Arc detection method and device for photovoltaic system and photovoltaic system

Abstract

The present application discloses an arc detection method, device and photovoltaic system for a photovoltaic system, and belongs to the technical field of photovoltaic systems. The arc detection method for the photovoltaic system comprises: performing fast Fourier transform on the sampled values ​​of the output sampling signal of the photovoltaic system to obtain the amplitude-frequency information and phase-frequency information of the output sampling signal; filtering the sampled values ​​to obtain the time domain information of the output sampling signal; performing feature extraction on the amplitude-frequency information, phase-frequency information and time domain information respectively to obtain the amplitude-frequency feature vector, phase-frequency feature vector and time domain feature vector; inputting the amplitude-frequency feature vector, phase-frequency feature vector and time domain feature vector into the target classification model to obtain the arc detection result output by the target classification model. The arc detection method for the photovoltaic system of the present application obtains arc detection results based on more comprehensive features, improves the accuracy of arc detection, and reduces missed detection and false detection.

Description

Technical Field

[0001] The present application belongs to the technical field of photovoltaic systems, and in particular, relates to an arc detection method and device for a photovoltaic system, and a photovoltaic system. Background Art

[0002] When the current noise of the photovoltaic system is large, the accuracy of arc detection is low, and it is easy to misdetect under long cables and high current. In the related technology, arc detection is mainly carried out by collecting noise data and arc data, training artificial intelligence algorithms. However, the data features extracted by the artificial intelligence algorithm are not comprehensive, and the extracted features are uncontrollable, making it difficult to obtain key features in harsh scenarios, resulting in misdetection and missed detection in the arc detection results, and low arc detection accuracy. Summary of the invention

[0003] The present application aims to solve at least one of the technical problems existing in the related art. To this end, the present application proposes an arc detection method, device and photovoltaic system for a photovoltaic system, which obtains arc detection results based on more comprehensive features, improves the accuracy of arc detection, and reduces missed detection and false detection.

[0004] In a first aspect, the present application provides an arc detection method for a photovoltaic system, the method comprising:

[0005] Performing fast Fourier transform on the sampled values ​​of the output sampled signal of the photovoltaic system to obtain amplitude-frequency information and phase-frequency information of the output sampled signal; performing filtering on the sampled values ​​to obtain time domain information of the output sampled signal;

[0006] Extracting features from the amplitude-frequency information, the phase-frequency information and the time-domain information respectively to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector;

[0007] Inputting the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector into a target classification model to obtain an arc detection result output by the target classification model, wherein the arc detection result includes: the presence of an arc or the absence of an arc;

[0008] The target classification model is trained using historical amplitude-frequency feature vectors, historical phase-frequency feature vectors and historical time-domain feature vectors as samples and historical arc detection results as sample labels.

[0009] According to the arc detection method for a photovoltaic system of the present application, by performing data processing on the sampled values ​​of the output sampling signal of the collected photovoltaic system, amplitude-frequency information, phase-frequency information and time domain information are obtained, and more comprehensive information with different attributes is obtained. Feature extraction is performed on the information with different attributes to obtain the feature vector of each attribute, that is, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are obtained. The feature vectors of different attributes are processed through a target classification model, and arc detection results are obtained based on more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detections and false detections.

[0010] According to the arc detection method for a photovoltaic system of the present application, the fast Fourier transform is performed on the sampled value of the output sampling signal of the photovoltaic system to obtain the amplitude-frequency information and phase-frequency information of the output sampling signal, including:

[0011] Performing the fast Fourier transform on the sampled value of the output sampled signal of the photovoltaic system to obtain initial amplitude-frequency information and initial phase-frequency information of the output sampled signal;

[0012] The noise frequency bands in the initial amplitude-frequency information and the initial phase-frequency information are deleted respectively to obtain the amplitude-frequency information and the phase-frequency information.

[0013] According to the arc detection method for a photovoltaic system of the present application, the filtering process on the sampled value to obtain the time domain information of the output sampled signal includes:

[0014] The sampling values ​​are subjected to notch processing and bandpass filtering to obtain the time domain information.

[0015] According to the arc detection method for a photovoltaic system of the present application, the noise frequency band in the sampled value is subjected to the notch processing to obtain time domain information after the notch processing;

[0016] Performing the bandpass filtering process on the time domain information after the notch processing to obtain the time domain information;

[0017] or;

[0018] Performing the bandpass filtering process on the sampled values ​​to obtain time domain information after the bandpass filtering process;

[0019] A notch process is performed on the noise frequency band in the time domain information after the bandpass filtering process to obtain the time domain information.

[0020] According to the arc detection method for a photovoltaic system of the present application, the feature extraction is performed on the amplitude-frequency information, the phase-frequency information and the time domain information respectively to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time domain feature vector, including:

[0021] The amplitude-frequency information is input into a first feature extraction model to obtain the amplitude-frequency feature vector output by the first feature extraction model; the first feature extraction model is trained using historical amplitude-frequency information as samples and arc detection results corresponding to the historical amplitude-frequency information as sample labels;

[0022] The phase-frequency information is input into a second feature extraction model to obtain the phase-frequency feature vector output by the second feature extraction model; the second feature extraction model is trained using historical phase-frequency information as samples and arc detection results corresponding to the historical phase-frequency information as sample labels;

[0023] The time domain information is input into a third feature extraction model to obtain the time domain feature vector output by the third feature extraction model; the third feature extraction model is trained using historical time domain information as samples and arc detection results corresponding to the historical time domain information as sample labels.

[0024] According to the arc detection method for a photovoltaic system of the present application, the inputting of the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector into a target classification model to obtain an arc detection result output by the target classification model includes:

[0025] Fusing any two of the amplitude-frequency feature vector, the phase-frequency feature vector, and the time-domain feature vector to obtain at least three fused feature matrices;

[0026] The at least three fused feature matrices are input into the target classification model to obtain the arc detection result output by the target classification model.

[0027] According to the arc detection method for a photovoltaic system of the present application, the fusion of any two feature vectors among the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector obtains at least three fusion feature matrices, including:

[0028] The element values ​​of the target position in the target feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector are added to each element value of another feature vector to obtain a sub-fusion feature vector corresponding to the target feature vector; the other feature vector is any feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector except the target feature vector;

[0029] The fused feature matrix is ​​constructed based on the sub-fused feature vectors corresponding to the target feature vector.

[0030] In a second aspect, the present application provides an arc detection device for a photovoltaic system, the device comprising:

[0031] The first processing module is used to perform fast Fourier transform on the sampled values ​​of the output sampled signal of the photovoltaic system to obtain the amplitude-frequency information and phase-frequency information of the output sampled signal; and perform filtering processing on the sampled values ​​to obtain the time domain information of the output sampled signal;

[0032] A second processing module is used to extract features from the amplitude-frequency information, the phase-frequency information and the time-domain information respectively to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector;

[0033] A third processing module is used to input the amplitude-frequency feature vector, the amplitude-frequency feature vector and the time domain feature vector into a target classification model to obtain an arc detection result output by the target classification model, wherein the arc detection result includes: the presence of an arc or the absence of an arc;

[0034] The target classification model is trained using historical amplitude-frequency feature vectors, historical phase-frequency feature vectors and historical time-domain feature vectors as samples and historical arc detection results as sample labels.

[0035] According to the arc detection device of the photovoltaic system of the present application, by performing data processing on the sampled values ​​of the output sampling signal of the collected photovoltaic system, amplitude-frequency information, phase-frequency information and time domain information are obtained, and more comprehensive information with different attributes is obtained. Feature extraction is performed on the information with different attributes to obtain the feature vector of each attribute, that is, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are obtained. The feature vectors of different attributes are processed through a target classification model, and arc detection results are obtained based on more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detections and false detections.

[0036] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the arc detection method for the photovoltaic system as described in the first aspect above is implemented.

[0037] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the arc detection method for a photovoltaic system as described in the first aspect above is implemented.

[0038] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the arc detection method for a photovoltaic system as described in the first aspect above.

[0039] The above one or more technical solutions in the embodiments of the present application have at least one of the following technical effects:

[0040] By processing the sampled values ​​of the output sampling signal of the collected photovoltaic system, the amplitude-frequency information, phase-frequency information and time domain information are obtained, and more comprehensive information with different attributes is obtained. Feature extraction is performed on the information with different attributes to obtain the feature vector of each attribute, that is, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are obtained. The feature vectors of different attributes are processed through the target classification model, and the arc detection result is obtained based on more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detection and false detection.

[0041] Furthermore, by performing fast Fourier transform on the sampling values ​​of the output sampling signal of the photovoltaic system, the initial amplitude-frequency information and initial phase-frequency information of the output sampling signal in the frequency domain are obtained, and by deleting the noise frequency band in the initial amplitude-frequency information and the initial phase frequency, the frequency band related to arc detection is highlighted to obtain the amplitude-frequency information and the phase-frequency information, thereby improving the data quality of the amplitude-frequency information and the phase-frequency information, reducing the influence of noise on arc detection, and improving the accuracy of subsequent arc detection.

[0042] Furthermore, by performing notch processing and bandpass filtering on the acquired sampling values, the time domain information of unnecessary frequency bands can be effectively weakened, the key frequency bands that can distinguish arcs from non-arcs can be selected, and the efficiency of feature extraction of time domain information can be improved.

[0043] Furthermore, by inputting the amplitude-frequency information, phase-frequency information and time domain information into the first feature extraction model, the second feature extraction model and the third feature extraction model one by one, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are effectively obtained, the characteristics of different attributes of the output sampling signal are obtained, more comprehensive features are obtained, and feature extraction is performed through the feature extraction model to improve the efficiency of feature extraction.

[0044] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0046] Figure 1 It is one of the flow charts of the arc detection method for a photovoltaic system provided in an embodiment of the present application;

[0047] Figure 2 This is one of the principle schematic diagrams of the arc detection method for a photovoltaic system provided in an embodiment of the present application;

[0048] Figure 3 This is the second flow chart of the arc detection method for a photovoltaic system provided in an embodiment of the present application;

[0049] Figure 4 This is the second principle schematic diagram of the arc detection method for a photovoltaic system provided in an embodiment of the present application;

[0050] Figure 5 This is the third principle schematic diagram of the arc detection method for a photovoltaic system provided in an embodiment of the present application;

[0051] Figure 6 This is the fourth principle schematic diagram of the arc detection method for a photovoltaic system provided in an embodiment of the present application;

[0052] Figure 7 is a schematic structural diagram of an arc detection device for a photovoltaic system provided in an embodiment of the present application;

[0053] Figure 8 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0055] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0056] In combination with the accompanying drawings, the arc detection method for a photovoltaic system, the arc detection device for a photovoltaic system, the photovoltaic system, the electronic device and the readable storage medium provided in the embodiments of the present application are described in detail below through specific embodiments and their application scenarios.

[0057] The arc detection method of the photovoltaic system may be applied to a terminal, and may be specifically executed by hardware or software in the terminal.

[0058] The terminal includes but is not limited to portable communication devices such as mobile phones or tablet computers. It should also be understood that in some embodiments, the terminal may not be a portable communication device, but a desktop computer.

[0059] In the following various embodiments, a terminal including a display and a touch-sensitive surface is described. However, it should be understood that the terminal may include one or more other physical user interface devices such as a physical keyboard, a mouse and a joystick.

[0060] The embodiment of the present application provides an arc detection method for a photovoltaic system, and the executor of the arc detection method for the photovoltaic system can be the photovoltaic system, or can be an arc detection device of the photovoltaic system arranged on the photovoltaic system, or can also be a server electrically connected to the photovoltaic system (wired or wireless connection), or can also be a user terminal communicatively connected to the photovoltaic system, including but not limited to mobile terminals and non-mobile terminals.

[0061] like Figure 1 As shown, the arc detection method of the photovoltaic system includes: step 110, step 120 and step 130.

[0062] A photovoltaic system includes photovoltaic modules.

[0063] Step 110: Perform fast Fourier transform on the sampled values ​​of the output sampled signal of the photovoltaic system to obtain amplitude-frequency information and phase-frequency information of the output sampled signal; perform filtering on the sampled values ​​to obtain time domain information of the output sampled signal;

[0064] In this step, the output sampling signal is a signal used to detect whether there is an arc in the photovoltaic system.

[0065] The output sampling signal may include: the output current of the photovoltaic system and the output voltage of the photovoltaic system.

[0066] The sampled value is the actual data of the output sampled signal obtained by sampling.

[0067] The sampling value of the output sampling signal of the photovoltaic system can be as follows Figure 2 Data shown.

[0068] Fast Fourier transform is to convert the sampled values ​​of the output sampled signal into the frequency domain.

[0069] The amplitude information and the phase information are the frequency domain related features of the output sampling signal.

[0070] The amplitude information is the amplitude of the output sampling signal in the frequency domain, reflecting the strength of the output sampling signal at each frequency.

[0071] The phase information is the phase information of each frequency component of the output sampling signal in the frequency domain, reflecting the time offset of the output sampling signal at each frequency component. The phase information can be represented by an angle.

[0072] In the actual implementation process, the output sampling signal of the photovoltaic system can be collected through devices such as current sensors, Hall elements, voltage sensors, and voltage sampling devices to obtain sampling values.

[0073] Of course, the sampling value of the output sampling signal of the photovoltaic system may also be obtained by any other feasible method, which is not limited in this application.

[0074] like Figure 3 As shown, after obtaining the sampling value of the output sampling signal, the sampling value can be subjected to fast Fourier transform through the fast Fourier transform algorithm, and the sampling value can be converted to the frequency domain to obtain the amplitude information and phase information of the output sampling signal, so as to perform frequency domain analysis on the output sampling signal.

[0075] Fast Fourier transform algorithms include, but are not limited to, Cooley-Tukey algorithm, Radix-2 algorithm, and Radix-4 algorithm.

[0076] In the actual implementation process, a suitable fast Fourier transform algorithm can be selected based on the actual situation of the output sampling signal, and this application does not limit it.

[0077] The filtering process is to weaken the frequency bands with unclear characteristics in the sampling values ​​of the output sampling signal and enhance the frequency bands with obvious characteristics in the sampling values.

[0078] The time domain information is the specific value of the sampling value in the time domain.

[0079] After the sampling value of the output sampling signal is obtained, the sampling value can be filtered using a filtering algorithm to obtain the processed time domain information.

[0080] The filtering algorithms include but are not limited to: limiting filtering algorithm, arithmetic averaging filtering algorithm, recursive averaging filtering algorithm and de-jitter filtering algorithm.

[0081] In the actual implementation process, a suitable filtering algorithm can be selected based on actual usage requirements, and this application does not limit it.

[0082] Step 120, extracting features from the amplitude-frequency information, the phase-frequency information and the time-domain information respectively to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector;

[0083] In this step, the amplitude-frequency feature vector is a vector that can characterize the amplitude-frequency characteristics of the output sampling signal.

[0084] The phase-frequency characteristic vector is a vector that can characterize the phase-frequency characteristic of the output sampling signal.

[0085] The time domain feature vector is a vector that characterizes the characteristics of the output sampling signal in the time series.

[0086] In the actual execution process, feature extraction is performed on the amplitude-frequency information, and features such as the mean, standard deviation, maximum value, minimum value, spectrum bandwidth, and specific frequency band energy of the amplitude-frequency information can be extracted. The obtained mean, standard deviation, maximum value, spectrum bandwidth, and specific frequency band energy are arranged in a certain order to form a feature vector, thereby obtaining an amplitude-frequency feature vector.

[0087] By performing feature extraction on the phase-frequency information, features such as the mean, standard deviation, maximum value, minimum value, spectrum bandwidth, specific frequency band energy, phase difference between adjacent frequency points, and phase stability of the phase-frequency information can be extracted. The acquired features such as the mean, standard deviation, maximum value, minimum value, spectrum bandwidth, specific frequency band energy, phase difference between adjacent frequency points, and phase stability are arranged in a certain order to form a feature vector, thereby obtaining a phase-frequency feature vector.

[0088] By performing feature extraction on the time domain information, the mean, standard deviation, maximum value, minimum value, peak-to-peak value, skewness, kurtosis and short-time energy and other features of the time domain information can be extracted. The mean, standard deviation, maximum value, minimum value, peak-to-peak value, skewness, kurtosis and short-time energy and other features of the acquired time domain information are arranged in a certain order to form a feature vector, thereby obtaining a time domain feature vector.

[0089] It should be noted that the specific category of extracted features can be determined based on actual conditions and is not limited in this application.

[0090] Step 130: input the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector into the target classification model to obtain the arc detection result output by the target classification model.

[0091] In this step, the target classification model is a model for analyzing whether an arc exists in the input sampling signal.

[0092] The target classification model can be a classification model such as a convolutional neural network model, a machine learning model, and an artificial intelligence model.

[0093] The specific type of the target classification model can be determined based on actual conditions and is not limited in this application.

[0094] In some embodiments, the dimension of the target classification model output layer may be greater than or equal to 2.

[0095] When the dimension of the output layer of the target classification model is 2, the arc detection result includes: the presence of an arc or the absence of an arc.

[0096] When the dimension of the output layer of the target classification model is 3, the arc detection results include: the presence of an arc, the presence of ordinary non-arc noise, or the presence of severe non-arc noise.

[0097] The dimension of the target classification model output layer can be determined based on the actual working conditions. For example, when the features representing whether an arc exists are obvious in the sampling values, the dimension of the target classification model output layer can be set to 2. When the features representing whether an arc exists are not obvious in the sampling values, the dimension of the target classification model output layer can be set to 3. This is not limited in this application.

[0098] The target classification model is trained with historical amplitude-frequency feature vectors, historical phase-frequency feature vectors and historical time-domain feature vectors as samples and historical arc detection results as sample labels.

[0099] The historical amplitude-frequency feature vector is an amplitude-frequency feature vector obtained by extracting the amplitude-frequency information of the historical output sampling signal.

[0100] The historical output sampling signal is a sample output sampling signal collected before the output sampling signal is obtained.

[0101] The historical phase-frequency feature vector is a phase-frequency feature vector obtained by extracting features from the phase-frequency information of the historical output sampling signal.

[0102] The historical time domain feature vector is a time domain feature vector obtained by extracting features from the time domain information of the historical output sampling signal.

[0103] The historical arc detection results are the detection results corresponding to the historical output sampling signals.

[0104] When the dimension of the output layer of the target classification model is 2, the historical arc detection results include: the existence of an arc or the absence of an arc.

[0105] When the dimension of the output layer of the target classification model is 3, the historical arc detection results include: the presence of an arc, the presence of ordinary non-arc noise, or the presence of severe non-arc noise.

[0106] During the actual execution process, sampling values ​​of multiple historical output sampling signals can be collected, and each sampling value can be subjected to fast Fourier transform and filtering processing. Feature extraction is performed on the data after fast Fourier transform and filtering processing to obtain the historical amplitude-frequency feature vector, historical phase-frequency feature vector and historical time domain feature vector of each historical output sampling signal. The obtained historical amplitude-frequency feature vector, historical phase-frequency feature vector and historical time domain feature vector are used as samples, and the historical arc detection results are used as sample labels to train the target classification model.

[0107] Taking the target classification model as the AlexNet neural network model and the dimension of the output layer of the AlexNet neural network model as 2 as an example, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are input into the AlexNet neural network model. The AlexNet neural network model will give the classification result of the output sampling signal, that is, the output sampling signal has an arc, or the output sampling signal does not have an arc.

[0108] Taking the target classification model as the ResNet convolutional neural network model and the dimension of the output layer of the ResNet convolutional neural network model as 3 as an example, the amplitude-frequency feature vector, phase-frequency feature vector and time domain feature vector are input into the ResNet convolutional neural network model. The ResNet convolutional neural network model will give the classification result of the output sampling signal, that is, there is an arc in the output sampling signal, there is ordinary non-arc noise in the output sampling signal, or there is severe non-arc noise in the output sampling signal.

[0109] The inventors found during the research and development process that in the related technologies, arc detection is mainly performed by collecting noise data and arc data, training artificial intelligence algorithms. However, the data features extracted by the artificial intelligence algorithm are not comprehensive, and the extracted features are uncontrollable, making it difficult to obtain key features in harsh scenarios, such as drastic changes in lighting conditions, switching of photovoltaic systems or photovoltaic storage integrated machine operation modes, and inconsistent specifications of photovoltaic panels in photovoltaic systems, resulting in false detections and missed detections in arc detection results, and low arc detection accuracy.

[0110] The present application obtains information of the output sampling signal in the frequency domain, namely, amplitude-frequency information and phase-frequency information, by performing fast Fourier transform on the sampling values ​​of the output sampling signal, and performs filtering processing on the sampling values ​​to obtain time domain information of key information of the enhanced sampling values, and obtains information capable of analyzing the characteristics and laws of the output sampling signal from different angles; by performing feature extraction on the acquired amplitude-frequency information, phase-frequency information and time domain information, an amplitude-frequency feature vector, a phase-frequency feature vector and a time domain feature vector are obtained, and key information of the sampling signal on different characteristics is obtained, thereby processing the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector through a target classification model to obtain arc detection results, and performing arc detection through more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detections and false detections.

[0111] According to the arc detection method for a photovoltaic system provided in an embodiment of the present application, by performing data processing on the sampled values ​​of the output sampling signal of the collected photovoltaic system, amplitude-frequency information, phase-frequency information and time domain information are obtained, and more comprehensive information with different attributes is obtained. Feature extraction is performed on the information with different attributes to obtain a feature vector of each attribute, that is, an amplitude-frequency feature vector, a phase-frequency feature vector and a time domain feature vector are obtained. The feature vectors of different attributes are processed through a target classification model, and an arc detection result is obtained based on more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detections and false detections.

[0112] In some embodiments, step 110 may include:

[0113] Performing fast Fourier transform on the sampling value of the output sampling signal of the photovoltaic system to obtain initial amplitude-frequency information and initial phase-frequency information of the output sampling signal;

[0114] The noise frequency bands in the initial amplitude-frequency information and the initial phase-frequency information are deleted respectively to obtain the amplitude-frequency information and the phase-frequency information.

[0115] In this embodiment, the initial amplitude-frequency information is the amplitude-frequency information including a frequency point or frequency band with relatively large noise.

[0116] The initial phase-frequency information is the phase-frequency information including the frequency point or frequency band with relatively large noise.

[0117] In the actual implementation process, after the sampling value of the output sampling signal is obtained, the sampling value can be processed by a fast Fourier algorithm to obtain initial amplitude-frequency information and initial phase-frequency information.

[0118] In some embodiments, performing a fast Fourier transform on the sampled values ​​of the output sampled signal of the photovoltaic system to obtain the initial amplitude-frequency information and the initial phase-frequency information of the output sampled signal may also include:

[0119] Performing fast Fourier transform on the sampled values ​​of the output sampled signal of the photovoltaic system to obtain the amplitude-frequency information of the complete frequency band and the phase-frequency information of the complete frequency band;

[0120] The amplitude-frequency information of the target frequency band range is extracted from the complete frequency band amplitude-frequency information to obtain the initial amplitude-frequency information, and the phase-frequency information of the target frequency band range is extracted from the complete frequency band phase-frequency information to obtain the initial phase-frequency information.

[0121] In this embodiment, the complete frequency band is a frequency band determined based on a sampling frequency at which the output sampling signal is sampled.

[0122] The full frequency band amplitude-frequency information is the distribution information of the amplitude in the full frequency band obtained by performing fast Fourier transform on the sampling values.

[0123] The phase-frequency information of the complete frequency band is the distribution information of the phase in the complete frequency band obtained by performing fast Fourier transform on the sampling values.

[0124] The target frequency band range is the frequency band range with obvious arc characteristics in the complete frequency band.

[0125] The target frequency band range may be user-defined or determined based on actual conditions. For example, the target frequency band range may be 10kHz to 60kHz or 20kHz to 70kHz, which is not limited in this application.

[0126] The initial amplitude-frequency information is the amplitude-frequency information with obvious arc detection characteristics.

[0127] The initial phase-frequency information is the phase-frequency information with obvious arc detection characteristics.

[0128] In the actual implementation process, after the sampling value of the output sampling signal is obtained, the sampling value is subjected to a fast Fourier transform to obtain the amplitude-frequency information of the complete frequency band and the phase-frequency information of the complete frequency band.

[0129] The amplitude-frequency information of 10kHz to 60kHz is selected from the amplitude-frequency information of the complete frequency band to obtain the initial amplitude-frequency information.

[0130] The phase-frequency information of 10kHz to 60kHz is selected from the phase-frequency information of the complete frequency band to obtain the initial phase-frequency information.

[0131] According to the arc detection method for a photovoltaic system provided in an embodiment of the present application, the amplitude-frequency information and phase-frequency information of the entire frequency band are obtained by acquiring the amplitude-frequency information of the complete frequency band and the phase-frequency information of the complete frequency band, and the amplitude-frequency information and phase-frequency information of the target frequency band range are further extracted from the amplitude-frequency information of the complete frequency band and the phase-frequency information of the complete frequency band to obtain the initial amplitude-frequency information and the initial phase-frequency information, thereby effectively deleting the frequency band range where the arc characteristics are not obvious, reducing the amount of data that needs to be processed in the subsequent arc detection process, and improving the arc detection efficiency.

[0132] The noise frequency band is the frequency band within the target range of the amplitude or phase protrusion frequency point.

[0133] The specific value of the target range can be determined based on actual conditions. For example, the target range can be 20 Hz-50 Hz, or 100 Hz-200 Hz, which is not limited in this application.

[0134] For example Figure 2 The fast Fourier transform of the sampled values ​​shown in Figure 4 The initial amplitude-frequency information shown, Figure 4 The three protruding peaks are the noise frequency band.

[0135] The noise frequency bands in the initial amplitude-frequency information and the initial phase-frequency information are deleted respectively to obtain the amplitude-frequency information and the phase-frequency information.

[0136] In the actual implementation process, Figure 4 Taking the initial amplitude-frequency information shown as an example, the noise frequency bands are the frequency bands within the 20kHz target range, the frequency bands within the 40kHz target range, and the frequency bands within the 60kHz target range. The frequency bands within the 20kHz target range, the frequency bands within the 40kHz target range, and the frequency bands within the 60kHz target range in the 10kHz to 60kHz frequency band can be deleted to obtain the amplitude-frequency information.

[0137] Similarly, after obtaining the initial phase-frequency information, the frequency band within the 20 kHz target range, the frequency band within the 40 kHz target range, and the frequency band within the 60 kHz target range in the 10 kHz to 60 kHz frequency band may be deleted to obtain the phase-frequency information.

[0138] According to the arc detection method for a photovoltaic system provided in an embodiment of the present application, by performing a fast Fourier transform on the sampling values ​​of the output sampling signal of the photovoltaic system, initial amplitude-frequency information and initial phase-frequency information of the output sampling signal in the frequency domain are obtained, and by deleting the noise frequency bands in the initial amplitude-frequency information and the initial phase frequency, the frequency bands related to arc detection are highlighted to obtain amplitude-frequency information and phase-frequency information, thereby improving the data quality of the amplitude-frequency information and the phase-frequency information, reducing the influence of noise on arc detection, and improving the accuracy of subsequent arc detection.

[0139] In some embodiments, step 110 may further include:

[0140] The sampling values ​​are processed by notch processing and bandpass filtering to obtain time domain information.

[0141] In this step, notch processing is used to attenuate or suppress the noise frequency band.

[0142] In actual implementation, notch processing can be performed by means of a notch processor and a notch band-stop filtering algorithm.

[0143] Bandpass filtering is used to pass signals within a target frequency band and attenuate signals outside the target frequency band.

[0144] The specific selection of the target frequency band can be determined based on actual conditions and is not limited in this application.

[0145] In the actual implementation process, bandpass filtering can be performed by means of a bandpass filter processor and a digital bandpass filter.

[0146] In the actual execution process, after obtaining the sampling value, the sampling value can be input into the notch processor to quickly attenuate the input signal at a certain frequency point to block the passage of this frequency signal, and the sampling value can be input into the bandpass filter processor to allow the signal of the target frequency band to pass through to attenuate signals of other frequencies.

[0147] In the actual implementation process, the time domain information after the notch processing and the time domain information after the bandpass filtering processing can be fused, and the fused time domain information can be determined as the time domain information for subsequent arc detection.

[0148] In the actual implementation process, Figure 5 The sampled values ​​shown in the figure are processed by notch processing and bandpass filtering, and the following Figure 6 The time domain information is shown.

[0149] When the sampling value includes 100 data points, after notch processing and bandpass filtering processing, the obtained time domain information also includes 100 data points.

[0150] According to the arc detection method for a photovoltaic system provided in an embodiment of the present application, by performing notch processing and bandpass filtering on the acquired sampling values, time domain information of unnecessary frequency bands is effectively weakened, key frequency bands that can distinguish between arcs and non-arcs are selected, and the efficiency of feature extraction of time domain information is improved.

[0151] In some embodiments, before the sampling values ​​are notched and bandpass filtered to obtain time domain information, the method may further include:

[0152] The sampling values ​​within the target frequency band range are extracted from the sampling values ​​to obtain the initial time domain information.

[0153] In this embodiment, the initial time domain information is time domain information with obvious arc detection characteristics.

[0154] The frequency band range of the initial time domain information may be the same as the frequency band corresponding to the initial amplitude-frequency information and the initial phase-frequency information.

[0155] In the actual implementation process, time domain information of 10kHz to 60kHz can be selected from the sampled values ​​to obtain initial time domain information.

[0156] According to the arc detection method for a photovoltaic system provided in an embodiment of the present application, by extracting sampling values ​​within a target frequency band from the sampling values, time domain information with obvious arc detection characteristics can be effectively selected.

[0157] In some embodiments, performing notch processing and bandpass filtering on the sampled values ​​to obtain time domain information may also include:

[0158] Perform notch processing on the noise frequency band in the sampling value to obtain time domain information after notch processing;

[0159] The time domain information after the notch processing is subjected to bandpass filtering to obtain the time domain information.

[0160] In this embodiment, the time domain information after notch processing is the time domain information obtained after attenuating the signal in the noise frequency band.

[0161] During the actual execution process, the sampling value can be first input into the notch processor, and the signal in the noise frequency band can be weakened by the notch processor to obtain the time domain information after the notch processing. The time domain information after the notch processing can be input into the bandpass filter processor to obtain the time domain information of the frequency band with obvious arc detection characteristics.

[0162] Continuing with the example of the noise frequency bands being the frequency bands within the 20kHz target range, the frequency bands within the 40kHz target range, and the frequency bands within the 60kHz target range, notch processing can be performed on the frequency bands within the 20kHz target range, the frequency bands within the 40kHz target range, and the frequency bands within the 60kHz target range to obtain time domain information after notch processing to weaken the noise frequency band.

[0163] The time domain information after notch processing is subjected to bandpass filtering, and the time domain information after notch processing is the time domain information in the frequency band of 10kHz to 60kHz. The time domain information after notch processing in the frequency band of 10kHz to 60kHz can be subjected to bandpass filtering to obtain the time domain information in the frequency band of 10kHz to 60kHz after bandpass filtering, that is, the time domain information used for subsequent arc detection.

[0164] In some embodiments, performing notch processing and bandpass filtering on the sampled values ​​to obtain time domain information may also include:

[0165] Performing bandpass filtering on the sampled values ​​to obtain time domain information after bandpass filtering;

[0166] The noise frequency band in the time domain information after the bandpass filtering is notched to obtain the time domain information.

[0167] In this embodiment, the time domain information after the bandpass filtering process is the time domain information obtained by selecting the signals within the target frequency band and attenuating the signals outside the target frequency band.

[0168] During the actual execution process, the sampling value can be first input into the bandpass filter processor, and the signal within the target frequency band can be selected through the bandpass filter processor, and the signal outside the target frequency band can be attenuated to obtain the time domain information after the bandpass filter processing. The time domain information after the bandpass filter processing can be input into the notch processor to obtain the time domain information of the frequency band with obvious arc detection characteristics.

[0169] Taking the sampling value frequency band range of 10kHz to 60kHz, and the noise frequency band being the frequency band within the target range of 20kHz, the frequency band within the target range of 40kHz, and the frequency band within the target range of 60kHz as an example, the sampling values ​​in the frequency band of 10kHz to 60kHz can be first band-pass filtered by a band-pass filter processor to obtain time domain information after band-pass filtering, and the time domain information after band-pass filtering is input into a notch processor, and notch processing is performed on the frequency band within the target range of 20kHz, the frequency band within the target range of 40kHz, and the frequency band within the target range of 60kHz in the time domain information after band-pass filtering to obtain the required time domain information.

[0170] According to the arc detection method for a photovoltaic system provided in an embodiment of the present application, a noise frequency band in a sampling value is first notched to obtain time domain information after the notch processing, and the time domain information after the notch processing is band-pass filtered to obtain time domain information, or the sampling value is first band-pass filtered to obtain time domain information after the band-pass filtering processing, and the time domain information after the band-pass filtering processing is notched to obtain time domain information. A variety of processing methods are provided to weaken frequency bands with unclear characteristics and retain frequency bands with obvious characteristics. The processing method is flexible and can reduce interference from irrelevant data, thereby improving arc detection efficiency and arc detection accuracy.

[0171] Continue to refer Figure 2 In some embodiments, step 120 may further include:

[0172] Inputting the amplitude-frequency information into the first feature extraction model to obtain the amplitude-frequency feature vector output by the first feature extraction model;

[0173] Inputting the phase-frequency information into the second feature extraction model to obtain a phase-frequency feature vector output by the second feature extraction model;

[0174] The time domain information is input into the third feature extraction model to obtain the time domain feature vector output by the third feature extraction model.

[0175] In this embodiment, the first feature extraction model is trained using historical amplitude-frequency information as samples and arc detection results corresponding to the historical amplitude-frequency information as sample labels.

[0176] The historical amplitude-frequency information is the amplitude-frequency information obtained by performing fast Fourier transform on the sampling values ​​of the historical output sampling signal.

[0177] During the actual execution process, a large number of signals with and without arcs can be collected, and fast Fourier transform can be performed on the arc signals and non-arc signals to obtain historical amplitude-frequency information of each signal. The dimension of the historical amplitude-frequency information can be 500. Multiple historical amplitude-frequency information and arc detection results corresponding to each historical amplitude-frequency information are used as the amplitude-frequency information training set of the first feature extraction model to train the first feature extraction model.

[0178] The amplitude-frequency information is input into the trained first feature extraction model to obtain the amplitude-frequency feature vector output by the first feature extraction model.

[0179] The second feature extraction model is trained using historical phase-frequency information as samples and arc detection results corresponding to the historical phase-frequency information as sample labels.

[0180] The historical phase-frequency information is the phase-frequency information obtained by performing fast Fourier transform on the sampling values ​​of the historical output sampling signal.

[0181] During the actual execution process, a large number of signals with and without arcs can be collected, and fast Fourier transform can be performed on the arc signals and non-arc signals to obtain the historical phase-frequency information of each signal. The dimension of the historical phase-frequency information can be 500. Multiple historical phase-frequency information and the arc detection results corresponding to each historical phase-frequency information are used as the phase-frequency information training set of the second feature extraction model to train the second feature extraction model.

[0182] The phase-frequency information is input into the trained second feature extraction model to obtain the phase-frequency feature vector output by the second feature extraction model.

[0183] The third feature extraction model is trained using the historical time domain information as samples and the arc detection results corresponding to the historical time domain information as sample labels.

[0184] The historical time domain information is the time domain information obtained by filtering the sampling values ​​of the historical output sampling signals.

[0185] During the actual execution process, a large number of signals with and without arcs can be collected to obtain the historical time domain information of each arc signal and the historical time domain information of non-arc signals. The dimension of the historical time domain information can be 4096. The multiple historical time domain information and the arc detection results corresponding to each historical time domain information are used as the time domain information training set of the third feature extraction model to train the third feature extraction model.

[0186] The time domain information is input into the trained third feature extraction model to obtain the time domain feature vector output by the third feature extraction model.

[0187] The first feature extraction model, the second feature extraction model and the third feature extraction model may be models such as convolutional neural networks and machine learning models.

[0188] Of course, the first feature extraction model, the second feature extraction model and the third feature extraction model may also be any other models that can distinguish whether an arc exists in the target information or not.

[0189] The target information includes: amplitude-frequency information, phase-frequency information and time domain information.

[0190] Taking the CNN-1 model as the first feature extraction model, the CNN-2 model as the second feature extraction model, and the RNN model as the third feature extraction model as an example, the fully connected layers of the CNN-1 model, the CNN-2 model, and the RNN model can be exactly the same, including two sublayers, the number of neurons in the output layer is 2, and the number of neurons in the previous layer of the output layer is 64.

[0191] The amplitude-frequency information, phase-frequency information and time domain information of the output sampling signal are respectively input into CNN-1, CNN-2 and RNN for feature extraction. The feature extraction location is in the layer before the output layer of the fully connected layer, that is, the layer with 64 neurons.

[0192] The amplitude-frequency information, phase-frequency information and time-domain information are converted into amplitude-frequency feature vectors, phase-frequency feature vectors and time-domain feature vectors respectively, and the dimensions of the amplitude-frequency feature vectors, phase-frequency feature vectors and time-domain feature vectors are the same, which can be 64.

[0193] According to the arc detection method for photovoltaic systems provided in an embodiment of the present application, by inputting the amplitude-frequency information, phase-frequency information and time domain information into the first feature extraction model, the second feature extraction model and the third feature extraction model one by one, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are effectively obtained, the characteristics of different attributes of the output sampling signal are obtained, more comprehensive features are obtained, and feature extraction is performed through the feature extraction model to improve the efficiency of feature extraction.

[0194] Continue to refer Figure 2 In some embodiments, step 130 may further include:

[0195] Fusing any two eigenvectors among the amplitude-frequency eigenvector, the phase-frequency eigenvector and the time-domain eigenvector to obtain at least three fused feature matrices;

[0196] At least three fused feature matrices are input into a target classification model to obtain an arc detection result output by the target classification model.

[0197] In this embodiment, the fused feature matrix is ​​a feature matrix obtained by including any two features of amplitude-frequency information, phase-frequency information, and time-domain information.

[0198] In some embodiments, a fused feature matrix may be obtained by directly concatenating any two feature vectors among the amplitude-frequency feature vector, the phase-frequency feature vector, and the time-domain feature vector.

[0199] In some other embodiments, an outer product operation may be performed on any two eigenvectors among the amplitude-frequency eigenvector, the phase-frequency eigenvector and the time-domain eigenvector to obtain a fused feature matrix.

[0200] In the actual implementation process, the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector can be fused through a feature fusion algorithm to obtain a fused feature matrix.

[0201] Of course, in the actual implementation process, any two feature vectors among the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector can be fused in any other feasible way, and this application does not limit this.

[0202] By fusing the amplitude-frequency feature vector and the phase-frequency feature vector, a fused feature matrix can be obtained.

[0203] By fusing the phase-frequency eigenvector and the amplitude-frequency eigenvector, a fused feature matrix can be obtained.

[0204] By fusing the amplitude-frequency feature vector and the time-domain feature vector, a fused feature matrix can be obtained.

[0205] By fusing the time domain feature vector and the amplitude-frequency feature vector, a fused feature matrix can be obtained.

[0206] By fusing the phase-frequency feature vector and the time-domain feature vector, a fused feature matrix can be obtained.

[0207] By fusing the time domain feature vector and the phase-frequency feature vector, a fused feature matrix can be obtained.

[0208] It can be understood that when the values ​​in the amplitude-frequency eigenvector are different from the values ​​in the phase-frequency eigenvector, the amplitude-frequency eigenvector and the phase-frequency eigenvector are fused, and the resulting fused feature matrix is ​​different from the fused feature matrix obtained by fusion of the phase-frequency eigenvector and the amplitude-frequency eigenvector.

[0209] Similarly, if the fusion order of any other two feature vectors is different, the resulting fusion feature matrix will also be different.

[0210] After obtaining multiple fusion features, three fusion feature matrices with different feature vectors used in the multiple fusion feature matrices can be obtained and input into the target classification model to obtain the arc detection result output by the target classification model.

[0211] According to the arc detection method for photovoltaic systems provided in the embodiments of the present application, by fusing any two feature vectors among the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector, multiple fused feature matrices can be obtained, effectively fusing information of two different attributes, and inputting the fused feature vector into the target classification model to obtain the arc detection result output by the target classification model. There is no need to train multiple classification models for different feature vectors and further analyze the detection results of each classification model, thereby reducing the cost of arc detection, simplifying and facilitating the operation, and improving the user experience.

[0212] In some embodiments, any two eigenvectors among the amplitude-frequency eigenvector, the phase-frequency eigenvector, and the time-domain eigenvector are fused to obtain at least three fused feature matrices, which may further include:

[0213] The element value of the target position in the target feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector is added to each element value of another feature vector to obtain a sub-fusion feature vector corresponding to the target feature vector;

[0214] Based on the sub-fusion feature vectors corresponding to the target feature vector, a fusion feature matrix is ​​constructed.

[0215] In this embodiment, the target feature vector is any feature vector among the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector.

[0216] The other eigenvector is any eigenvector among the amplitude-frequency eigenvector, the phase-frequency eigenvector and the time-domain eigenvector except the target eigenvector.

[0217] The target position is any position in the target feature vector.

[0218] Taking the target feature vector as the time domain feature vector and the other feature vector as the amplitude-frequency feature vector as an example, when the target position is the first element value of the target feature vector, the first element value of the time domain feature vector can be added to the element values ​​in the amplitude-frequency feature vector to obtain a sub-fusion feature vector, that is, the element values ​​of the first row of the fusion feature vector.

[0219] When the target position is the second element value of the target feature vector, the second element value of the time domain feature vector can be added to each element value in the amplitude-frequency feature vector to obtain a sub-fusion feature vector, that is, each element value of the second row of the fusion feature vector.

[0220] Through the above processing method, the element values ​​in the time domain feature vector and the amplitude-frequency feature vector are fused to obtain a fused feature vector.

[0221] It should be noted that if the target feature vector is different from another feature vector, the obtained fused feature vector will also be different.

[0222] According to the arc detection method for photovoltaic systems provided in an embodiment of the present application, a sub-fusion feature vector corresponding to the target feature vector is obtained by adding the element value of the target position in the target feature vector to the element value of each other feature vector. Based on multiple sub-fusion feature vectors, a fusion feature matrix is ​​obtained. While fusing features of different attributes, the correspondence between features of different attributes and the relationship between adjacent values ​​of the same feature are expressed to obtain richer information.

[0223] The arc detection method for a photovoltaic system provided in the embodiment of the present application can be performed by an arc detection device for a photovoltaic system. The arc detection method for a photovoltaic system performed by an arc detection device for a photovoltaic system is used as an example to illustrate the arc detection device for a photovoltaic system provided in the embodiment of the present application.

[0224] An embodiment of the present application also provides an arc detection device for a photovoltaic system.

[0225] like Figure 7 As shown, the arc detection device of the photovoltaic system includes: a first processing module 710 , a second processing module 720 and a third processing module 730 .

[0226] The first processing module 710 is used to perform fast Fourier transform on the sampled values ​​of the output sampled signal of the photovoltaic system to obtain the amplitude-frequency information and phase-frequency information of the output sampled signal; and perform filtering on the sampled values ​​to obtain the time domain information of the output sampled signal;

[0227] The second processing module 720 is used to extract features from the amplitude-frequency information, the phase-frequency information and the time-domain information to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector;

[0228] The third processing module 730 is used to input the amplitude-frequency feature vector, the amplitude-frequency feature vector and the time-domain feature vector into the target classification model to obtain the arc detection result output by the target classification model, and the arc detection result includes: the existence of an arc or the absence of an arc;

[0229] Among them, the target classification model is trained with historical amplitude-frequency feature vectors, historical phase-frequency feature vectors and historical time-domain feature vectors as samples and historical arc detection results as sample labels.

[0230] According to the arc detection device for a photovoltaic system provided in an embodiment of the present application, by performing data processing on the sampled values ​​of the output sampling signal of the collected photovoltaic system, amplitude-frequency information, phase-frequency information and time domain information are obtained, and more comprehensive information with different attributes is obtained. Feature extraction is performed on the information with different attributes to obtain a feature vector of each attribute, that is, an amplitude-frequency feature vector, a phase-frequency feature vector and a time domain feature vector are obtained. The feature vectors of different attributes are processed through a target classification model, and an arc detection result is obtained based on more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detections and false detections.

[0231] In some embodiments, the first processing module 710 may also be used to:

[0232] Performing fast Fourier transform on the sampling value of the output sampling signal of the photovoltaic system to obtain initial amplitude-frequency information and initial phase-frequency information of the output sampling signal;

[0233] The noise frequency bands in the initial amplitude-frequency information and the initial phase-frequency information are deleted respectively to obtain the amplitude-frequency information and the phase-frequency information.

[0234] In some embodiments, the first processing module 710 may also be used to:

[0235] The sampling values ​​are processed by notch processing and bandpass filtering to obtain time domain information.

[0236] In some embodiments, the first processing module 710 may also be used to:

[0237] Perform notch processing on the noise frequency band in the sampling value to obtain time domain information after notch processing;

[0238] Performing bandpass filtering on the time domain information after notch processing to obtain time domain information;

[0239] or;

[0240] Performing bandpass filtering on the sampled values ​​to obtain time domain information after bandpass filtering;

[0241] The noise frequency band in the time domain information after the bandpass filtering is notched to obtain the time domain information.

[0242] In some embodiments, the second processing module 720 may also be used to:

[0243] The amplitude-frequency information is input into the first feature extraction model to obtain the amplitude-frequency feature vector output by the first feature extraction model; the first feature extraction model is trained using the historical amplitude-frequency information as a sample and the arc detection result corresponding to the historical amplitude-frequency information as a sample label;

[0244] The phase-frequency information is input into the second feature extraction model to obtain a phase-frequency feature vector output by the second feature extraction model; the second feature extraction model is trained using the historical phase-frequency information as a sample and the arc detection result corresponding to the historical phase-frequency information as a sample label;

[0245] The time domain information is input into the third feature extraction model to obtain the time domain feature vector output by the third feature extraction model; the third feature extraction model is trained using the historical time domain information as samples and the arc detection results corresponding to the historical time domain information as sample labels.

[0246] In some embodiments, the third processing module 730 may also be used to:

[0247] Fusing any two eigenvectors among the amplitude-frequency eigenvector, the phase-frequency eigenvector and the time-domain eigenvector to obtain at least three fused feature matrices;

[0248] At least three fused feature matrices are input into a target classification model to obtain an arc detection result output by the target classification model.

[0249] In some embodiments, the third processing module 730 may also be used to:

[0250] The element value of the target position in the target feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector is added to each element value of another feature vector to obtain a sub-fusion feature vector corresponding to the target feature vector; the other feature vector is any feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector except the target feature vector;

[0251] Based on the sub-fusion feature vectors corresponding to the target feature vector, a fusion feature matrix is ​​constructed.

[0252] The arc detection device of the photovoltaic system in the embodiment of the present application can be a photovoltaic system, or can be an electronic device connected to the photovoltaic system for communication, or can be a component in the photovoltaic system or electronic device, such as an integrated circuit or chip. The electronic device can be a terminal, or can be other devices other than the terminal. Exemplarily, the electronic device can be a mobile phone, a tablet computer, a laptop computer, a handheld computer, a vehicle-mounted electronic device, a mobile Internet device (Mobile Internet Device, MID), an augmented reality (augmented reality, AR) / virtual reality (virtual reality, VR) device, a robot, a wearable device, an ultra-mobile personal computer (ultra-mobile personal computer, UMPC), a netbook or a personal digital assistant (personal digital assistant, PDA), etc., and can also be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television (television, TV), a teller machine or a self-service machine, etc., which is not specifically limited in the embodiment of the present application.

[0253] The arc detection device of the photovoltaic system in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an IOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.

[0254] The arc detection device for a photovoltaic system provided in the embodiment of the present application can achieve Figures 1 to 6 To avoid repetition, the various processes implemented by the method embodiment are not described here.

[0255] In some embodiments, Figure 8 As shown, an embodiment of the present application further provides an electronic device 800, including a processor 801, a memory 802, and a computer program stored in the memory 802 and executable on the processor 801. When the program is executed by the processor 801, each process of the arc detection method embodiment of the above-mentioned photovoltaic system is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0256] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0257] The embodiment of the present application also provides a photovoltaic system.

[0258] In some embodiments, the photovoltaic system includes: at least one photovoltaic module.

[0259] In this embodiment, the photovoltaic component is a component that converts light energy into electrical energy.

[0260] The photovoltaic system is detected based on the arc detection method for the photovoltaic system as described in any of the above embodiments.

[0261] According to the photovoltaic system provided in the embodiment of the present application, by performing data processing on the sampled values ​​of the output sampling signal of the collected photovoltaic system, amplitude-frequency information, phase-frequency information and time domain information are obtained, and more comprehensive information with different attributes is obtained. Feature extraction is performed on the information with different attributes to obtain the feature vector of each attribute, that is, the amplitude-frequency feature vector, the phase-frequency feature vector and the time domain feature vector are obtained. The feature vectors of different attributes are processed through a target classification model, and arc detection results are obtained based on more comprehensive features, thereby improving the accuracy of arc detection and reducing missed detections and false detections.

[0262] An embodiment of the present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the above-mentioned photovoltaic system arc detection method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0263] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0264] An embodiment of the present application also provides a computer program product, including a computer program, which implements the above-mentioned arc detection method for the photovoltaic system when executed by a processor.

[0265] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory ROM, a random access memory RAM, a magnetic disk or an optical disk.

[0266] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned photovoltaic system arc detection method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0267] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0268] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0269] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0270] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0271] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "illustrative embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0272] Although the embodiments of the present application have been shown and described, those skilled in the art will appreciate that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the claims and their equivalents.

Claims

1. A method for detecting arc in a photovoltaic system, characterized in that: include: Performing a fast Fourier transform on the sampled value of the output sampled signal of the photovoltaic system to obtain amplitude-frequency information and phase-frequency information of the output sampled signal; Performing filtering on the sampled values ​​to obtain time domain information of the output sampled signal; Extracting features from the amplitude-frequency information, the phase-frequency information and the time-domain information respectively to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector; Inputting the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector into a target classification model to obtain an arc detection result output by the target classification model, wherein the arc detection result includes: the presence of an arc or the absence of an arc; The target classification model is trained using historical amplitude-frequency feature vectors, historical phase-frequency feature vectors and historical time-domain feature vectors as samples and historical arc detection results as sample labels.

2. The arc detection method of a photovoltaic system according to claim 1, characterized in that: The method of performing a fast Fourier transform on the sampled value of the output sampled signal of the photovoltaic system to obtain the amplitude-frequency information and the phase-frequency information of the output sampled signal includes: Performing the fast Fourier transform on the sampled value of the output sampled signal of the photovoltaic system to obtain initial amplitude-frequency information and initial phase-frequency information of the output sampled signal; The noise frequency bands in the initial amplitude-frequency information and the initial phase-frequency information are deleted respectively to obtain the amplitude-frequency information and the phase-frequency information.

3. The arc detection method of a photovoltaic system according to claim 1, characterized in that: The filtering process on the sampled values ​​to obtain the time domain information of the output sampled signal includes: The sampling values ​​are subjected to notch processing and bandpass filtering to obtain the time domain information.

4. The arc detection method of a photovoltaic system according to claim 3, characterized in that: The performing notch processing and bandpass filtering on the sampled values ​​to obtain the time domain information includes: Performing the notch processing on the noise frequency band in the sampling value to obtain time domain information after the notch processing; Performing the bandpass filtering process on the time domain information after the notch processing to obtain the time domain information; or; Performing the bandpass filtering process on the sampled values ​​to obtain time domain information after the bandpass filtering process; A notch process is performed on the noise frequency band in the time domain information after the bandpass filtering process to obtain the time domain information.

5. The arc detection method for a photovoltaic system according to any one of claims 1 to 4, characterized in that: The extracting features of the amplitude-frequency information, the phase-frequency information and the time-domain information to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector respectively includes: The amplitude-frequency information is input into a first feature extraction model to obtain the amplitude-frequency feature vector output by the first feature extraction model; the first feature extraction model is trained using historical amplitude-frequency information as samples and arc detection results corresponding to the historical amplitude-frequency information as sample labels; The phase-frequency information is input into a second feature extraction model to obtain the phase-frequency feature vector output by the second feature extraction model; the second feature extraction model is trained using historical phase-frequency information as samples and arc detection results corresponding to the historical phase-frequency information as sample labels; The time domain information is input into a third feature extraction model to obtain the time domain feature vector output by the third feature extraction model; the third feature extraction model is trained using historical time domain information as samples and arc detection results corresponding to the historical time domain information as sample labels.

6. The arc detection method for a photovoltaic system according to any one of claims 1 to 4, characterized in that: The step of inputting the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector into a target classification model to obtain an arc detection result output by the target classification model includes: Fusing any two of the amplitude-frequency feature vector, the phase-frequency feature vector, and the time-domain feature vector to obtain at least three fused feature matrices; The at least three fused feature matrices are input into the target classification model to obtain the arc detection result output by the target classification model.

7. The arc detection method of a photovoltaic system according to claim 6, characterized in that: The fusing of any two feature vectors among the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector obtains at least three fused feature matrices, including: The element values ​​of the target position in the target feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector are added to each element value of another feature vector to obtain a sub-fusion feature vector corresponding to the target feature vector; the other feature vector is any feature vector in the amplitude-frequency feature vector, the phase-frequency feature vector and the time-domain feature vector except the target feature vector; The fused feature matrix is ​​constructed based on the sub-fused feature vectors corresponding to the target feature vector.

8. An arc detection device for a photovoltaic system, characterized in that: include: A first processing module is used to perform a fast Fourier transform on the sampled value of the output sampled signal of the photovoltaic system to obtain the amplitude-frequency information and phase-frequency information of the output sampled signal; Performing filtering on the sampled values ​​to obtain time domain information of the output sampled signal; A second processing module is used to extract features from the amplitude-frequency information, the phase-frequency information and the time-domain information respectively to obtain an amplitude-frequency feature vector, a phase-frequency feature vector and a time-domain feature vector; A third processing module is used to input the amplitude-frequency feature vector, the amplitude-frequency feature vector and the time domain feature vector into a target classification model to obtain an arc detection result output by the target classification model, wherein the arc detection result includes: the presence of an arc or the absence of an arc; The target classification model is trained using historical amplitude-frequency feature vectors, historical phase-frequency feature vectors and historical time-domain feature vectors as samples and historical arc detection results as sample labels.

9. A photovoltaic system, characterized in that: include: at least one photovoltaic module; The photovoltaic system performs arc detection based on the arc detection method for the photovoltaic system according to any one of claims 1 to 7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the arc detection method for the photovoltaic system according to any one of claims 1 to 7 is implemented.

11. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the arc detection method for a photovoltaic system as described in any one of claims 1 to 7 is implemented.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the arc detection method for a photovoltaic system according to any one of claims 1 to 7 is implemented.

Images