under different conditions.

[0103] Figure 4 is a flowchart illustrating another arc fault detection method for a photovoltaic system provided in this application. Both the method shown in Figure 3 and the method shown in Figure 4 aim to select a suitable arc fault detection model in real-time fault detection, but they differ in their decision-making logic. The difference between Figure 4 and Figure 3 is that when the operation point is determined to be located in the first operation area, the operation point is directly adjusted to the third operation area. The method shown in Figure 4 is applicable, for example, to situations where arc faults are difficult to detect in the first operation area, or where there is no corresponding arc fault detection model in the first operation area. For instance, when factors such as the length of the photovoltaic panel cable and the noise floor of the photovoltaic converter significantly affect data collection, making it difficult to establish an effective arc fault identification model and resulting in a high false positive rate. In these situations, directly adjusting the operation point to the third operation area for judgment is more reasonable.

[0104] Referring to Figure 4, the method includes the following steps.

[0105] S410, Get the current operation point.

[0106] S420. Determine if the current operation point belongs to the first operation area. If the current operation point belongs to the first operation area, execute S430; if the current operation point does not belong to the first operation area, execute S450.

[0107] S430, Adjust the target operation point to the third operation area.

[0108] S440. The third arc fault detection model is used for fault detection.

[0109] S450. Determine whether the current operation point belongs to the second operation area. If the current operation point belongs to the second operation area, execute S460; if the current operation point does not belong to the second operation area, execute S440.

[0110] S460. Determine whether the current operation point meets the second operation point hold condition. If the current operation point meets the second operation point hold condition, then execute S470; if the current operation point does not meet the second operation point hold condition, then execute S430.

[0111] S470. The second arc fault detection model is used for fault detection.

[0112] This application embodiment uses another fault detection decision logic provided by S410-S470.

[0113] In summary, these two methods provide a flexible approach to optimize the fault detection process of photovoltaic systems, thereby ensuring the efficient operation and safety of the systems.

[0114] Based on the above-described embodiments, optionally, to avoid frequent operation point adjustments, after determining that the current operation point does not meet the operation point holding condition corresponding to the operation area to which the current operation point belongs, the method may further include:

[0115] The determination result is whether the number of consecutive times that the current operation point does not meet the operation point holding condition corresponding to the operation area to which the current operation point belongs has reached the first preset number;

[0116] If the number of consecutive times that the current operation point does not meet the operation point holding condition corresponding to the operation area to which the current operation point belongs reaches the first preset number, then the operation point in the operation area corresponding to the arc fault detection model with the highest confidence of the arc fault detection result is selected as the target operation point.

[0117] If the number of consecutive times that the current operation point does not meet the operation point holding condition corresponding to the operation area to which the current operation point belongs has not reached the first preset number, then the collection and judgment of the current operation point will continue.

[0118] The first preset number of times can be set according to actual needs, such as 3-8 times, or even 5 times. This setting can avoid the influence of interference factors. The operation point will only be switched when the operation point needs to be adjusted after several consecutive determinations. This can ensure the stability of system operation while ensuring the reliability of fault detection.

[0119] Based on the above-described multiple implementation methods, optionally, to avoid misjudgment, the target operating point is detected based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, and the determination of whether an arc fault has occurred in the photovoltaic system is based on the detection results, which may include:

[0120] An arc fault is determined to have occurred in the photovoltaic system when the detection results are all arc faults in n consecutive sampling periods; n is a positive integer greater than 1.

[0121] For example, n can be 3-8, or even 3. This setting avoids the influence of interference factors, and only when multiple consecutive detection results indicate an arc fault does the photovoltaic system definitively determine that an arc fault has occurred, thus avoiding misjudgment.

[0122] It should be noted that the content and order of the multiple steps in the above embodiments are only illustrative examples and are not intended to limit this application. In actual applications, the order can be changed and the steps can be added or removed according to the needs. For example, in Figure 2, the process of determining which operation area the operation point belongs to (i.e., S206, S208 and S210) can be performed simultaneously or in other orders, such as before, after, or simultaneously with S205, etc.

[0123] Based on the above-described embodiments, optionally, the time domain and spectrum of DC arcing can vary significantly under different weather conditions, arc location, voltage and current ranges, and power converter noise levels. The above embodiments describe the division of the arc fault detection area, model establishment, and detection process in a solar photovoltaic system. These operational area divisions are based on comprehensive acquisition and analysis of voltage and current data from the photovoltaic system within specific frequency bands, the establishment of a DC arcing sample library, and neural network model training. Through in-depth analysis of voltage and current data, these data can be divided into different regions with small, medium, and large differences. The following example, dividing the entire range into three operational areas, illustrates the detection process using the three division methods shown in Figures 5-7. It should be noted that these operational area divisions may vary depending on different photovoltaic converters, operating conditions, and system operating states. Therefore, specific modeling and analysis based on actual on-site data are necessary. By comprehensively collecting voltage and current data from the photovoltaic system, analyzing and processing data within specific frequency bands, establishing a DC arcing sample library, and training neural network models, the entire operating range can be divided into three operating regions. Figure 5 uses full-range voltage and current analysis as the basis for this division, Figure 6 uses voltage as the basis, and Figure 7 uses current as the basis, all dividing the entire operating range into three operating regions. The first operating region, zone I, corresponds to a region with small arc signal characteristics, high detection difficulty, and low confidence. The second operating region, zone II, represents a region with medium arc signal characteristics, medium detection difficulty, and medium confidence. The third operating region, zone III, represents a region with large arc signal characteristics, low detection difficulty, and high confidence. Different operating regions are displayed using different filling styles: diagonally filled areas represent the first operating region, zone I; square-filled areas represent the second operating region, zone II; and dotted-filled areas represent the third operating region, zone III. The solid black curve represents the photovoltaic system's operating characteristic curve, which is the current-voltage (IV) characteristic curve, and the dashed black curve represents the maximum power point tracking curve.

[0124] Referring to Figure 5, in one embodiment, the division of the operating region is based on a sample library under arcing conditions of full voltage (minimum and maximum operating voltage) and full current (minimum and maximum operating current), and a neural network model is built and trained using the arcing sample library. After the data from the arcing group and the normal group are input into the model, three operating regions are divided according to the difference in the model output results. Operating point A11 represents an operating point where an arc may exist. It is located in the first operating region (zoon I). It can be considered that the time-domain and frequency-domain characteristics of the operating point in this operating region are not very obvious during arcing, or that the distinguishability is insufficient. Therefore, the operating point is directly adjusted, and the operating point in the third operating region (zoon III) is selected as the target operating point. Alternatively, after determining that the operating point A11 does not meet the first operating point holding condition, the operating point can be adjusted, for example, the operating point can be adjusted from A11 to the target operating point B11. The target operating point B11 is located in the third operating region (zoon III), where arc faults are easily detected, for more accurate fault judgment. The situation at operating point A12 is similar to that at operating point A11; fault diagnosis can also be performed by adjusting the operating point, for example, adjusting it to the target operating point B12. The target operating point can be achieved by adjusting the voltage and current. Since operating point A13 is located in the second operating region (zoon II), if, after assessment, the characteristics of the arc signal are obvious and meet the conditions for maintaining the second operating point, the arc fault detection model in the second operating region (zoon II) can continue to be used for fault diagnosis. If the conditions for maintaining the second operating point are not met, operating point A13 still needs to be adjusted, and an operating point in the third operating region (zoon III) should be selected as the target operating point.

[0125] Referring to Figure 6, in another embodiment, the division can still be based on a sample library under full voltage and full current arcing conditions. Operating point A21 operates at the maximum power point and is located in the third operating region (zoon III), where arc faults are easily detected. When arc detection is required, the third arc fault detection model can be directly used for judgment. Furthermore, for operating points not in the third operating region (zoon III), if based on the operating point holding conditions corresponding to the operating region, it is determined that operating points A22, A23, and A24 all need adjustment, then these operating points can be adjusted to the third operating region (zoon III), for example, to target operating points B22, B23, and B24 respectively, for more accurate arc fault judgment. The target operating point can be achieved by adjusting the operating voltage. For operating points not in the third operating region (zoon III), if their arc signal characteristics are obvious and meet the corresponding operating point holding conditions, the arc fault detection model of their original operating region can continue to be used for fault judgment, without adjusting the operating point.

[0126] Referring to Figure 7, in another embodiment, the division can still be based on a sample library under full-voltage and full-current arcing conditions. Operating point A31 is located in the third operating region (zoon III), where arc faults are easily detected, and no operating point switching is required. If operating point A31 operates at the maximum power point tracking (MPPT), the current output power and efficiency of the photovoltaic system can still be guaranteed. Furthermore, for operating points not located in the third operating region (zoon III), if it is determined that operating points A32 and A33 both need adjustment based on the corresponding operating point holding conditions, these operating points can be adjusted to the third operating region (zoon III), for example, to target operating points B32 and B33 respectively, for more accurate arc fault detection. The target operating point can be achieved by adjusting the operating voltage and operating current. For operating points not located in the third operating region (zoon III), if their arc signal characteristics are obvious and meet the corresponding operating point holding conditions, the arc fault detection model of their original operating region can continue to be used for fault detection without operating point adjustment.

[0127] In summary, this application provides a leading dynamic intelligent arc fault detection method. This method is based on different operating regions of the current-voltage (IV) curve, offering a flexible and efficient fault detection mechanism for solar photovoltaic systems. While ensuring high accuracy in fault diagnosis, it reduces power loss and helps optimize system operating efficiency and safety. This method is applicable to all series of string inverters. Based on the analysis of the DC current arcing process in photovoltaic systems, this method simulates an arc by superimposing a high-frequency vibration signal onto the PV current. By acquiring DC arcing signals and performing Fast Fourier Transform (FFT) operations, feature values ​​are extracted in both the time and frequency domains to establish a DC arc sample library, which is then used for neural network training. Thus, by dividing multiple operating regions and establishing multiple arc fault detection models, effective DC arcing detection can be achieved using machine learning and pattern recognition technologies. Furthermore, during the decision-making process, the system operating point can be adjusted in real time based on current and voltage measurements to apply the most suitable arc model for accurate arc fault determination.

[0128] This application also provides a photovoltaic system, and the arc fault detection method for the photovoltaic system provided in any embodiment of this application can be applied to this photovoltaic system. Figure 8 is a schematic diagram of the structure of a photovoltaic system provided in an embodiment of this application. Referring to Figure 8, the photovoltaic system includes: a photovoltaic array 10, a photovoltaic converter 20, a signal acquisition module 30, and a control module 40. The photovoltaic converter 20 is connected to the photovoltaic array 10 and is configured to control the operating point of the photovoltaic system; the signal acquisition module 30 is connected to the photovoltaic array 10 and is configured to acquire the output voltage and output current of the photovoltaic array 10 to determine the operating point of the photovoltaic system; the control module 40 is connected to the photovoltaic converter 20 and the signal acquisition module 30 respectively and is configured to execute the arc fault detection method of the photovoltaic system.

[0129] For example, the control module 40 may include a microcontroller configured with FFT operation functionality. The signal acquisition module 30 may include a current sensing circuit, a bandpass filter, a signal amplifier, and an analog-to-digital converter. The signal acquisition module 30 and the control module 40 may be configured to detect photovoltaic current and voltage signals. The sensed current signal, after being processed by the bandpass filter and signal amplifier, is converted into a digital signal by the analog-to-digital converter. Then, the microcontroller uses a Fast Fourier Transform (FFT) to process these digital signals, analyzing the spectral characteristics in the frequency domain under arcing and non-arcting conditions. By comparing the signal strength in different frequency bands, feature values ​​for a neural network model are extracted. These feature values ​​are used to build a sample library and to construct an arc fault detection model, where signal strength is expressed in decibels (dB).

[0130] It should be noted that in several embodiments of the arc fault detection method for photovoltaic systems, some descriptions of the structure of the photovoltaic system have been provided. For any content not described here, please refer to the explanations of the above embodiments. Repeated content will not be repeated.

[0131] It should be understood that the various processes shown above can be used to rearrange, add, or delete steps. For example, the multiple steps described in this application can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this application can be achieved, and this is not limited herein.

Claims

1. A method for detecting arc faults in a photovoltaic system, comprising: The current operating point of the photovoltaic system is obtained, and the operating area to which the current operating point belongs is determined; wherein, the entire operating point range of the photovoltaic system is divided into at least two operating areas, and at least some operating areas have corresponding arc fault detection models; the confidence level of the arc fault detection results corresponding to different operating areas is different; The target operation point is determined based on the current operation point and the operation area to which the current operation point belongs; wherein, the operation area to which the target operation point belongs has a corresponding arc fault detection model, and the confidence level of the arc fault detection result of the operation area to which the target operation point belongs is higher than or equal to the confidence level of the arc fault detection result of the operation area to which the current operation point belongs. The target operating point is detected based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, and the photovoltaic system is determined to have an arc fault based on the detection results.

2. The method for detecting arc faults in a photovoltaic system according to claim 1, wherein, Determining the target operation point based on the current operation point and the operation area to which the current operation point belongs includes: If the operating area to which the current operating point belongs is the operating area with the highest confidence level of the arc fault detection result, the current operating point shall be taken as the target operating point. If the operating area to which the current operating point belongs is not the operating area with the highest confidence level of the arc fault detection result: In response to the fact that there is no corresponding arc fault detection model in the operating area to which the current operating point belongs, the operating point in the operating area with the highest confidence of the arc fault detection result is selected as the target operating point; In response to the existence of a corresponding arc fault detection model in the operating area to which the current operating point belongs, the target operating point is determined according to the operating point holding conditions corresponding to the operating area to which the current operating point belongs.

3. The method for detecting arc faults in a photovoltaic system according to claim 2, wherein, The step of determining the target operation point based on the operation point holding conditions corresponding to the operation area to which the current operation point belongs includes: Determine whether the current operation point meets the operation point holding condition corresponding to the operation area to which the current operation point belongs; In response to the current operation point satisfying the operation point holding condition corresponding to the operation area to which the current operation point belongs, the current operation point is taken as the target operation point; In response to the current operating point not meeting the operating point holding condition corresponding to the operating area to which the current operating point belongs, the operating point in the operating area with the highest confidence of the arc fault detection result is selected as the target operating point.

4. The method for detecting arc faults in a photovoltaic system according to claim 3, wherein, The operating point holding condition includes: (DB-Threshold) / Threshold > P; Wherein, DB is the intensity of the current signal at the current operating point at a preset frequency, Threshold is the preset intensity threshold corresponding to the operating area to which the current operating point belongs at the preset frequency, and P is the preset percentage corresponding to the operating area to which the current operating point belongs at the preset frequency.

5. The arc fault detection method for a photovoltaic system according to claim 3, further comprising, in response to the current operating point not meeting the operating point hold condition corresponding to the operating area to which the current operating point belongs: The determination result is whether the number of consecutive times that the current operation point does not meet the operation point holding condition corresponding to the operation area to which the current operation point belongs reaches a first preset number; In response to the current operating point not meeting the operating point holding condition corresponding to the operating area to which the current operating point belongs for a first preset number of consecutive times, the operating point in the operating area with the highest confidence of the arc fault detection result is selected as the target operating point.

6. The method for detecting arc faults in a photovoltaic system according to claim 1, wherein, The step of detecting the target operating point based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, and determining whether the photovoltaic system has experienced an arc fault based on the detection results, includes: If the detection results for n consecutive sampling periods are all of arc fault, the photovoltaic system is determined to have an arc fault; n is a positive integer greater than 1.

7. The method for detecting arc faults in a photovoltaic system according to claim 1, wherein, The target operating point is located on the maximum power point tracking curve of the photovoltaic system.

8. The method for detecting arc faults in a photovoltaic system according to claim 1, wherein, If the target operating point is different from the current operating point, before detecting the target operating point based on the arc fault detection model corresponding to the operating area to which the target operating point belongs, the method further includes: switching the operating point of the photovoltaic system to the target operating point.

9. The method for detecting arc faults in a photovoltaic system according to claim 1, wherein, The detection of the target operating point based on the arc fault detection model corresponding to the operating area to which the target operating point belongs includes: The detection signal in decibel format corresponding to the target operating point is fed into the arc fault detection model corresponding to the operating area to which the target operating point belongs for detection. The process of acquiring the detection signal in decibel format includes: Extract the current signal from the target operating point; Amplify the current signal corresponding to a preset frequency range to obtain a characteristic signal; The characteristic signal is subjected to analog-to-digital conversion to obtain the characteristic digital signal; The characteristic digital signal is subjected to a fast Fourier transform to obtain the detection signal in decibel format.

10. The method for detecting arc faults in a photovoltaic system according to claim 1, further comprising, before determining the operating area to which the current operating point belongs: Determine whether the current operation point meets the arc fault detection triggering conditions; In response to the current operating point meeting the arc fault judgment triggering condition, the step of determining the operating area to which the current operating point belongs is executed; In response to the current operating point not meeting the arc fault judgment triggering condition, the operating point of the photovoltaic system is obtained according to the preset sampling frequency.

11. The method for detecting arc faults in a photovoltaic system according to claim 1 further includes, when the operation of the photovoltaic system meets the model update conditions, updating the arc fault detection model corresponding to each of the operating areas, or, re-dividing the operating areas of the photovoltaic system and determining the arc fault detection model corresponding to each of the re-divided operating areas.

12. A photovoltaic system, comprising: Photovoltaic array; A photovoltaic converter, connected to the photovoltaic array, is configured as an operating point to control the photovoltaic system; A signal acquisition module, connected to the photovoltaic array, is configured to acquire the output voltage and output current of the photovoltaic array in order to determine the operating point of the photovoltaic system; The control module is connected to the photovoltaic converter and the signal acquisition module respectively, and is configured to execute the arc fault detection method of the photovoltaic system as described in any one of claims 1-11.

Images